UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Determinants of ASP success MacKenzie, Winston Joseph 2001

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_2001-0240.pdf [ 4.89MB ]
Metadata
JSON: 831-1.0089933.json
JSON-LD: 831-1.0089933-ld.json
RDF/XML (Pretty): 831-1.0089933-rdf.xml
RDF/JSON: 831-1.0089933-rdf.json
Turtle: 831-1.0089933-turtle.txt
N-Triples: 831-1.0089933-rdf-ntriples.txt
Original Record: 831-1.0089933-source.json
Full Text
831-1.0089933-fulltext.txt
Citation
831-1.0089933.ris

Full Text

DETERMINANTS O F A S P S U C C E S S by W I N S T O N J O S E P H M A C K E N Z I E B . S c . d . , T h e Universi ty Co l lege of C a p e Breton, 1997 A T H E S I S S U B M I T T E D IN P A R T I A L F U L F I L M E N T O F T H E R E Q U I R E M E N T S F O R T H E D E G R E E O F M A S T E R O F S C I E N C E ( B U S I N E S S A D M I N I S T R A T I O N ) in T H E F A C U L T Y O F G R A D U A T E S T U D I E S (Managemen t Information S y s t e m s Div is ion, Facul ty of C o m m e r c e and Bus iness Administrat ion) W e accep t this thesis as conforming i ^ the . requ i red s tandard . T H E U N I V E R S I T Y O F B R I T I S H C O L U M B I A Apri l 2001 © W i n s t o n J o s e p h M a c K e n z i e , 2001 In present ing this thes is in partial fulfil lment of the requi rements for an a d v a n c e d degree at the Univers i ty of Brit ish C o l u m b i a , I ag ree that the Library shal l m a k e it f reely avai lab le for reference and study. I further agree that permiss ion for ex tens ive copy ing of this thes is for scholar ly pu rposes may be granted by the head of my depar tment or by his or her representat ives. It is unders tood that copy ing or publ icat ion of this thes is for f inancia l gain shal l not be a l lowed without my written permiss ion . C 3 JDepartment'-oK T h e Universi ty of Brit ish C o l u m b i a Vancouve r , C a n a d a Abstract Appl icat ion Se rv i ce Prov iders ( A S P ) are a recent phenomenon in the sphe re of information technology. Leverag ing the advan tages of the universal ly a c c e s s i b l e and ubiqui tous nature of the Internet, appl icat ion serv ice providers c la im to lower the cos t of ownersh ip of appl icat ions. Genera l l y operat ing f rom a central and remote da ta centre, A S P s 'rent' their appl icat ions to cl ients in a 'pay a s you go ' fash ion . Th is has severa l advan tages to the client bes ides cost sav ings . Th is thes is introduces the reader to A S P s and exam ines the reasons c o m p a n i e s hire the serv ices of an A S P ; furthermore it sugges ts an analyt ic f ramework b a s e d upon economic concep ts in an attempt to predict what bus iness mode ls and strategies will be success fu l in the A S P market and what s h a p e that market might take. T h e thes is a lso d i s c u s s e s the poss ib le effect of s o m e emerg ing techno log ies . ii Table Of Contents A b s t r a c t ii T a b l e O f C o n t e n t s iii L i s t O f T a b l e s v L i s t O f F i g u r e s vi A c k n o w l e d g e m e n t s vi i D e d i c a t i o n viii C h a p t e r 1. Introduct ion 1 C h a p t e r 2. D e f i n i n g T h e A S P C o n c e p t 3 2.1 . Wha t Is A n A S P ? 3 2.2. Wha t T y p e s Of A s p s A re Out The re? 6 2.3 . W h o A r e T h e P layers In T h e A S P E c o s y s t e m ? 8 2.4. B u s i n e s s Mode ls 14 ASP Market Internal Operations 14 A New Stratification OfASPs 18 2.5. Compar i ng T h e A S P Mode l T o In H o u s e Solut ions 26 2.6. Techno log ica l P rob lems A n d A d v a n c e s 28 2.7. E x a m p l e s Of A S P s 3 3 ASP Market Shares 33 C h a p t e r 3. T h e A S P M a r k e t 39 3.1 . H o w Big Is T h e Market For A s p s ? 39 How Much Revenue Is There For Asps? 39 Who Is Investing And How Much? 42 iii 3.2. Market T rends 44 What Is The Economic Outlook For Asps?.. 46 Changing Customer Attitudes 47 C h a p t e r 4. A S P F r a m e w o r k : D e c i d i n g A S P T y p e 49 4 .1 . E c o n o m i c C o n c e p t s U s e d In A S P Framework 49 4 .2 . A S P Framework 52 Outcome Variable 53 Input Variables 60 Interrelationships Of Input Variables.. 63 4.3 . F ramework Predic t ions 68 Summary Of Framework Predictions 80 C h a p t e r 5. A S P F r a m e w o r k : D e c i d i n g W h e t h e r T o U s e A n A S P 81 5.1. F o r c e s That A r e C a u s i n g Or Hinder ing A S P Growth 82 5.2. S u m m a r y Of Predic t ions A n d Find ings Of A S P Framework 98 Contrast Between ASPs And Outsourcing 99 5.3. A S P Stra teg ies A s Dictated By A S P Framework 100 5.4. S u m m a r y 104 C h a p t e r 6. C o n c l u s i o n 105 B i b l i o g r a p h y 107 A p p e n d i x A L i s t O f C o m p a n i e s 112 A p p e n d i x B L i s t O f A b b r e v i a t i o n s U s e d 115 iv List Of Tables Tab le 1. Categor izat ion By Market A n d S p a n 18 Tab le 2. Sub types Of N iche V S P s 20 Tab le 3. C o m p a r i s o n Be tween A S P A n d Tradit ional IT 27 Tab le 4. A S P s By T y p e A n d Target Cus tomer S i z e 38 Tab le 5. T h e C h a n g i n g Project ions Of A S P R e v e n u e By Major R e s e a r c h Corpora t ions . 40 Tab le 6. Appl icat ions By Percen tage Intending T o Buy F rom A n A S P 45 Tab le 7. C o d e s For A S P T y p e U s e d In A S P Framework 54 Tab le 8. S u m m a r y Of Var iab le Interrelationships 68 Tab le 9. Cl ient A c c e p t a n c e A n d Lock-In By Crit icality A n d Substitutabil i ty 76 Tab le 10. F ramework Predict ions 80 Tab le 11. Fac to rs Increasing T h e Likel ihood Of A S P U s a g e 98 Tab le 12. Fac to rs Increasing T h e Likel ihood Of Not Us ing A n A S P 99 Tab le 13. S u m m a r y Of Success fu l St rategies 106 v List Of Figures Figure 1. G e n e r i c A S P Mode l 8 F igure 2. Cus tomer , Cl ient A n d A S P Interaction 11 F igure 3. Microsoft A S P Serv ice Del ivery Mode l 13 F igure 4 . C o m p a r i s o n Of A S P Market Wi th A n d Without Aggrega to rs . 16 F igure 5. Market S h a r e By Number Of Cus tomers [ S C N , 2000] 34 F igure 6. Market S h a r e By Number Of Cont rac ts [ S C N , 2000] 34 F igure 7. Commun ica t i ons Startups Venture Fund ing By Industry S e g m e n t 4 3 F igure 8. Attr ibutes Affect ing T h e Fo rm Of A n Organizat ion 52 F igure 9. A S P T y p e Matrix 54 F igure 10. Funct ional i t ies C o m b i n e d Into A n Appl icat ion By Cl ient 56 F igure 11 . Appl ica t ions Del ivered T o Cl ient 57 F igure 12. P a c k a g e Del ivered T o Cl ient 58 F igure 13. Cl ient C h o i c e Of A S P T y p e A s Transact iona l C o s t s Increase 69 F igure 14. Cl ient C h o i c e Of A S P T y p e For P(i) A s A s s e t Speci f ic i ty Domina tes 71 F igure 15. Cl ient C h o i c e Of A S P Type For P(i) A s A s s e t Speci f ic i ty Increases 72 F igure 16. Cl ient C h o i c e Of A S P Type For f(i) Or A(i) A s Speci f ic i ty Increases 73 F igure 17. Cl ient C h o i c e Of A S P Type For P(i) A s Crit icality Domina tes 77 F igure 18. Cl ient C h o i c e Of A S P Type For P(i) A s Substitutabil i ty Increases 79 vi Acknowledgements I would like to acknow ledge Ya i r W a n d for his many contr ibut ions in shap ing this thes is . My utmost thanks to Albert Dexter, for his adv ice and support a long the w a y and at the end . Many thanks to Mart in Mroz for his edit ing, input and moral support . M y gratitude to K a lpna So lank i and Suni l C h a u h a n for their late night proofreading. My utmost gratitude to Ju l ie N icho ls , Jenni fer P a p k e and Margot F rase r for providing information, direct ion and their ever-cheerfu l d isposi t ions. My thanks to S i m o n F rase r Universi ty Library for giving me a c c e s s to their cons iderab le , and up to date, resources . A n d lastly, for my parents An i ta and Wins ton M a c K e n z i e , for their love and support . vii Dedication T o M o m and D a d , without whom I would not have star ted. T o L iv ia, without w h o m I would not have f in ished. viii Chapter 1. Introduction Appl icat ion Serv i ce Prov iders ( A S P s ) are a growing phenomenon in the IT wor ld . A S P s rent a c c e s s to appl icat ions over the Internet. In the past much hype about their future s u c c e s s and current bad t imes have been in the news. Investors and cus tomers have sh ied away f rom them after watch ing the failure of dot corns. W h a t are A S P s ? A n d how will they fare in the IT marke tp lace? Wi l l cus tomers really buy their appl icat ions 'over the wire '? A n d most importantly, what a re the determinants of A S P s u c c e s s ? Th is thes is at tempts to answer these quest ions by examin ing why potential cus tomers favour this del ivery method for software. It sugges ts an analyt ic f ramework b a s e d upon an extens ive literature review . Th is at tempts to predict what type of A S P bus iness model will be success fu l and what strategies would be most appropr iate for e a c h A S P type. T h e f ramework accomp l i shes this by examin ing how the var ious types of A S P s will relate to one another and to their c l ients. In Chap te r 2 , a definit ion of A S P s is set forth. A S P s are briefly c o m p a r e d to outsourcers and are then ca tegor ized by ownersh ip of appl icat ion. T h e focus then shifts to the var ious entit ies interacting with A S P s . B u s i n e s s mode ls are set forth and a new classi f icat ion s c h e m e b a s e d on appl icat ion market ve rsus s c o r e is put forward. T h e chapter cont inues on with a compar i son of hosted serv ices against in -house solut ions. S o m e upcoming technolog ica l solut ions to per formance prob lems (packet label l ing), new connectivi ty s tandards such a s U D D I / S O A P , and a d v a n c e s in software deve lopment techn iques are exp lored and how they so lve s o m e of the prob lems with the A S P mode l a re shown . T h e chapter conc ludes with s o m e examp les of A S P s and their market sha re . 1 The market for ASPs is discussed in Chapter 3. Total revenue projections are stated, as are investment figures and trends. Predictions with regards to ASPs in the year 2001 and beyond are presented. Changes in customer attitudes toward software as utilities instead of a product or service are discussed. An analytic framework for examining the determinants of an ASP's success is created in Chapter 4 . Some economic concepts such as transaction costs and asset specificity are reviewed since they are used throughout the framework that is developed. In Chapter 4 the outcome variable is the decision not to use an ASP or to use some type of ASP. Several variables are used to determine the ASP type decided upon. This comes together to from a framework of ASP success. Predictions are made as to what types of ASPs will succeed in the future and the strategies they should follow. Strategies that an ASP should follow, according to the ASP type and what the model states, in order to draw customers to their business and keep them, are discussed. Chapter 5 puts forward the forces driving clients toward ASP usage. Also, the possible reasons that clients are not using ASPs are examined in greater detail than they are in Chapter 4 . These forces are analysed in light of the framework constructed in Chapter 4, and several of the variables used in Chapter 4 are applied to determine the client's decision to use an ASP or not. This is done in hopes of furthering the predictive abilities of the model. Furthermore, Chapter 5 suggests the strategies an ASP might follow in light of the findings of Chapters 4 and 5 to maximize the likelihood of success. Mix of services, costs and relationship to other ASPs are discussed. Chapter 6 concludes the thesis by summarizing our findings and presenting some possible research directions and extensions to this work. 2 Chapter 2. Defining The ASP Concept In this chapter the A S P concept is def ined and the types of A S P s that are operat ing is examined . T h e A S P ' ecosys tem ' d i s c u s s e d , a s well a s s o m e A S P bus iness mode ls . A stratif ication of A S P s that examines them based on a market s p a n ve rsus functionality s p a n bas is is p roposed . A compar i son of hosted serv ices against in -house solut ions is made . S o m e upcoming technolog ica l a d v a n c e s are d i s c u s s e d that purport to el iminate of s o m e Q o S (Quality of Serv ice) conce rns with respect to the Internet new connect iv i ty s tandards that will c h a n g e the ways functionality is de l ivered over the Internet. Fur thermore, new deve lopment methods and technolog ies that will c reate appl icat ions des igned to be del ivered over the Internet are a lso examined . T h e chapter then conc ludes with s o m e examp les of A S P s and their market share . 2.1. What Is An ASP? International Da ta Corporat ion supposed ly co ined the term Appl icat ion Se rv i ce Provider , or A S P , in 1998 in a study on the future of outsourc ing [Butler 2000] . In [Portera 2000] an A S P , or a Hos ted Se rv i ce Provider , is def ined a s : " . . . an independent , third-party provider of so f tware-base serv ices that are del ivered to cus tomers ac ross a w ide -a rea network, typically the Internet." Fur thermore, Por tera differentiates the A S P mode l f rom the traditional c l ient /server app roach to software: "Unl ike a traditional c l ient /server sof tware vendor , an A S P typical ly instal ls and hosts appl icat ions exc lus ive ly within its own da ta centre." It shal l be shown that the da ta centre may not be long to the A S P itself, but to a third party, a s long a s it is not at the user 's si te. 3 Exp los ive growth in network capaci ty has el iminated the tradit ional a rgument that a hosted appl icat ion would be too s low. Th is comb ined with the dropping pr ice of bandwidth, and up coming improvements like packet label l ing, has enab led the A S P industry to ove rcome s o m e of the user conce rns about s p e e d . S i n c e bandwidth is currently be ing pr iced in a commodi ty- l ike fash ion , the economies that the A S P model c a n real ize will be translated into sav ings for the cus tomer , rather than buried under connect ion cos ts . A S P s t hemse lves are not really a new concept . O n e author op ines the fol lowing about A S P s : ASP, or application service provider, is one of the more common terms used to describe the companies working to free businesses from the burden of designing, implementing, and maintaining IT systems. Applications hosting is another popular term. I prefer outsourcing; a time-honoured term for what is really a time-honoured practice, despite efforts to the contrary to make this market look all new and shiny. [Greenbaum, 2000] Al though this opinion may s e e m d ismiss ive , it is accura te with regard to the A S P compan ies themse lves . However , the author d o e s miss a few bas ic facts. For starters, A S P s today are offering major mult i -user appl icat ions whereas back in the 70s these were most ly s ingle user appl icat ions [Slavid, 2000] . Second ly , hosted appl icat ions represent a c h a n g e f rom thinking of sof tware a s a 'product ' to thinking of software a s a 'serv ice ' . Th is fundamenta l c h a n g e in the way c o n s u m e r s v iew software has s o m e important ramif icat ions for the market . Last ly, a n d most importantly, A S P s are offered over the Internet. Let us cons ider the impl icat ions of A S P s being offered over the Internet a little further. B e c a u s e the Internet is ubiquitous, and there are few computer use rs without Internet a c c e s s , they al low universal a c c e s s to A S P sys tems by suppl iers and vendors . 4 M a n y cons ider A S P s a s essent ia l ly an outsourc ing of the appl icat ion ma in tenance , des ign and provis ioning funct ions of the MIS or IT department [G reenbaum 2000 , Butler 2000] . Fur thermore, the A S P model bares a remarkable resemb lance to another form of outsourc ing: 't ime shar ing ' of computers in the early 1970s . [Slavid 2000] "In those days , computers were expens ive , and the expert ise required to p rogram and maintain them w a s s c a r c e " [Butler 2000] . S o m e of these forces are still in effect today. C o m p u t e r s migrated f rom the t ime-shar ing host to strict outsourc ing compan ies in the ear ly 1980s . M a n y factors p layed a role in the movement of compan ies a s the u s a g e of the t ime-shared computer app roached 24 hours per day and the price of computers started to dec l ine , thus mak ing t ime-shar ing more expens ive than buying and running a data centre. A l though A S P s c lear ly act in an outsourc ing role, this thes is shal l not focus on i ssues regarding gener ic outsourc ing but exc lus ive ly upon i ssues relating to A S P s . However , our d i scuss ion of A S P s d o e s have strong paral lels to the research in the f ield of outsourc ing. Outsourc ing in genera l is an a rea of intensive research and there ex is ts a large body of knowledge on the subject. T h o s e who are interested in more information regarding outsourc ing are d i rected to Winn ing the Outsourc ing G a m e by Jane t Butler, Strategic Outsourc ing Dec is ions written by W . R . K ing , or Outsourc ing : A C I Q ' s Perspec t i ve written by O a k i e Wi l l i ams. Throughout this thes is we shal l highlight relevant f indings in the field of outsourc ing. For the pu rposes of this thes is , an A S P shal l be def ined a s : an organizat ion that offers a contractual serv ice to deploy, host, m a n a g e and rent a c c e s s (via the Internet) to an appl icat ion f rom a remote facility. A S P s are respons ib le for either directly or indirectly providing all the speci f ic activit ies and expert ise a imed at manag ing a sof tware appl icat ion or set of appl icat ions [Douglass 2000] . 5 2.2. What Types Of Asps Are Out There? Accord ing to T e l e C h o i c e ' s Scot t Heinlein there are four major ca tegor ies of A S P s [Heinlein, 2000]. T h e s e groupings are b a s e d mainly on appl icat ion ownersh ip or c o m p a n y origins and are fairly representat ive of the c lassi f icat ion of A S P s that most ana lys ts a re current ly us ing in examin ing this market . It is useful to examine this at this juncture b e c a u s e it prov ides the reader with an insight into the categor izat ion of A S P s that a re currently in use . Let us look at e a c h , and d i s c u s s the strengths and w e a k n e s s e s of e a c h . Resellers: P a c k a g e another A S P ' s product and resel l it to end cus tomers for a d iscounted cost . T h e y typical ly will bill the end cus tomer and provide the first level of support . Many resel lers are not respons ib le for maintaining the appl icat ion f rom either a technica l or product perspect ive , beyond first level of support . T h e y do not have m u c h control over the appl icat ion but do own the customer . Th is model c a n be good for c o m p a n i e s want ing to offer appl icat ions but not having the exper ience to do s o . Th is prov ides them with the ability to comb ine their own products with the resold appl icat ions in a bundle p a c k a g e offering. T h e downs ides to this mode l are lack of product control and often smal l marg ins . T h e resel lers M ic roage and Viasof t fall in this category. Aggregators: Integrate multiple appl icat ions f rom different A S P s and offer them to cus tomers . T h e appl icat ions are usual ly offered through a portal with a s ingle s ign-on. Uti l izing multiple A S P s a l lows aggregators to offer vert ical p a c k a g e s or many horizontal p a c k a g e s . Th is limits the number of A S P s that bus i nesses must dea l with a s they add more appl icat ions. T h e d isadvan tage is that the aggregator mode l tends to b e rather comp lex b e c a u s e of the integration requirements of multiple A S P s . E x a m p l e s of aggregators include J a m c r a c k e r , Agil i t i , and Bluetrain. 6 Pure-plays: Re ly least on partnerships to del iver their appl icat ions. T h e y ideal ly have their own data cent res where they maintain the appl icat ions themse lves . Th is g ives them a high level of control over the serv ice and cus tomer exper ience. S o m e pure-p lays are partnering with Internet serv ice providers. Th is g ives them total end- to-end accountabi l i ty to the cus tomer . T h e d isadvantage to this model is that it requires exper t ise in many a reas . Pure -p lays must know the software s ide a s well a s the data centre s ide . It is a l so the most cost ly of the mode ls . However , many cus tomers will want their A S P to have end- to-end accountabi l i ty if they are running critical bus iness appl icat ions. E x a m p l e s of pure-p lays are U S i , Interliant, and Cor io . Business Process Outsourcers: can be either a resel ler, aggregator , or pure-play. T h e dist inction is that they del iver bus iness p rocess outsourc ing se rv i ces in addit ion to appl icat ion se rv i ces to the customer . They may provide an ou tsourced H R organizat ion, P R firm, sh ipp ing company , account ing group, etc. B u s i n e s s p rocess ou tsourcers may partner with managemen t consul t ing f irms for these outsourc ing se rv i ces a s well a s c o m p a n i e s like K inkos , Of f ice Depot , Federa l E x p r e s s , and J B Hunt. B u s i n e s s p rocess outsourcers can be compa red to a profess ional se rv ices organizat ion. T h e d isadvantage to this mode l is it rel ies on severa l par tnerships simi lar to the aggregator mode l and d e m a n d s creativity and simplici ty in packag ing . For the client of an A S P , the distinction between a B u s i n e s s P r o c e s s Ou tsource r and any of the other three A S P types would be c lear. In addit ion to the del ivery of appl icat ion functionality, B u s i n e s s P r o c e s s Outsourcers would a lso del iver outsourc ing capac i ty that ou tsourcers have a lways del ivered. However the distinction be tween resel lers , aggregators and pure-p lays b e c o m e s c lear when c h a n g e s have to be m a d e , or a re m a d e , to the appl icat ion(s) which forms the bas is of the serv ice. Rese l l e rs and aggregators have little or 7 no control over the appl icat ion s ince they are pass ing a long the se rv i ces of other A S P s . Pure -p lays create their own appl icat ions s o they have end- to-end control a s ment ioned previously. 2.3. Who Are The Players In The ASP Ecosystem? T h e bus iness s p a c e surrounding A S P s is often referred to a s a n e c o s y s t e m . T h e reason for this is the level of non- l inear interactions and chang ing roles that the var ious p layers in this market undergo [Charley, 2000] . W e shal l begin with a look at those var ious p layers and where they are headed within this market. Here we s e e the structure of a gener ic A S P interacting with one of its c l ients: Figure 1. Generic ASP Model Data Data Centre ISP Rental Payment Application There are four key p layers involved in del ivering the appl icat ion serv ice to the client in this f igure. T h e ISV (Independent Sof tware Vendor ) , A S P (Appl icat ion Se rv i ce Provider) , ISP 8 (Internet Se rv i ce Provider) and the Data Cent re . Note that these are not necessar i l y distinct organizat ions. For U S i , the A S P and Data Cen t re are all in one organizat ion, for M y S A P . c o m the ISV and A S P are all within one organizat ion. If a te lco (telco is a c o m m o n short form of te lecommunica t ions company) such a s A T & T were to buy U S i , only two c o m p a n i e s wou ld be represented in the gener ic A S P d iagram above . Not depic ted in this d iagram are the cl ient 's cus tomers . A S P s impact the cus tomer in certain ways . Let us examine each of these p layers in s o m e detai l . Independent Software Vendors (ISVs) Historical ly, ISVs mainta ined a direct relationship with the users of their sof tware. However , under the A S P del ivery mode l , the ISV is re legated to only a smal l componen t in the overal l solut ion in which the A S P m a n a g e s the entire cus tomer relat ionship. For this reason , major sof tware h o u s e s s u c h a s Orac le , S A P and IBM have a l ready entered the A S P market a s direct compet i tors . Microsoft has given mixed s igna ls a s to whether or not it will enter the market directly. However , Microsoft has s ignal led a signif icant back ing of p layers a l ready in this market with its recent . N E T platform and through its A S P Par tner Cert i f icat ion P rog ram. Internet Service Providers (ISPs) T e l c o s and I S P s are the 'connectivity' component of the A S P mode l . M a n y of the larger te lcos have the network and da ta centre capabi l i t ies, but typical ly lack the appl icat ion expert ise that an A S P cal ls for. A s we shal l s e e in later in this chapter , many ana lys ts think the te lcos shou ld , and will, buy an A S P b e c a u s e of their d e p r e s s e d market capi tal va lues . Exaggera t ing the push for te lcos to buy A S P s , b roadband serv ices are quick ly becoming a commodi ty with low marg ins , and adding a hosted serv ice would inc rease their return on 9 network investment [King, Sep t 2000] . Qwes t , A T & T and Sprint are all involved in joint A S P ventures. Data Centre Often referred to a s Se rve r F a r m s , data cent res can be owned by the A S P (as with USi ) or owned by s o m e o n e other than the A S P (as with Cor io) . Of ten T h e da ta cent res are often owned by te lcos, however E x o d u s and Concent r i c a re both data centre only serv ices . Whe the r or not they will move into the A S P market is unknown at this t ime. The re are two main ca tegor ies of data cent res : 1) co- locat ion providers (co-lo's) and 2) m a n a g e d host ing providers. C o - l o ' s offer cl ients a secu re envi ronment with reliably c lean power in which to house servers . M a n a g e d host ing providers take the next s tep by add ing IT support and hardware, bas ica l ly offering cl ients turnkey host ing solut ions. Forrester R e s e a r c h predicts m a n a g e d host ing will account for 86 percent of an ant ic ipated $12 .5 bill ion data centre revenue in 2 0 0 3 [Bernard, Feb . 2001] . A n d y Hunn , a senior director of B u s i n e s s Deve lopment at M H P Digex has openly stated that "Co- loca t ion is a l ready getting commodi t i zed , " [Bernard, F e b . 2001] and the users of da ta cent res des i re the extras such a s t rouble-shoot ing serv ices that m a n a g e d hosts provide. Fur thermore, many A S P s are comparat ive ly c a s h s tarved and investors are not a s patient a s they were with the dot corns when it c o m e s to turning a profit [Wainewright, Ma rch 2001] . Fortunately for the A S P s that are not p lanning on building their own da ta cent res , the market for da ta centre se rv i ces is f looded at the moment . Da ta centre serv ice providers were expect ing the A S P market to be much larger than it currently is and they were a l so expec ted to be serv ic ing the dot c o m market at this point in t ime [Bernard, F e b 2000] . Neither of these assumpt ions has held true and e x c e s s data centre capac i ty has therefore resul ted. 10 Customer A S P s affect the way compan ies dea l with their cus tomers . For ins tance, if a c o m p a n y ou tsourced its sh ipment t racking software to an A S P , it would m a k e s e n s e to leverage the fact that A S P s del iver their se rv ices over the net. O n e way to take advan tage of this would be to have the cus tomer contact the A S P directly to get information, or m a k e their payments for goods . However this can be done in a t ransparent fash ion . C o n s i d e r the fol lowing d iagram: Figure 2. Customer, Client And ASP Interaction How the System Appears to the Customer Billing, Order Info, Goods Payment, Info Requests How the System is really working Billing, Order Info, Goods r ASP Payment, Info Requests / 11 Here the cus tomer has visi ted the cl ient 's websi te to obtain information about its order. C u s t o m e r s may bel ieve they are interacting with the cl ient 's IT sys tems . However , t ransparent to the cus tomers themse lves , they are really interacting with the cl ient 's A S P . T h e A S P has all the data and functionality to rel ieve the client of the burden of support ing cus tomer information requests . Note that a suppl ier, retailer or any other entity that is interacting with the client cou ld replace the 'customer ' . Application Service Providers (ASPs) T h e A S P general ly d o e s not own the software it l eases out to c l ients. C o m p a n i e s like U S i and Cor io (both in the top five for number of cl ients and contract s i ze [ S C N , 2000]) do build s o m e of the software and integrate the solut ions they market . Howeve r they are in ag reements with Microsoft to use Microsoft 's deve lopment tools a n d if they are going to move toward aggregat ion or resel l ing, they will use one of Microsof t 's A S P par tners : 12 Figure 3. Microsoft ASP Service Delivery Model Small, Medium and Large Organizations Solution Delivery Application Developers ISVs Web Developers ASP - Solution Developer Develop Packaged and/or Custom ASP Solutions on ASP Platform Enabler Systems, Own Customer Relationship Systems Integrators Value-Added Resellers ASP - Agent Resell Packaged ASP Services for ASP Application Aggregators 8 2 = O » V) o ASP - Platform Enabler Managed Hosting, Platform, Provisioning, Billing, Telco -Licensing ASP - Application Aggregator Messaging, Office, LOB - Licensing This d iagram s h o w s the many d iverse activit ies involved in the del ivery of an A S P serv ice a s put forward by Microsoft on its A S P partner websi te or a s p resented in [Cornf ield, 2000]. S i n c e A S P s interact with the software vendors and te lecommunica t ions c o m p a n i e s (or their ISP div is ions) and face integration, deve lopment and networking i s sues on a dai ly bas is , a d iverse set of ski l ls are required to get an A S P running smooth ly . For this reason , A S P s often str ike dea ls with I S V s , te lcos and consul t ing f i rms in order to leverage those f i rms' understanding of those a reas . Qwes t Cyber .So lu t ions del ivers infrastructure, sof tware, appl icat ion deve lopment / implementat ion, ma in tenance and enhancemen ts , all in an end- to-end p a c k a g e . Cyber .so lu t ions is a joint venture between Qwes t commun ica t ions and K P M G . Cyber .so lu t ions has an obv ious advantage by having s u c h an intimate relat ionship with these c o m p a n i e s : Q w e s t Cyber .so lu t ions has a c c e s s to Q w e s t ' s networking ski l ls and 13 capac i t ies at a preferential rate and K P M G brings in a "wealth of intel lectual property". Both have invested a great dea l of c a s h into the c o m p a n y s o even though it w a s in a negat ive c a s h f low posit ion for a long t ime, it has out lasted most other start ups [King, Sep t 2000] . Or even P a n d e s i c , born a s a joint venture between Intel and S A P in 1997, P a n d e s i c des igned its hosted e -bus iness solut ion for smal l and mids ize b u s i n e s s e s . P a n d e s i c leveraged its relat ionship with both parents and had what most in the industry cons ide red a great serv ice [ M c C a b e , Nov. 2000]. However , P a n d e s i c has gone out of bus iness . T h e channe l confl ict on S A P ' s part (between P a n d e s i c and M y S A P . c o m ) is often b lamed for this dem ise . Fo r A S P s , their re lat ionships with the var ious industry p layers a re ex t remely important for the survival of the c o m p a n y and adopt ion of its se rv ices within the industry. W e shal l cont inue our d i scuss ion of A S P s in sect ion 2.4 where we shal l cons ider the var ious types of A S P bus iness mode ls . 2.4. Business Models Th is sect ion exam ines the ways in which A S P s genera te i ncome and their internal operat ions. Se rv i ce Leve l Ag reemen ts (SLA) and the role they play in the A S P / C l i e n t relat ionship shal l a l so be d i s c u s s e d . ASP Market Internal Operations W e c a n s e e in the Microsoft A S P Serv ice Del ivery Mode l (figure 3) that Microsof t subsc r ibes to four spec ia l iza t ions of labour under the A S P market. T h e s e shal l be cons ide red a s the ' bus iness mode ls ' for the remainder of this thes is . 14 Platform Enabler Data cent res and te lcos fall under this category. Essent ia l ly we are talking about organizat ions that provide the hardware and infrastructure that m a k e the A S P model poss ib le . Plat form enab lers general ly provide m a n a g e d host ing to 3 r d party A S P s to deve lop and del iver their solut ions a s network serv ices . Genera l l y they channe l their se rv ices to end cus tomer through I S P s , A S P s and deve lopers . Qwes t , D igex and Da ta Return are all p layers in this market . P lat form enab lers c a n m a k e their revenue by charg ing for da ta centre u s a g e and network serv ices on the back end and then charg ing the cus tomer directly for Internet a c c e s s when they provide it. Solution developer (or Provider) Not ice that this c a n be either an ISV (Such a s Microsoft itself), a p rogrammer des ign ing web p a g e s that dr ive the appl icat ion, or a ful l-blown A S P . However under this s c h e m e it d o e s not matter. Solut ion providers are what first c o m e to mind when one th inks of an A S P and they c lose ly resemble the gener ic A S P at the beginning of this chapter . The re is heavy over lap be tween this group and Appl icat ion Aggrega to rs . There fo re , Microsoft often uses the 'developer ' mon iker to m a k e the dist inction be tween these a n d aggrega to rs b e c a u s e the deve loper has built the solut ion it is sel l ing. In the earl ier days of the market , the provider would own the cus tomer relat ionship. However , providers are becoming d isp laced by the growth of aggregators in manag ing cus tomer relat ionships. Solut ion deve lopers can charge for their se rv ices directly or more often their cha rges are to aggregators . T h e s e cha rges are then p a s s e d on to the cus tomers of the aggregator . 15 Application Aggregators Cor io , U S i and Interliant all fall under this categor izat ion. A l though they do sel l directly they can a lso use V A R s (Value A d d e d Rese l le rs) or traditional sys tem integration providers. T h e agent 's most important roles are the managemen t and ma in tenance of the software being provided a s a hosted serv ice . Aggrega to rs al low cus tomers to take the A S P mode l to a higher level of eff ic iency, by adding simplicity to the A S P cus tomer relat ionship. Cons ide r the fol lowing schemat ic : Figure 4. Comparison Of ASP Market With And Without Aggregators Market without aggregators ISVs have been left out to simplify the d iagram, however s o m e or all of the entit ies in the d iag ram may be deal ing with an ISV. Without an Aggregator involved, the client may have to ou tsource to severa l A S P s . However there are many shor tcomings to this set -up. First of 16 al l , most of the sav ings on the appl icat ion managemen t s ide are sudden ly lost in contact managemen t and integration if the A S P s are not coordinat ing this amongs t themse lves . W h e r e a s with a n aggregator involved, the cl ient c a n provide a wider se lec t ion of se rv ices than any one A S P provider cou ld . Fur thermore, the client only has one contact to m a n a g e , and integration is not an i ssue the client has to dea l with. Aggrega to rs often s imply bundle cos ts together and p a s s them on to the client. For the remainder of this thes is whenever aggregator is referred to, it is a s in the Microsoft use of the word , not Scot t Hein le in 's categor izat ions introduced at the beginning of sect ion 2.2. ASP Agent Agen ts tradit ionally f ocused their bus iness on integration and/or other pro fess iona l serv ices . T o get a better idea of the role that agents play cons ider the relat ionship A n d e r s e n Consu l t ing (now ca l led Accenture) en joyed with S A P . Wi th the prev ious ar rangement , A n d e r s o n contac ted the c o m p a n y and cus tomized the software to opera te at its si te. Instead under the A S P model will determine what p re -packaged appl icat ions are needed and commun ica te the customizat ions needed for the appl icat ion to work at the cl ient 's site. T h e reason we introduce these new categor izat ions is that we will use these (with s o m e slight modif icat ions) a s the bas is of the f ramework we shal l set forth in chapter 4. A l though this mode l d o e s not compar tmenta l ize the A S P industry p layers a s c lean ly a s w e would l ike, we would be hard p ressed to find a s c h e m e that would s ince many p layers in this industry fall under many categor izat ions b e c a u s e they have different interact ions with their c l ients. 17 A New Stratification Of ASPs For the purpose of our ana lys is in chapter 4, we have created a stratif ication b a s e d on market (vertical or horizontal) and s p a n of functionality provided (limited to enterpr ise wide). W e def ine vert ical to be relating to, involving, or integrating economic activity f rom bas ic product ion to point of sa le . Hor izontal is def ined a s bus iness funct ions a c r o s s severa l vert ical markets . Cons ide r the fol lowing two by two matrix, where 'market ' is vert ical or horizontal and ' span ' is the number of bus iness funct ions the se rv i ces prov ided by that A S P support : T a b l e 1. C a t e g o r i z a t i o n B y Marke t A n d S p a n M a r k e t S p a n H o r i z o n t a l V e r t i c a l N i c h e Niche V S P s F o c u s e d Solut ion E n t e r p r i s e Enterpr ise Suppor t Se rv i ces ( E S S ) Full Se rv i ce V S P s T h e s e four group ings are d i s c u s s e d in detail below. Focused Solutions F o c u s e d Solu t ions would so lve one issue within one of the internal vert ical depar tments (for examp le a d e m a n d predict ion algori thm provider would be a F o c u s e d Solut ion within supply cha in management ) . T h e serv ice provided by a F o c u s e d Solut ion wou ld be integrated into an in -house appl icat ion or into the serv ice of any of the other three A S P types. e B a y has recently a n n o u n c e d that it will be joining Microsoft 's . N E T f ramework. Th is m e a n s that any solut ion or serv ice requiring an auct ion eng ine that e B a y prov ides cou ld be a c c e s s e d and integrated into that serv ice a s descr ibed on the Microsoft web site. USinternetwork ing is another A S P who offers its se rv ices in this fash ion . B e c a u s e many c o m p a n i e s are currently 18 adopt ing this f ramework, and few have announced it, it is hard to s a y which will be F o c u s e d Solut ions and which will not. Full Vertical Industry Services Full Se rv i ce Vert ical Se rv i ce Prov iders or V S P s are starting to e m e r g e at a very rapid p a c e . T h e reason for this is s imple : they intimately know both the industry they are in and the software they se l l . Th is results in lower customizat ion cos ts and better market accep tance (if one is in the restaurant bus iness one would want the pe rson providing software to unders tand the restaurant bus iness) . G o v H o s t . c o m , ( focuses on Government ) , A d W a r e S y s t e m s (advertising) and M I R U S (multi-unit restaurant cha ins) a re all examp les of Ful l Se rv i ce V S P s . W h a t a vert ical A S P (also ca l led V S P s ) br ings to the table a s its pr imary attribute is industry expert ise. It offers cus tomers more than just a new way to pu rchase sof tware; it prov ides the speci f ic industry knowledge needed to m a k e a part icular appl icat ion so lve a particular p rob lem. " A n d , s ince every industry has its own set of unique needs , it is this knowledge that will be the differentiator at day 's e n d " [Bernard, J a n 2001] . T h e poss ib le d rawback is that this may el iminate a compet i t ive advantage if adopt ion rate is high within an industry (Cons ider S A P and the petro leum industry). A l though this wou ld m e a n that the c o m p a n y offering this serv ice would be very success fu l , its s u c c e s s cou ld zeni th and start to dec l ine a s S A P ' s for tunes did when market saturat ion w a s reached . B e c a u s e these serv ices are a l ready speci f ic to a part icular industry, industry pre-customizat ion lowers the t ransact ion cos ts of contract ing these se rv i ces . A n d b e c a u s e the operat ing cos ts of the A S P are distr ibuted over many cl ients within one industry, knowledge that is ga ined at other cl ients t ranslates into sav ings for all the cl ients of that A S P . 19 Niche VSPs Niche V S P s are the A S P s that typically offer emai l , account ing p a c k a g e s or supply cha in sof tware; se rv ices that a re broadly appl icab le but would not sat isfy the IT needs of any one organizat ion. N iche V S P s are n a m e d V S P s s ince internal depar tments s u c h a s account ing or human resources are often cons idered vert ical within the company . T h e y al low the cus tomer to focus on core bus iness i ssues (Account ing and f inancia ls c a m e in s e c o n d in cus tomer d e m a n d [Newcomb, Feb 2001]). Ne tLedger is an examp le of a N iche V S P offering account ing appl icat ions over the Internet and Virtual H R is a N iche V S P offering human resources over the Internet. O n e sub grouping of N iche V S P s would be a s noted in the fol lowing table. Table 2. Subtypes Of Niche VSPs Enterpr ise support S e r v i c e s Functional i ty that f ocuses on support ing the enterpr ise and integrating its p r o c e s s e s . Tradi t ional E R P sof tware would naturally grow into this s p a c e . Ana lys i s and Infrequent use (One off) A n y serv ices that are des igned to ana l yse a bus iness prob lem, da ta warehous ing and data min ing, dec is ion support sys tems , wa rehouse layout p lanning sof tware and route planning software. External col laborat ion support Th is inc ludes serv ices that rep lace C R M , e C o m m e r c e and Procurement . Se rv i ces that enab le interaction be tween the client of the A S P and the external wor ld. Internal workf low support Th is inc ludes e-mai l , groupware, knowledge managemen t and document management . Pe rsona l Product ivi ty support Of f ice productivity se rv ices that would rep lace products s u c h a s M S Off ice or Co re l WordPe r fec t (in their current format) Let us explore e a c h of these subtypes. Analysis and Infrequent use: W h y buy an appl icat ion if it is only used o n c e , or very infrequently? W h y buy and support an appl icat ion that is outs ide of the bus iness ' s a rea of exper t ise or that requires hiring new talent? T h e s e quest ions are driving cl ients to ana lys is 20 and infrequent use sof tware. B e c a u s e the client only pays for the t ransact ions it m a k e s and not for a l icence for the software each t ransact ion has a lower cos t assoc ia ted with it. Many c o m p a n i e s use w a r e h o u s e s but few spec ia l i ze in opt imizing their layout, h e n c e buying floor layout sof tware m a k e s little s e n s e for these compan ies . A S P for infrequent use appl icat ions provide a 'per use ' a c c e s s to this appl icat ion. Most compan ies today would like to have a data warehouse for reporting pu rposes , but few have the in-house expert ise to build and maintain one (Cognos ' offer ings are an examp le of such software). A n A S P providing an onl ine data warehouse a l lows these c o m p a n i e s a c c e s s to these se rv i ces while shar ing the cost of infrastructure a c r o s s many cus tomers . Current ly, connect iv i ty i ssues are holding cus tomers back f rom using these se rv i ces directly without an aggregator ; however new technolog ies s u c h a s UDDI (Universa l D i scove ry Descr ip t ion a n d Interface) and S O A P (Simple Object A c c e s s Protocol) cou ld c h a n g e this by reducing the cos ts of coordinat ion and connect ion between new partners, a s we shal l s e e in sect ion 2.6. High specif ic i ty h inders profitability for this market s ince customizat ion cos t s for short- term engagemen ts eas i ly eat into margin [Bushaus , Sep t 2000] . It is notable that e C o m m e r c e and C R M were third and fourth in terms of number of cus tomers p lanning on us ing an A S P to provide a part icular serv ice [Newcomb, F e b 2001]) Long- term ana lys is sof tware f a c e s a different cha l lenge. It is b e c a u s e this software is usual ly very hard to install and cus tomize to each client that A S P s in this a r e a del iver cost sav ings . B e c a u s e A S P s price their software post -customizat ion [Trager, F e b . 2001] they reduce the uncertainty of the client with respect to cos ts of instal lat ion. External collaboration support: External col laborat ion support , s imi lar to infrequent use serv ices , a l lows the outsourc ing of serv ices outs ide the c o m p a n y ' s a rea of expert ise. B e c a u s e of the t ransparency of an A S P ' s serv ices to cus tomers and other entit ies (see 21 cus tomer sect ion in Chap te r 2), the A S P can al leviate the burden of suppor t ing cus tomers , and des ign ing and maintaining a websi te or onl ine store. Cus tomiza t ion of serv ices is cost ing this segmen t by increas ing its operat ing cos ts . C o m p a n i e s s u c h a s P a n d e s i c and R e d Gor i l la both offered e C o m m e r c e serv ices , both t ied their bill ing to t ransact ions v ia the onl ine s tores they had built for cus tomers , a n d both a re now out of b u s i n e s s b e c a u s e they over est imated onl ine purchas ing . Internal workflow support: Internal workf low inc ludes emai l and documen t management . Th is ca tegory inc ludes ema i l , the serv ice that most buyers p lan o n gett ing f rom a n A S P . Most of these serv ices are work enablers , smooth ing the f low of work within the c o m p a n y and h e n c e lowering internal t ransact ion cos ts , and their d iscont inuat ion wou ld c a u s e disruption of work, however the serv ices are very low in specif ic i ty c o m p a r e d to an enterpr ise wide serv ice . Th is type of serv ice has a lot of ups ide for the cus tomer and little potential down s ide. Personal Productivity support: Of f ice productivity se rv ices are se rv i ces that would replace products such a s M S Off ice or Core l WordPer fec t . T h e s e se rv i ces have been s low to grow b e c a u s e they tend to be user interface intensive, which in turn often results in numerous R P C s [ARIBA, 2000 (b)]. O n e solut ion has been to build these appl icat ions us ing D H T M L (Dynamic H T M L ) b e c a u s e it uti l izes the client computer 's power. T h e s e se rv i ces are fairly e a s y to maintain and are relativity e a s y to replace or substi tute if n e e d be . Get t ing these serv ices onl ine prov ides a cost sav ings (you only pay for what you use) and upgrades are s o m e o n e e l se ' s conce rn . B e c a u s e the amount of specif ici ty var ies with the serv ice offered we shal l return to examine N iche V S P s , however s o m e genera l izat ions can be m a d e at this t ime. N iche V S P s lower t ransact ion cos t s in a way simi lar to Ful l V S P s , they spec ia l i ze in o n e a r e a (hence greater 22 doma in knowledge) and enhance the connectivi ty of the product a s much a s poss ib le . T h e appl icat ion their serv ice is b a s e d upon would general ly be outs ide the cl ient 's core a rea of expert ise or bus iness . T h e y al low t ransact ion cost reduct ions by al lowing the client to focus on its key a r e a of expert ise. Enterprise Support Services (ESS) If the A S P a lso falls into the Aggregator or Pure-p lay subtype, it c a n offer a full suite of se rv ices for entire c o m p a n y using best-of -breed solut ions (consider U S i or Cor io ) . If the E S S is an ISV (such a s Orac le ) the solut ion may not be a s end- to-end b e c a u s e the A S P will attempt to only use its own products. E S S s behave the s a m e a s a Ful l V S P with respect to any g iven client; the di f ference is that their cl ients are in var ious industr ies whe reas the V S P s cl ients are all in one industry. E S S , inc ludes M y S A P . c o m and Peop lesof t Onl ine, however these c a n be industry-centr ic. For example , S A P ' s sof tware is u s e d a c r o s s many industr ies, however S A P can c o m e in industry speci f ic f lavours (which may be cons ide red vertical). T h e s e c o m p a n i e s s u c c e e d b a s e d on the huge t ransact ion cos t sav ings factor they provide through simpli f icat ion of cl ient 's IT relat ionships (ideally one E S S would del iver all the cl ient 's appl icat ion needs by repackag ing and integrating appl icat ions f rom numerous sources) giving the client one point of contact. La rge and mids ize c o m p a n i e s are their most c o m m o n cus tomers . T h e largest d rawback to their s u c c e s s is the vital role they would play in client operat ions [Bushaus , J a n 2001] . C u s t o m e r s real ize these solut ions have very high specif ic i ty and are vital to cont inued operat ions o n c e they are instal led, and cus tomers may be wary of hav ing s u c h a d e p e n d e n c e . A n extens ive review of the literature reveals that cus tomers tend to examine the A S P offering these serv ices for f inancial stability to gauge whether or not they will fai l . Often these solut ions d isp lace in-house IT staff a s well a s further increas ing that lock-in may 2 3 al ready high due to specif ici ty. C u s t o m e r fears of lock- in to a part icular serv ice provider comb ined with cus tomer conce rns about the life span of the A S P has held this market back. Here is what we have learned f rom recent A S P exper iences : Full V S P s are succeed ing in solv ing the needs of cl ients in their part icular industr ies. T h e market is shap ing up s o that N iche V S P s (e.g. an account ing sys tem provider) will often have their se rv i ces resold or aggrega ted and then so ld to cl ients [ M c C a b e , F e b 2001] . A l though there are a large number of V S P s (19% of A S P s were V S P s in 2000) , few are in direct compet i t ion with one another [ S C N , 2000] . T h e aggregator mode l that s o m e Full V S P s and most E S S providers fol low s e e m s to be the most success fu l in the market today. Wh i le only 3 . 6% of c o m p a n i e s p lanned on purchas ing an onl ine E R P sys tem [Newcomb, F e b 2001] , U S i is the industry leader. W h y is this? B e c a u s e the aggregator model a l lows 'one-stop shopp ing ' and flexibility that none of the other mode ls provide, espec ia l ly if the A S P is not a lso an ISV or is not too commit ted to one software vendor 's products. Aggrega to rs a lso al low a cus tomer to ' e a s e ' into A S P serv ices 'step by s tep ' rather then having a 'sink or sw im ' date. Th is is related back to specif ic i ty i ssues regarding lock- in and customizat ion (which are d i s c u s s e d in Chap te r 4 and 5). Service Level Agreement (SLA) S i n c e A S P s , by their very name, are the providers of a serv ice and play an outsourc ing role, we would be remiss to not cons ider the role Serv i ce Leve l A g r e e m e n t s play in the relat ionship be tween client and A S P . Tradit ionally, an S L A would conta in metr ics by which an outsourcer (and somet imes the internal IT department) wou ld be m e a s u r e d [Butler, 2000] . T h e y would have goa ls b a s e d on those metr ics. In an A S P envi ronment , the metr ics inc lude, but a re not l imited to, sys tem availabil ity, r esponse t imes, i ssue priority and quality s tandards [Butler, 2000] . For example , a cal l centre cou ld p romise to respond with a 24 solut ion to an IT prob lem within two hours for 6 5 % of all ca l ls . S L A s c a n provide the bas is for measur ing the outsourcer 's per formance by outl ining the metr ics by which they m e a s u r e d . However , when deal ing with appl icat ions, c lear-cut metr ics gaug ing functionality c a n be hard to c o m e by. Issues regarding chang ing bus iness needs c a n a lmost never be totally i ronclad into the S L A . Fur thermore, many compan ies that will be us ing A S P s will be outsourc ing for the first t ime. S L A s have not el iminated prob lems in the tradit ional outsourc ing world and they will not el iminate confl ict in the A S P realm. That having been sa id , S L A s go a long way toward el iminat ing unstructured confl icts and providing a f ramework within wh ich to resolve these conf l icts. E v e n the very act of forcing the part ies to sit down and lay out their expectat ions on paper make S L A s benef ic ia l . W e ment ion S L A s only to m a k e the reader aware of their ex is tence and poss ib le impact on A S P / u s e r relat ions. O n e interesting start-up in the field of S L A managemen t is the c o m p a n y Obl icore . Founded in Janua ry of 2000 , Ob l icore purports to provide A S P s and m a n a g e m e n t serv ice providers with the tools to draft measurab le S L A s for all cus tomers . T h e Ob l i core software will measu re the resources used to meet each S L A and help the A S P concent ra te on fixing prob lem S L A s without affecting other cus tomers ' serv ice [Dubie, 2000] . Ob l icore s a y s its sof tware tells A S P s the appl icat ion response t ime and the levels of network availabil i ty and latency the A S P can guarantee. Obl icore will a l so give serv ice providers and their cus tomers a v iew into the network and how its performing in terms of the S L A requi rements throughout the billing cyc le . If p rob lems are encounte red , the A S P c a n take act ion and poss ib ly avoid penal t ies. Thus , S L A structuring is now of fered a s a serv ice , to serv ice providers. For further information regarding S L A s the reader is referred to Outsourc ing : A C I Q ' s Perspec t ive by O a k i e Wi l l iams. 25 2.5. Comparing The ASP Model To In House Solutions At first g lance , the A S P model bears many striking similari t ies to the main f rame model of o ld . It is ana logous to the in-house network getting rep laced with the Internet. However A S P s are much more than that. A S P s have the capabi l i ty to del iver the high power GUI appl icat ions users have b e c o m e accus tomed to whereas ma in f rames have s tayed distinctly on the number crunch ing s ide of the street [Slavid, J a n . 2001] . A S P del ivery methods and Cl ien t /Server sha re a lot of the networking componen ts , however p rocess ing and data are scat tered all over the sys tem. A S P s update software regularly and automat ical ly (as far a s the user is concerned) [King, Sep t . 2000] . C lear ly all three mode ls have c o m m o n traits, however they clear ly have s o m e very different character is t ics a s wel l . Let us compare the benef i ts and d rawbacks of A S P s a s compared with sys tems that have been des igned and built ' in -house ' . W e start by examin ing the benef i ts the user of an A S P c a n expect and c o m p a r e that to the benef i ts of a n in house sys tem. T h e fol lowing table c o m p a r e s the A S P model to the tradit ional sof tware del ivery model b a s e d on the strength of each mode l . 26 Table 3. Comparison Between ASP And Traditional IT Benefits of the ASP model Benefits of Traditional Model R e d u c e d in-house IT d e m a n d s Contro l over server s ide archi tecture S a v e d M o n e y and T ime More cus tom report ing N o Addi t ional Server software or hardware Better per fo rmance of C P U intensive or heavyweight R P C p rograms F e w client appl icat ions to support Customizabi l i ty Un iversa l A c c e s s Better offline per fo rmance Shor ter Appl icat ion C y c l e s Commun ica t ion intensive appl icat ions E a s e of Integration No o b s o l e s c e n c e Better distr ibuted per fo rmance Better cus tomer serv ice than in-house Better for handl ing c h a n g e s in d e m a n d levels Th is table revea ls that the A S P provides more benef i ts, however the d rawbacks revealed by the st rengths of the older sys tem can be difficult to ignore. C u s t o m reporting and customizat ion are th ings that c a n eat into an A S P s margin b e c a u s e an A S P s profitability d e c r e a s e s with the customizat ion each client requires [ M c C a b e , Nov. 2000] . A S P s der ive income from the 'repeatabil i ty' of their instal lations [ M c C a b e , Nov. 2000] . Commun ica t i on , or bus intensive appl icat ions such a s graph ics p rograms and C A D do not readi ly lend 27 t hemse lves to the A S P mode ls either. T h e A S P can provide file and da ta s torage support to these appl icat ions, however it d o e s not provide a v iable rep lacement for these appl icat ions. 2.6. Technological Problems And Advances A S P s have s o m e technica l cons iderat ions that still have to be a d d r e s s e d . M a n y of the techno log ies to maintain and support A S P s more efficiently are still be ing deve loped . Th ree of the top technolog ica l p rob lems that have impact on cus tomer a c c e p t a n c e and appl icat ion per fo rmance and are currently holding back the growth of A S P s are a s fo l lows: 1) conce rns over appl icat ion per fo rmance b e c a u s e of networking i ssues 2) T h e lack of connectivi ty s tandards between appl icat ions often m a k e s integration of se rv i ces expens ive and 3) Most sof tware w a s not deve loped for the web and its onl ine per fo rmance is lack ing a s a result. W e d i s c u s s these i ssues b e c a u s e there are factors in s o m e cl ient 's dec is ion a s to whether or not to pu rchase the serv ices of an A S P . Let us examine e a c h of t hese prob lems and d i s c u s s poss ib le solut ions. Network Performance Network ing Issues inc lude both response t ime uncertainty and reliability. A l though the Internet is at a s tage in its deve lopment where it is ab le to act a s a del ivery channe l for software, it is by no m e a n s an ideal del ivery platform. The re are too many uncontrol led factors for any one A S P , or even any ISP to m a k e a serv ice level guaran tee with regard to response t imes that it can really back up (other than to try to max im ize the amount of bandwidth avai lab le to the user and hope for the best). N e e d l e s s to say , this is not an acceptab le way of deal ing with this i ssue if the appl icat ions that an A S P ' s cus tomers are us ing are bandwidth intensive. Connec t i on s p e e d between any two points on the Internet is determined by the s lowest point on the route c h o s e n by the 28 routers between those two points. Wi th current technology it is not poss ib le to 'priorit ize' data to insure cus tomers requiring a high 'Qual i ty of Se rv i ce ' ( Q o S ) . Th is m e a n s that a l though other high priority has to get there immediately, da ta like v ideoconferenc ing has to compe te with lower priority traffic on the net. Th is prob lem is be ing dealt with by serv ices such a s mult i- label protocol switching ( M L P S ) and Dif fServ. Both a re a imed at al lowing vo ice and data to co-exist on the s a m e network, and a s author Mark Va i l puts it: The emergence of M P L S and DiffServ is enabling new packet-based access solutions that can support and deliver toll-quality voice and data services. M P L S and DiffServ together enable network operators to maintain a guaranteed QoS for IP, legacy voice and data transmissions while providing tools for managing traffic, deploying and monitoring SLAs and provisioning secure data VPNs across IP backbones. Ultimately, M P L S and DiffServ will enable service providers to provision QoS across entire networks. [Vail, Jan. 2000] T h e s e se rv i ces a lso al low for a c c e s s quality to be dial led up or down depend ing on client needs , thus te lcos would be able to del iver var ious guaran teed levels of serv ice and bill e a c h cus tomer segmen t for the per formance it d e m a n d s , or even bill for high Q o S t ime per iods in a fash ion similar to the way te lcos bill for long d is tance [Skemer , 2000] . W h e n these se rv i ces b e c o m e avai lable, I S P s can offer A S P s and their cus tomers guaran teed S L A s . Th is will al leviate most cus tomer conce rns with respect to response t ime and reliability and thus more cl ients with s e e A S P s a s a v iable alternative to in-house sys tems Lack Of Connectivity Standards Between Applications T h e s e drive up cos ts for any IT user or provider, whether the appl icat ions are onl ine or not. E a s y integration would be of particular benefit to the aggregator or any client looking to ou tsource only s o m e of its IT appl icat ion needs . 29 N e w e n a b l e r s : UDDI a n d S O A P With the e m e r g e n c e of bus iness over the Internet, severa l bus i ness prob lems e m e r g e d for buyers and se l lers of goods and serv ices . N e w suppl iers and buyers were on the Internet, however f inding them, be ing sure of who they were and identifying what they are looking for or offering cou ld be an onerous chore . N e w industry-wide pane ls c a m e into ex is tence to determine how to so lve these prob lems. Consor t iums such a s W 3 C ( W 3 Consor t ium) and UDDI .org (Universal Descr ip t ion, D iscovery and Integration) which is a joint panel fo rmed by Ar iba , IBM and Microsoft) [Ariba, 2001(a)] took on these prob lems and c a m e forward with the UDDI spec i f ica t ion. UDDI offers b u s i n e s s e s a s tandard ized w a y to connec t a n d interact over the Internet. Here is how UDDI def ines itself: To address this challenge, a group of technology and business leaders have come together to develop the Universal Description, Discovery and Integration [UDDI] specification - a sweeping initiative that creates a global, platform-independent, open framework to enable businesses to (1) discover each other, (2) define how they interact over the Internet, and (3) share information in a global registry that will more rapidly accelerate the global adoption of B2B eCommerce. Each incremental advance in Web-enabled commerce has carried deep implications for business processes and organizational culture. UDDI is a major advance - the first cross-industry effort driven by platform providers, software developers, marketplace operators, and eCommerce and business leaders that comprehensively addresses the problems limiting the growth of B2B eCommerce, and that will benefit businesses of all sizes by creating this global, platform-independent, open framework. UDDI is a building block to enable businesses to quickly, easily and dynamically find and transact with one another via their preferred applications. [Ariba, 2001(a)] T h e UDDI speci f icat ion is b a s e d on the W 3 C ' s and Internet Eng ineer ing T a s k Fo rce ' s ( IETF) s tandards s u c h a s X M L (extens ib le Mark-up Language) and S O A P (S imple Object A c c e s s Protocol) , which we shal l d i s cuss momentar i ly. Wi th the UDDI s tandard , A S P s have a formal ized way to descr ibe their interface, and serv ices they provide. 30 Th is , comb ined with the a d v a n c e s in the software deve lopment a rena , a l lows for the A S P s that were aggrega ted by providers such a s U S i and Cor io , to now be 'd i scovered ' by cl ients and incorporated into solut ions that are built in -house with ou tsourced appl icat ions. S O A P is an industry wide rep lacement for Remo te P rocedu re Ca l l ( R P C ) mode ls such a s C O R B A and C O M (or D C O M ) . S O A P uses no new technology, and is currently suppor ted by both A p a c h e and IIS web servers . S O A P uses an http (hypertext transport protocol) a s a transport and X M L a s a data descr iptor [Microsoft, 2001] . Tradit ional ly, when one would connect to a 'server farm' over the net, http would be used a s the transport and then C O M or C O B R A would take over the transport inside the farm. S O A P , o v e r c o m e s the 'cultural ' r easons for use rs to stay with one R P C or the other. Th is a l lows for a new level of interconnectivi ty be tween appl icat ions over an http connect ion (note that this is not just l imited to the Internet itself). Th is comb ined with the UDDI f ramework a l lows for compan ies offering appl icat ions over the Internet to dynamica l ly link a n d opera te together [Ar iba, 2000(b)]. Th is is a boon for compan ies offering low criticality, commodi ty or s o ca l led 'one of f appl icat ions. Most Software Was Not Developed For The Web Most of the offer ings that A S P s are providing are s imply o lder appl icat ions with a web enab lement layer built onto them [Dering, J a n . 2001] . M a n y of the e c o n o m i e s of sca le that A S P s have been promis ing were lost in trying to del iver and interact with the software in a way for which it w a s not des igned . Sof tware vendors have started retool ing their offer ings to perform in this new environment. Citr ix is the chief proponent of this app roach and its sof tware ports exist ing software for the net. Microsof t is head ing the other way with its . N E T initiative. . N E T a l lows the bui lding of w e b native appl icat ions that utilize UDDI and S O A P to their ful lest [Microsoft, Nov . 2000] . T h e 31 . N E T deve lopment envi ronment is des igned s o that onl ine se rv i ces c a n be used in a fash ion similar to the way c l a s s l ibraries are currently used in the C++ envi ronment . Many c la im that . N E T and UDDI represent the model for future software deve lopment [Kerstetter, F e b . 2 0 0 1 ; F o n s e c a , Nov. 2000 ; Kucharvy , May 2000] . Microsoft is fac ing stiff compet i t ion f rom S u n and Orac le a s those vendors retool their offer ings a s well [Kerstetter, F e b . 2001] . Many analys ts are predict ing that the major p layers will have to do more than just repackage their offer ings for web sa les . Gar tner 's B e n Pr ing, lead A S P researcher at Gar tner , has stated the fol lowing: 2001 will be the year of net-native ASP solutions. The year in which the realization becomes apparent that ASP is not really an ERP play at all but just a stepping-stone from an ERP world to an Internet world. Net native applications fly on the current (let alone future) bandwidth in place in a way that Citrixised ERP applications will not be able to match. [Pring, Aug. 2000] This new connect iv i ty is someth ing that A S P s ignore at their own risk either way. Orac le is 'going it a lone ' in the A S P world, offering its software serv ices directly to the cus tomer . T h e prob lem? Quot ing Art Wi l l i ams, a director at G i g a Information G r o u p : Some analysts say Oracle should get rid of Business Online altogether and simply use ASP partners. "They need to realize the real ASP opportunity lies in accelerating the uptake of their applications via the use of partnerships," Williams says. "Oracle will never satisfy customers, because they only offer their own applications." [Maselli, Oct. 2000] Orac le is ignoring two market t rends: 1) In a interconnected wor ld the users will sea rch for best -of -breed [Dering, 2000] and 2) U s e r s are not going to be a s interested in brand n a m e s a s they are interested in functionality and the cons is tency of its del ivery [Kerstetter, Mar 2 0 0 1 ; Sper l ing , J a n . 20001] . 32 Aggrega to rs ignore this threat at their own peril a s wel l . U D D I / S O A P and . N E T type deve lopment methodo log ies al low users to find and plug into A S P s directly. A l though this is now poss ib le , it is the belief of the author that U D D I / S O A P will a l low u s e r s that a re hung up on sunk cos ts to ' e a s e ' into us ing A S P s p iece by p iece rather than having to immediate ly make the jump. Th is would satisfy the 'sunk cost effect', which shal l be def ined in Chap te r 3, b e c a u s e exist ing investments would be p h a s e d out due to their a g e . T h e s e c h a n g e s in techno logy and market are coming about at the right t ime to enab le A S P s to b e c o m e accep ted a s relevant p layers in the software del ivery market . T h e c h a n g e that is occurr ing in cus tomer attitude toward A S P s is an interesting one and it is unc lear what stature the I S V s will hold in this new market . 2.7. Examples Of ASPs This sect ion is ded ica ted to looking at compan ies out in the market today or in days gone by. W e start by looking at the market share of the larger A S P s . ASP Market Shares It is very difficult to accura te ly track the revenues genera ted in the A S P s p a c e s i nce there are few A S P only c o m p a n i e s and A S P revenues are blurred a s they are often bundled with other se rv i ces the c o m p a n y offers [ S C N , 2000] . S o in order to g a u g e market sha re we present two char ts extracted f rom A S P : Appl icat ion Serv i ce Prov id ing by S C N Educa t ion . T h e first represents the market share by the number of cus tomers a n d the s e c o n d shows market sha re by number of contracts (to account for cus tomers buying more than one serv ice) . Here we s e e market share by number of cus tomers 3 3 Figure 5. Market Share By Number Of Customers [SCN, 2000] T h e fol lowing f igure presents the number of contracts: Figure 6. Market Share By Number Of Contracts [SCN, 2000] Interpath, App l icas t , Q w e s t and more are inc luded in the other ca tegor ies . Many analys ts attribute the s u c c e s s of USinternetwork ing (US i is an appl icat ion aggregator reseller) to its having started an A S P model with a c lean slate and having built the c o m p a n y on a pure 34 A S P foundat ion. USinternetwork ing 's chief advan tage is big n a m e cl ients " U S i has been better at penetrat ing Fortune 50 compan ies " s a y s Burney of C a h n e r s In-stat [King, Sep t 2000] . U S i ' s i M A P serv ices have gone the opposi te route of B r e a k a w a y and Intacct b e c a u s e it has bought its data cent res. U S i offers enterpr ise appl icat ions f rom the big n a m e s s u c h a s Microsoft , Orac le and Peoplesof t . U S i fee ls that in order to get those big n a m e s the c o m p a n y has to own its data cent res [King, Sep t 2000] . M iche le Perry, sen ior V P of market ing at U S i , c la ims this s u c c e s s is b a s e d on the fol lowing three p romises U S i m a k e s to its c l ients: " W e guarantee what it will cost , when you' l l get it a n d that we'l l keep it running for the next three years , " s h e sa id in a Sep tembe r 2 0 0 0 interview. In November of 2000 U S i struck a dea l with Microsoft where in exchange for $50 mill ion in funding U S i ag reed to give Microsof t 's . N E T m a n a g e d appl icat ions serv ices platform [ F o n s e c a , 2000] . U S i ' s fourth quarter loss in 2000 w a s $12 mil l ion, down a third f rom Q 3 ' s $18 mill ion loss [Wainewright, F e b 2001] . B reakaway solut ions b e c a m e profitable in a hurry, well before most of its compet i tors. B r e a k a w a y began life a s a sys tems integrator and hopes to cont inue on a s an Aggregator A S P . Host ing revenues compr i sed only 1 5 % of B reakaway ' s $32 .5 mill ion revenue in the third quarter of 2000 . " W e ' d like to get revenue up to 3 0 % to 4 0 % of total revenue base , " Dev Ittycheria, sen ior V P of the A S P group at B reakaway [King, Sep t 2000] . T h e c o m p a n y has a g ross operat ing margin of 5 5 % and has done this by keep ing its cost structure low. T o do this, B r e a k a w a y has dec ided not to build its own data cent res and instead leases s p a c e [King, Sep t 2000] . T h e c o m p a n y has grown f rom 120 peop le to 900 emp loyees in little over a year . B reakaway bui lds solut ions tai lored to e a c h cus tomer b a s e d on software a l ready deve loped . B reakaway c la ims each new engagemen t usual ly requires a 2 0 % customizat ion of the software, however it is unclear a s to how this metr ic is g a u g e d . 35 Unfortunately, B r e a k a w a y has not remained profitable. Part of its funding agreement with Internet Capi ta l G r o u p forced B r e a k a w a y to take on dot corns that were a lso funded by I C G . Th is in turn s q u e e z e d out more viable potential cus tomers that B r e a k a w a y w a s cons ider ing . Wi th the dot corns die-out, many of B reakaways cus tomers turned out to be insolvent, dragging B r e a k a w a y back to the red s ide of the ledger [Bushaus , F e b . 2001] . Th is is a prob lem to wh ich many A S P s attribute poor year 2000 fourth quarter per fo rmance. Orac le at tempted the move f rom ISV to A S P relativity ear ly [G reenbaum, F e b . 2000] and has wave red between maintaining its A S P s tance or moving to the backg round and taking a more tradit ional ISV s tance [Masel l i , Oct . 2000] . Many analys ts were c la iming that Orac le shou ld drop its B u s i n e s s On l ine div is ion a n d al low A S P s s u c h a s U S i a n d Co r i o to se l l O rac le ' s appl icat ions, however Orac le has made severa l large s teps in shor ing up its A S P a rm by restructuring the S L A s accompany ing its contracts with cl ients [Masel l i , Oct . 2000] . T h e new agreement offers very generous serv ice guaran tees to cl ients that are b a c k e d by a 2 5 % rebate on monthly f ees . In late February 2 0 0 1 , O r a c l e l aunched onl ine tools for maintaining onl ine s tores just w e e k s after launching an entirely web b a s e d cus tomer support appl icat ion. T h e s e tools move Orac le c loser to providing users with an entire suite of appl icat ions a c c e s s e d only v ia the Orac le web site, rather than be hos ted by an A S P . Cor io is an Aggrega to r A S P with a consul t ing background . Cor io offers E R P appl icat ions f rom Peop lesof t and S A P . B e c a u s e Cor io cus tomers were most ly dot corns Cor io has yet to s e e a profit f rom its A S P serv ices . However , despi te the l osses , Mark Ve rbeck , an analyst with E p o c h Par tners , s a y s Cor io has s tayed on track by shift ing toward recurr ing appl icat ion-managemen t revenue while partnering with compan ies such a s C a p G e m i n i Ernst & Y o u n g for implementat ion se rv ices . T h e partnership lets Cor io win larger cus tomers , helping boost 36 its ave rage new-cus tomer contract to approximately $100 ,000 per month [Masel l i , F e b . 2001] . FutureLink del ivers software from E R P software f rom Citrix (in fact it is a top reseller) and has par tnerships with Microsoft and Grea t P la ins - recently Microsoft a n n o u n c e d p lans to buy Grea t P la ins [Alexander, J a n . 2001] . However , FutureL ink 's future is unclear . In the fall of 2000 , the L a k e Forest Cal i forn ia, c o m p a n y announced minor layoffs, most of which were to el iminate redundancy c a u s e d by an aggress i ve acquis i t ion strategy dur ing the previous 18 months. But it has yet to integrate its c o m p a n i e s effectively into one cohes ive organizat ion, and it must further expand its p resence outs ide the Uni ted S ta tes to stay up to s p e e d [Will is, F e b . 2001] . Other interesting p layers include: Qwes t Cyber .so lu t ions , a joint venture between Qwes t commun ica t ions and K P M G a s descr ibed previously. Intacct f o c u s e s on speci f ic vert ical markets or spec i f ic horizontal appl icat ions. Intacct main product is focus ing on W e b - b a s e d account ing for fast-growing compan ies . Intacct is not bui lding da ta cen t res ; instead it is focus ing its efforts on cus tomer serv ice. Pr ic ing starts at a relativity low cost of $50 per user month. E v e n at this pr ice, Intacct m a k e s an 8 0 % g ross marg in. Interliant h a s been an A S P s ince before the term A S P w a s co ined by IDC - in 1994: Interliant w a s host ing appl icat ions. Interliant is currently E B I T D A negat ive at negat ive 19.1 mill ion mainly attr ibuted to a great dea l of recent capital expendi tures. However , Interliant has garnered a weal th of knowledge about host ing and has a wide client base . [King, Sept . 2000] . W e round out this list of examp les by looking at where the major p layers in the market fall under the categor izat ion s c h e m e created earl ier in this chapter however this list is by no m e a n s exhaust ive . Fur thermore, let us segment the cus tomer b a s e by the s i ze that the A S P is target ing: 37 T a b l e 4. A S P s B y T y p e A n d T a r g e t C l i en t S i z e \ C l i e n t S i z e S m a l l M e d i u m L a r g e A S P T y p e \ S O H O ( less than 100 employees) (Between 100 and 1000 employees) (More than 1000 employees) eBay, USi eBay, USi eBay, USi F o c u s e d S o l u t i o n F u l l V e r t i c a l S e r v i c e P r o v i d e r Javelinx, atyourbusiness.com Javelinx, MIRUS, RetailAspect, atyourbusiness.com Javelinx, govHost.com N i c h e V e r t i c a l S e r v i c e P r o v i d e r Netledger, WebTurbo Tax NetLedger, NetAbacus, Virtual HR, Telecomputing, WebTurbo Tax, e-Benefits, Applicast, CoreHarbor, Salesforce.com Telecomputing, Trizetto, employease.com, Learningstation.com, NetAbacus, FutureLink, Breakaway, Applicast, CoreHarbor, Salesforce.com Telecomputing, Trizetto, employease.com E n t e r p r i s e S u p p o r t S e r v i c e s FutureLink, Corio, Peoplesoft FutureLink, Usinternetworking, mySAP.com, Corio, Applicast, eOnline, Peoplesoft Usinternetworking, mySAP.com Corio, AristaSoft, Oracle BO, Cyber.solutions, Peoplesoft Note: information f rom var ious sou rces throughout bibl iography and c o m p a n y webs i tes 38 Chapter 3. The A S P Market Th is chapter prov ides the motivation for examin ing the A S P p h e n o m e n o n by looking at the market in terms of revenue and investment. A s the reader shal l s e e , the market s i ze project ions are very volati le. However , all of them are fairly large, mak ing A S P s of interest to investors and a c a d e m i c s al ike. A l s o highl ighted is the apparent confus ion amongs t the ana lys ts of this market. T h e chapter conc ludes with an ana lys is of current market t rends and short- term project ions and predict ions for the market. 3.1. How Big Is The Market For Asps? This sect ion examines the s i ze of the A S P market in terms of revenue and investment. It shal l b e c o m e c lear that none of the 'experts ' know what the market is going to do in the short-run. How Much Revenue Is There For Asps? Author D a w n B u s h a u s s u m s up the confus ion a m o n g industry ana lys ts with the fol lowing statement: " W h e n it c o m e s to A S P s , the only thing analys ts c a n ag ree on is that the market will cont inue to grow" [Bushaus , Sep t 2000] . T h e fol lowing table w a s compl ied f rom var ious sou rces throughout the bibl iography and provides a snapsho t of project ions m a d e in the past with regard to how big the A S P market would b e c o m e , in te rms of revenue. 39 Table 5. The Changing Projections Of ASP Revenue By Major Research Corporations Year Pre-November 2000 (in bill ions) Post-November 2000 (in bil l ions) 2001 Forrester : $23 B (old projection) 2002 G i g a : $2-$6 B 2003 Gar tner : $22.7 B IDC: $2 B $23 Bil l ion by 2 0 0 3 over by severa l billion - B e n Pr ing , Gar tner L e a d A S P R e s e a r c h e r Forrester : $10 B Detoitte: $48 .5 B Yankee Group : $19.2 B 2004 Gar tner : $ 2 5 B Frost & Su l l i van: $12 - $17 B IDC: $7 .75 B Gar tner : $25 .3 B 2006 Frost & Su l l i van: $25 B Project ions m a d e prior to the spr ing 2000 market correct ion tended to be much higher then predict ions m a d e afterward. However , even after the correct ion, many ana lys ts s tood by their predict ions. Th ree factors c a m e into play in this behaviour : 1) Ana lys ts felt that dot corns were overva lued and the depress ion this c a u s e d in the market w a s temporary [Trager, D e c . 2000] 2) A S P s looked like sol id bus iness mode ls when c o m p a r e d to b u s i n e s s e s like S o c k s . c o m that so ld s o c k s onl ine 3) T h e y did not real ize all IT b u s i n e s s e s would be painted with the s a m e brush a s dot corns by cus tomers and investors [Trager, D e c . 2000] . In D e c e m b e r of 2000 , the A S P market project ions started to contract drast ical ly b e c a u s e of the fact that the chil l on IT investment and purchas ing did not lift and b e c a u s e talk of a recess ion in the U S m a d e analys ts think the set b a c k s that first s h o w e d up over the s u m m e r would be a round longer then they had ant ic ipated [Trager, D e c 2000] . 40 S i n c e most A S P s are fairly young , they need a c a s h rich envi ronment in order to surv ive the negat ive c a s h f low start up per iod. A s a result, many good appl icat ions will, and currently are, be ing so ld at wel l be low what their market va lue would be otherwise. Th is low pr ice exists not b e c a u s e the of the shor tcomings of those appl icat ions, but b e c a u s e the c o m p a n y that c rea ted it d id not have the n e e d e d capital to surv ive this contract ion [ M c C a b e , Nov . 2000] . C o n s i d e r the fol lowing s tatements Frost & Sullivan's Stan Prescott is forthright enough to wish he could get back his hanging curve of a projection. The company is on record predicting $26 billion in 2006 revenue for an industry broadly defined as including hosting and managed services...'Now that we're all seeing a slowdown in the economy, I might want to pull that back a little bit.' Prescott says. T m seeing numbers tossed around for 2004. The high numbers that are realistic are $17 billion, and the low numbers are around $12 billion.' [Trager, Dec. 2000] There is apparent confus ion even at Gar tner : "Pr ing w a s quoted last Ma rch [of 2000] a s say ing d isappoint ing d e m a n d m a d e his predict ion of $22.7 bill ion in A S P revenues by 2003 overb lown by severa l bil l ion. But in Augus t [of 2000] , Dataques t c a m e right back with a $25 .3 bill ion target for 2004" [Trager, D e c . 2000] For the majority of 2000 , Gar tner G roup repeatedly made the fol lowing project ions: Of the current 480 A S P s , 6 0 % (288) will be out of bus iness by the end of 2 0 0 1 , 2 0 % will survive a s prov iders of A S P products or vert ical se rv i ces in 2004 , 1 6 % Wi l l b e bought out by another A S P or be out of bus iness by 2004 , 4 % (20) will still be in bus iness a s ful l -service retail A S P s in 2004 . "Industry watchers general ly agree that the A S P bus iness mode l is worth the hype, but they caut ion that there is going to be a shake-ou t soon that will leave only a few serv ice providers s tanding. " [Bushaus , Nov. 2000] 41 This is comb ined with a currently lukewarm cus tomer response to the industry "An electronic aud ience poll dur ing the State of the Industry pane l [in S e p t e m b e r of 2000] indicated that 60 percent of the severa l hundred solut ion providers in a t tendance were act ively deve lop ing an A S P strategy. However , only 31 percent sa id they v iewed the A S P mode l a s 'extremely important' to their b u s i n e s s e s currently." [Longwel l , 2000] O r cons ider the damn ing with faint pra ise language G o l d m a n S a c h s analyst G r e g o r y G o u l d uses in a note to cl ients on newly publ ic A S P Cor io , w h o s e stock t rades a round $ 1 1 , three dol lars be low its late Ju ly offering pr ice. "Al though the A S P opportunity m a k e s s e n s e , we bel ieve it is n e c e s s a r y to exerc ise caut ion when a s s e s s i n g the speci f ic bus i ness mode ls , many of which are young and likely f lawed." [Lashinksy, 2000] Many potential cus tomers are wait ing on the s ide l ines watch ing to s e e who will survive this s tage of the market. However Mark Chestnut , who is Microsof t 's director bus iness deve lopment efforts with A S P s and I S P s , had this to s a y in a recent art icle in T h e Net E c o n o m y : " C u s t o m e r s are delay ing pu rchase dec is ions b e c a u s e they don't know whether the A S P s will s u r v i v e . . . A S P s aren't getting enough cus tomers to surv ive" [Trager, F e b 2001] . However , the te lcos, ana lys ts , authors and ISVs all maintain that this is the software del ivery model of the future. Who Is Investing And How Much? Figure 7 s h o w s V C (Venture Capi ta l ) investment in commun ica t ions star tups: 42 F i g u r e 7. C o m m u n i c a t i o n s S t a r t u p s V e n t u r e F u n d i n g B y Industry S e g m e n t 1Q2000 2Q2000 8.1% 45.6% 35.9% 38.6% ED Equipment and infrastructure • A S P s B E S P s S o u r c e : Eas te rn M a n a g e m e n t G r o u p [Blum, Sept . 2000] T h e s e char ts represent V C investment in commun ica t ions star tups. E v e n though the chart on the right represents four t imes the amount on of capi tal of that on the left, a s a percen tage A S P s have lost ground. Equ ipment and infrastructure inc lude network and b roadband del ivery equipment . E S P s (Emerg ing Se rv i ce Prov iders) are essent ia l ly Compet i t ive Loca l E x c h a n g e C o m p a n i e s ( C L E C s ) , providing mult iserv ice a c c e s s serv ices and integrated providers such a s in-Bui lding serv ice providers Eas te rn M a n a g e m e n t G roup , which monitors te lecom industry V C investment a s part of its Venture C o m p a s s serv ice reports V C investment in A S P s actual ly dec l ined f rom $1.54 billion in the first quarter to just under $1.21 billion in the s e c o n d quarter. Meanwhi le , investment in compet i t ive serv ice providers inc reased nearly fourfold over the s a m e per iod, f rom $446 mill ion to more than $1.81 bil l ion. F rom an interview with Rober t S a u n d e r s , an analyst at Eas te rn Managemen t , Jona than B lum reported the fol lowing: 4 3 'There is no question that the ASP sector has taken a hit,' says Saunders. 'Though ASP deals are still getting done, the V C community clearly feels that the loss of equity valuation for ASPs and the fewer IPOs coming to market have eroded the value of these businesses. The spring [2000] correction was the clear cause...whether the trend away from ASPs holds over the long run remains to be seen. I would not count ASPs out,' Saunders says. 'But clearly the market is cycling away from those businesses as of now.' [Blum, Sept 2000] Many industry watchers are speculat ing that the fall ing share pr ices of A S P s may al low big te lecoms like A T & T , W o r l d C o m and Sprint (all of w h o m have gone on record stat ing they would not b e c o m e A S P s ) to buy A S P s at who lesa le pr ices [Bushaus , D e c 2000] . T h e main push for buying A S P s is that the bas ic host ing offered by most te lcos is quick ly becoming a commodi ty , with rapidly shr inking marg ins and alternative, new s o u r c e s of revenue are likely being cons ide red . B e c a u s e A T & T and Sprint are weak in web host ing serv ices , they wou ld f ind A S P s s u c h a s U S i attractive b e c a u s e it owns its own Data Cen t res . W o r l d C o m , which is a l ready es tab l ished a s a top host ing provider through its U U N e t subs id iary and Digex would likely prefer A S P s s u c h a s B reakaway or Cor io b e c a u s e it d o e s not need to buy more data cent res , and both of these A S P s lease third party data cent res [Bushaus , D e c 2000] . A l though rarely mak ing any direct c a s h investments, Independent Sof tware V e n d o r s ( ISVs) are mak ing a signif icant investment in product re-engineer ing and training (see m y s a p . c o m , Microsof t 's . N E T , S u n ' s Suntone) . They are a lso lending their n a m e s , and hence their credibil ity, to startup A S P s that have p a s s e d their certif ication p rograms. 3.2. Market Trends W e start by examin ing what appl icat ions cl ients a re seek ing . Commun ica t i ons , f inancia ls and e - c o m m e r c e are the most c o m m o n types of appl icat ions be ing cont racted f rom A S P s 44 accord ing to an A S P Industry Consor t ium cus tomer t racking survey re leased on February 1 s t 2001 [Newcomb, F e b 2001] . Th is survey w a s of 137 senior and execut ive level manage rs and IT pro fess iona ls in the U . S . that have purchas ing authority for or involvement with genera l off ice productivity or software. Al l had indicated that they were currently us ing or had p lans to use the serv ices of an A S P within the next year . Here is a table outl ining the results (note that percen tages do not add up to one hundred s ince multiple answers were a l lowed). T a b l e 6. A p p l i c a t i o n s B y P e r c e n t a g e Intending T o B u y F r o m A n A S P A p p l i c a t i o n T y p e P e r c e n t a g e Commun ica t i ons (such a s email) 3 3 . 6 % F inanc ia ls and Accoun t ing 2 4 . 8 % E - c o m m e r c e (cata logs and transact ions) 2 1 . 2 % C R M 1 9 % Educat ion and Train ing 1 8 . 2 % H u m a n R e s o u r c e s 1 3 . 1 % Project M a n a g e m e n t 9 . 5 % S a l e s Fo rce Automat ion 8 .8% Pe rsona l Productivi ty (Off ice, word) 8 .8% Enterpr ise R e s o u r c e P lann ing 3 .6% Virtual trading communi t ies (E-Hubs) 3 .6% Supp ly C h a i n M a n a g e m e n t < 1% T h e s e numbers a re given here to provide the reader with an apprec iat ion of the appl icat ions that c l ients would likely s e e k f rom an A S P . In this sect ion we shal l cons ider what impact the part iculars of the appl icat ion will have on cus tomer adopt ion. W e shal l look at the serv ices 45 f rom the fol lowing aspec t s b e c a u s e they s e e m to be the crit ical factors shap ing client interaction with the software and client purchas ing dec is ions What Is The Economic Outlook For Asps? A S P s learned a lot in 2000 about how this market will operate . In response , many A S P s have c h a n g e d their bill ing structures. Both U S i and Cor io have started ask ing cus tomers to pay for part of the cost of software l i censes up front [Kerstetter, March 2001] . U S i has f in ished its three data cent res and is projected to be profitable by Q 4 2 0 0 1 . Cor io is projected to be profitable by Q 4 2002 . Of the top A S P s , only Cor io and Orac le have avo ided lay-offs [Bushaus , F e b . 2001] . However , most ana lys ts be l ieve 2001 will be the year A S P s round the bend and start to grow in terms of market penetrat ion, revenue and profitability [Bernard, F e b 2 0 0 1 ; Ko lbasuk M c G e e , J a n 2001] . W h y is this the c a s e ? Firstly, the p rob lems that dot corns c a u s e d A S P s are starting to p a s s and A S P s are no longer relying on dot corns a s potential cus tomers [Bushaus , F e b . 2000] . A s ment ioned, s o m e A S P s got roped into having to take on dot corns a s part of their f inancing ag reements - B r e a k a w a y is a pr ime examp le of this - and not only did they lose payments when the dot corns went out of bus iness , they a lso lost cus tomers that had to be turned away [Bushaus , F e b . 2000] . H e n c e , A S P s are turning to traditional b u s i n e s s e s , i.e. 'brick and mortar', a s new cus tomers and are activity jett isoning dot corns a s a drain on resources [Bushaus , F e b . 2001] . Second l y , A S P s are one of the few a reas in IT that s tand to benefit f rom the uncertainty that has hung over the North Amer i can economy at the end of 2000 and the start of 2 0 0 1 . T h e r e a s o n s ? "With execut ives nervous about commit t ing large investments and large investments and many capital spend ing budgets f rozen, enterpr ises will favour the dr ip- feed, 46 pay-as -you-go , A S P subscr ipt ion mode l . " [Wainewright, March 2001] . A s if to back this analyst up, Q w e s t Cyber . so lu t i ons landed a $22 million dol lar dea l with Expane ts . Th is tops the prev ious record-break ing dea l for the A S P industry of $18 mill ion (also by Qwes t Cyber .so lu t ions ) with R e d b a c k Networks. Other factors that are bolster ing the long term viability of the market is that 9 2 % of the cus tomers of A S P s have reported that they are sat isf ied with the serv ice they are being provided with accord ing to an A S P Industry Consor t ium survey [Newcomb, Feb 2001]. T h e main point be ing m a d e is that the future for A S P s is not a s g loomy a s the present . San fo rd B rown , v ice president and genera l manager of A T & T ' s host ing se rv i ces , f ramed it nicely when he s a y "There 's been no denying the trough of d is i l lus ionment, but despi te the tough t imes, the future really d o e s look quite promis ing. T h e bus iness i ssues that c o m p a n i e s are fac ing - they're not going to go away" [Koblentz, F e b . 2001] . Changing Customer Attitudes C u s t o m e r s are starting to c h a n g e the ways in which they interact with A S P s . A s ment ioned in Chap te r 2 , A S P s represent a fundamenta l c h a n g e in v iewing sof tware a s a serv ice and not a product. Th is has s o m e rather interesting c o n s e q u e n c e s . Firstly, sof tware is starting to fol low a utility mode l [Gruhn, J u n e 2000] . Th is commodi t izat ion of sof tware has been noted by other authors a s well [Sperl ing 2000 , Kerstetter 2001] . C u s t o m e r s are fol lowing a progress ion f rom product to serv ice and then serv ice to utility. Wi th a serv ice , method of del ivery is an important factor, however under a utility mode l , the only conce rn is del ivery [Gruhn, J u n e 2000] . Today , cus tomers perce ive the host ing mode l a s a new type of appl icat ions del ivery mode l . Jus t a s in the past they are still activity e n g a g e d in se lect ing the speci f ic appl icat ion they want to use . In a utility mode l the cus tomer ' s percept ions, expectat ions and investment style c h a n g e s . C u s t o m e r s do not take ownersh ip of the 47 software or the infrastructure the appl icat ions the appl icat ions runs on . Instead, they buy the functionality and serv ices it prov ides and have little conce rn about how that functionality is del ivered [Gruhn, J u n e 2000] . C o n s i d e r the fol lowing: USi , one of the pioneers of hosted applications, reports a marked change in the way customers are inquiring about their offerings. Last year, customer inquiries were always couched in terms of the branded application (e.g., "I want to talk to you about hosting Siebel software"). Today, only 25% of inquiries mention the software brand. More than half (52%) simply asks about USi's financial, sales-force automation or e-commerce services. USi sees this customer behavior change as an indication that customers are accepting the hosted paradigm and are seeking out the company based on its expertise rather than the brand of application; but this also suggests a longer-term change in customer behavior. Customers are disconnecting from the application brand and looking at software as a service that provides them with desired functionality. From this point, it's only a small to disregarding the application's underlying hardware, operating systems and middleware. And it's yet another small step for customers to begin looking at these solutions as a utility that delivers desired functionality. [Gruhn, June 2000] Data mining and O L A P fol lowed this s a m e route and are now cons ide red s tandard functionality requi rements. Microsoft sh ips S Q L Serve r 2 0 0 0 with O L A P and data mining features a s part of the D B M S . O n e author worr ies that A S P s may m o v e to be ing commodi ty s o quickly that they may never be able to recover their initial investments [Sper l ing, 2001] . Unl ike other utilities, A S P s would not have a stable commodi ty to del iver to the cus tomer . Th is conc ludes Chap te r 3. In chapter 4 an analyt ic f ramework to determine the factors in A S P s u c c e s s will be c reated. 48 Chapter 4. A S P Framework: Dec id ing A S P Type Th is chapter se ts forth an analyt ic f ramework for explain ing current t rends in the A S P market s p a c e and for mak ing predict ions a s to the s u c c e s s of var ious p layers in that market ( success of cou rse being def ined in terms of cont inued bus iness operat ion and growth of both revenue and profit). S i n c e we are attempting to predict o u t c o m e s that have not yet occur red , our fo recas ts are b a s e d on a s imple conceptua l ana lys is rather than on a sys temat ic empir ical study. A conc lus ive test ing of this model and the predict ions m a d e will therefore require further empir ical and analyt ical work. Never the less , s o m e examp les that would s e e m to support the predict ions made are given wherever poss ib le . T h e full ana lys is of the model is sp read over Chap te rs 4 and 5. Chap te r 4 examines the dec is ion a s to whether or not to buy the se rv i ces of a n A S P . Fur thermore, if the se rv i ces of an A S P are go ing to be pu rchased , f rom what type of A S P would they be pu rchased . However Chap te r 4 f o c u s e s more on the interplay of factors in produc ing the type dec is ion than it d o e s the dec is ion a s to whether or not to hire an A S P . In Chap te r 5 we focus more on the fo rces push ing a client toward, or away f rom, buying se rv i ces f rom an A S P more than we have in this chapter by a review of the literature and then we use that literature review to refine the f indings of this chapter . O n c e this is comple ted we shal l have mode led the entire dec is ion p rocess behind the cl ient 's cho ices . W e can then focus upon the strategies that A S P s shou ld adopt in light of this dec is ion p rocess a s we do in Chap te r 6. 4.1. Economic Concepts Used In ASP Framework Th is sect ion introduces the economic theor ies upon which the analyt ical f ramework p resen ted in the next sect ion is partially b a s e d . O n e of the major e c o n o m i c concep ts 49 forming the bas is for this f ramework is t ransact ion cost theory a s exp lored by Ol iver Wi l l i amson [Wi l l iamson, 1975; Wi l l i amson , 1979]. Market and hierarchy theor ies are presented. W e provide the reader with a brief introduction to these theor ies s o that the A S P f ramework mode l can be more easi ly unders tood. Oliver Williamson's Transaction Cost Economics Transac t ion cos ts are cons ide red by economis t Ol iver Wi l l i amson to be the "economic equivalent of friction in physica l sys tems" [Wi l l iamson, 1985]. Eng lander quotes him a s stat ing: A Transaction occurs when a good or service is transferred across a technologically separable interface. One stage of processing or assembly activity terminates, and another begins. A well-working interface, like a well-working machine, is one where these transfers occur smoothly. In mechanical systems, we look for frictions: Do the gears mesh? Are the parts lubricated? Is there needless slippage or other loss of energy? The economic counterpart of friction is transaction cost: Do the parties to the exchange operate harmoniously, or are there frequent misunderstandings and conflicts that lead to delays, breakdowns, and other malfunctions? [Englander, 1988] For our pu rposes t ransact ion cos ts are the cos ts assoc ia ted with 1) determining the availabil i ty of se rv ices 2) determining relevant pr ices of se rv i ces and 3) the cos ts of negotiat ing a contract. Envi ronmenta l factors such a s complex i ty of a good or serv ice and uncertainty about cos ts inc rease transact ion cos ts [Wi l l iamson, 1975] s ince more informational b a g g a g e must be inc luded with e a c h t ransact ion. T ransac t ion cos ts have been used in the ana lys is of hierarchical ve rsus market structures in intra and c ross firm interact ions [Wi l l iamson, 1975, Eng lander , 1988; Ma lone 1987]. In the fol lowing sect ion, factors determining the s u c c e s s of A S P s will be ana lysed in light of t ransact ional cos ts , which a re c lose ly related to Wi l l i amson 's concept of the s a m e name . It is our argument that 50 t ransact ion cos ts are a fundamenta l force in the buyer 's dec i s ions a s to whether or not to contract the serv ices of an A S P ( s ) and play a profound role in the cho i ce of A S P s . Arguably , A S P s lower t ransact ion cos ts for their cl ients in at least two ways . O n e , they have speci f ic knowledge of the serv ices they are offering. Two , b e c a u s e limited human resources are be ing sp read a c r o s s many users , the cost of those asse t s attributed to e a c h t ransact ion is lowered. In [Malone, 1988] the authors put forth an analyt ic f ramework in an attempt to predict whether new information techno log ies will c a u s e the coordinat ion of e c o n o m i c activit ies, which are adjacent s teps in the va lue cha in . T h e y state that e c o n o m i e s have two bas ic m e c h a n i s m s for coordinat ion of the f low of goods and se rv i ces through these adjacent s teps : markets or h ierarchies. Markets coordinate the f low of g o o d s and serv ices through supply and d e m a n d fo rces . W h e n the buyer of goods and se rv i ces c o m p a r e s its many poss ib le sou rces and m a k e s the dec is ion b a s e d on their own cri ter ion, a market structure is rea l ized. H ierarch ies , on the other hand , coordinate the f low of mater ia ls through adjacent s teps by control l ing and directing the f low of goods or se rv i ces through a set of predetermined suppl iers . Manager ia l dec is ions , not the interaction of market fo rces, determine the nature of hierarchical interact ions. In [Malone, 1988] a f ramework stat ing that internal organizat ional structures favour markets when product-descr ipt ion complexi ty and asse t specif ic i ty a re both low; hierarchical structures are favoured w h e n these factors are high (both factors are exp la ined in greater detail further on in this sect ion) . T h e organizat ional form likely for the other two poss ib le combinat ions of these factors depend ing upon which force is dominant . Th is is presented graphical ly in the fol lowing f igure 51 Figure 8. Attributes Affecting The Form Of An Organization s Hierarchy a Market o Low High Asset Specificity In the f ramework put forward in this thes is , the concept of asse t specif ic i ty is a lso used (and its definit ion is p resented in the fol lowing sect ion) however we are apply ing a simi lar logic a s [Malone, 1988] to the A S P market s p a c e . Howeve r 'complexi ty of product descr ip t ion ' is rep laced with the broader concep ts of t ransact ion cos ts and swi tching cos ts . W e shal l re ference this f ramework and use logic similar to this in order to expla in s o m e occu r rences in the A S P doma in today. W e refine the concep ts of H ierarchy and Market in the next sec t ion . Rather than apply ing them to the structure of an organizat ion, we shal l be apply ing similar concep ts to the structure of the A S P market. 4.2. ASP Framework In this sect ion we apply and adapt the prior theory to determine the major fo rces in determin ing the s u c c e s s of a n A S P . W e shal l then attempt to bui ld a f ramework a round these fo rces and b a s e d on this we shal l try to determine when an A S P will be u s e d , and if y e s , wh ich A S P mode l is most appropr iate. 52 Our mode l is b a s e d on four input var iab les: t ransact ional cos ts , specif ic i ty, criticality and substitutabil ity. T h e s e var iab les a re used to justify our poss ib le ou t comes , wh ich are def ined in the fol lowing sect ion. Outcome Variable Our equivalent of a 'Market ' structure a s descr ibed in F igure 8, in the A S P doma in would be a s p a c e where client relat ionships with A S P s would be simi lar to a "spot market." Th is is where c l ients cou ld readi ly switch between A S P s a n d readi ly comb ine a n d integrate their se rv ices . It is our theory that t ransact ion cos ts and specif ic i ty would be low for this to be the c a s e . Under these condi t ions it is our argument that c l ients will buy se rv i ces f rom var ious sou rces , a s suits their needs , and link them together if the cost of do ing this is low. T h e cho ice of A S P type (or not to use an A S P ) on the part of the cl ient is the dependent var iable in our f ramework, and is what the client will dec ide by measur ing their needs against our input var iab les def ined in the next sec t ion . W e represent our stratif ication graphical ly in the fol lowing f igure: 5 3 F i g u r e 9. A S P T y p e Matrix o s co a. 01 o N O as o •C a) > Q2 Q3 Q1 Q4 Niche Enterprise Span of Services In this f igure e a c h quadrant represents an A S P type a s created in Tab le 1. Not represented in F igure 9 is the dec is ion not to use an A S P . W e shal l use the fol lowing c o d e s to act a s shor thand for e a c h q u a d r a n t / A S P type: T a b l e 7. C o d e s F o r A S P T y p e U s e d In A S P F r a m e w o r k A S P T y p e o r D e c i d e d not to u s e a n A S P Q u a d r a n t C o d e F o c u s e d Solut ion Q1 N V N iche V S P Q 2 N H E S S Q 3 E H Ful l V S P Q 4 E V Wi l l not use an A S P N/A X X 54 Let us examine the impact the cl ients cho ice a s to whether or not to use an A S P or use an A S P of type X further before we cont inue with the construct ion of our mode l . In this mode l , if most cl ients favour F o c u s e d Solut ions in the long run, then F o c u s e d Solut ions would be cons ide red the " success fu l " mode l . A S P s in N H , E H and E V would not be success fu l in this situation, b e c a u s e the cl ient 's cho ice wou ld effectively cut them out of the va lue cha in . If c l ients favour N iche V S P s , then E S S a n d Ful l V S P s a re still unsuccess fu l mode ls . However F o c u s e d Solut ions cou ld still exist by sel l ing their se rv i ces to the N H A S P s . Simi lar ly for long run cho ices of E H and E V , b e c a u s e E V is the full serv ice V S P mode l , it is unc lear whether or not they could dominate all vert ical industr ies. However , if most c l ients favour option X X , then N V , N H , E H and E V are all cons ide red unsuccess fu l . A s the cost of f inding, connect ing , coordinat ing and serv ice cus tomiza t ions of a new A S P d e c r e a s e s (i.e. lower t ransact ional cos ts and specif icity), c l ients will go f rom a mode where they do not use an A S P to interacting with the min imum number of A S P s (i.e. use a full serv ice V S P or E S S in quadrants 3 or 4) toward interacting with a number of N iche V S P s (NH). At the ext reme they would comb ine the serv ices of a number of F o c u s e d Solut ions ( N V or N V a n d NH) . S o m e A S P s (particularly U S i wh ich h a s adop ted the . N E T f ramework [Fonseca , Nov. 2000]) would s e e m to be hedging the r isks of uncertainty by offering both E S S and F o c u s e d Solu t ions, hence spann ing N V , N H and E H . N V , a n d to a somewha t l esse r extent N H , co r respond to the "market" in the ana lys is by [Malone, 1988] and would be character ized by free interaction be tween the client and numerous A S P s , with N V being a pure market where cl ients directly comb ine the serv ices of many F o c u s e d So lu t ions. Think ing of an appl icat ion a s a col lect ion of funct ional i t ies, with f(i) represent ing a unique p iece of functionality and A(i) represent ing an appl icat ion (in this representat ion T is an indexing value) , this would be represented graphical ly a s fo l lows: 55 Figure 10. Functionalities Combined Into An Application By Client Connec t i ons to the functional i t ies provided by var ious F o c u s e d So lu t ions and are brought together by the client to form appl icat ion A(1). Th is impl ies that the client is respons ib le for the integration of var ious software functionali t ies (for var ious sou rces ) . Th is is not an easy task, and shal l the impl icat ions of this shal l d i s c u s s e d further in this chapter and Chap te r 5. Note that the client is not l imited to just creat ing one appl icat ion and fur thermore any f(i) cou ld exist within the client if an A S P is not providing that functionality. In fact all f(i) would exist within the client if no A S P s are being u s e d . Th is ar rangement would be the result of very low t ransact ion cos ts and low specif ic i ty of se rv ices if the most f(i) are being provided by N V A S P s . T h e next s tep toward a hierarchy a s we have def ined it would be the use of N iche V S P s , which would appea r in the fol lowing f igure. A l s o note the client has created what we def ine a s the p a c k a g e deno ted P(i). Aga in the reader is reminded that T is an indexing va lue due to the fact that more than one p a c k a g e maybe required. A p a c k a g e is a col lect ion of an underdetermined number of appl icat ions: 56 Figure 1 1 . Applications Delivered To Client f(1) f(2) f(3) f(i) Here the Cl ient has contracted the serv ices of a number of N iche V S P s (similar to F igure 10, any A(i) cou ld be an in-house appl icat ion). T h e connec t ions be tween the layers in this f igure are not a s c lear a s in the previous f igure. T h e connec t ions be tween the functionality layer and the appl icat ion layer are not necessar i ly a c r o s s a S O A P connect ion to a F o c u s e d Solut ion and f(i) cou ld be provided by an ISV. T h e y cou ld exist so le ly within an appl icat ion hosted by the N iche V S P (in the c a s e of an ISV providing f(i)). In this m o d e , the client is contract ing by the appl icat ion and may have to interact with more A S P s in order to acqu i re more appl icat ions. Travel l ing further up this h ierarchy we arr ive at the level of us ing an E S S or Ful l V S P . Th is cho ice would be equivalent to the h ierarch ies to wh ich [Malone, 1988] refers b e c a u s e the client would have a long-term relat ionship with the A S P providing it with the des i red p a c k a g e of appl icat ions. Rega rd l ess of market fo rces s u c h a s pr ic ing, o n c e an E S S or Ful l V S P is be ing used the client is locked into that relat ionship. 57 Figure 12. Package Delivered To Client Note that the client may need more than one package (this is a shortcoming of Oracle Business Online as was discussed in Chapter 2) but ideally the client would need only one package. Any A(i) could simply be part of a collection of applications within the ESS or Full Service VSP (if A(i) is provided by an ISV) or they could be within Niche VSPs as would be the case if the top level ASP is an aggregator. Furthermore, in a situation where the client has made decision XX (remember this is the notation for not using an ASP) package P(i) would be provided by an ISV such as SAP or Peoplesoft. The advantage is that the client is 58 not conce rned about what is occurr ing at the lower levels, just that the se rv i ces they want in P(1) a re de l ivered, they have little concern how the A S P brought them together. However it is our theory that the complexi ty of product descr ipt ion and the asse t specif ic i ty of most appl icat ions are s o high that hierarchical industry structures (i.e. the client will prefer to dea l with E H or E V A S P s ) , with the p layers l ocked into long-term relat ionships, will exist between most cl ients and A S P s for the fo reseeab le future. Fur thermore, in order to min imise t ransact ional cos ts many cl ients will opt for an E S S aggregator or Ful l V S P that can offer appl icat ions to support the highest number of bus iness funct ions poss ib le b e c a u s e this reduces their coordinat ion cos ts to a min imum. B e c a u s e of the difficultly assoc ia ted with software integrat ion, the wou ld be one of the pr imary fo rces driving the cl ient toward us ing a n E S S or Ful l V S P . Th is is accoun ted for through the t ransact ional cost var iable def ined in the next sec t ion . It is c lear f rom these f igures that interacting with A S P s at the functionality or appl icat ion level that the client has more relat ionships to maintain with A S P s , or that if only a few componen ts of their IT infrastructure is provided by one or more A S P s , these serv ices must be integrated with their exist ing in -house appl icat ions. Th is m e a n s that the client is support ing more relat ionships and integrating more sof tware. A s the progress ion is m a d e toward E H or E V , the client is directly support ing fewer and fewer of these relat ionships and in terconnect ions. However , if the dec is ion is m a d e to conduct all operat ions in -house, even though no relat ionships to A S P s must be mainta ined, the cost of this are higher then us ing an A S P (or s o the proponents of this model argue) . W e caut ion the reader that, a l though there are very strong paral le ls, we are not us ing these concep ts of 'Market ' and 'Hierarchy' in the s a m e way a s [Malone, 1988]. If the client dec ides to use a col lect ion of F o c u s e d Solut ions, this would most c lose ly resemble the 59 'Market ' structures def ined in [Malone, 1988] a s for the pu rposes of this thes is we re-use the term market but reapply it to the situation depic ted in F igure 10. If the client dec ides to use a E S S or V S P then this most c lose ly resemb les the 'Hierarchy ' st ructures a s def ined by [Malone, 1988], however we are reapplying this term to the situation dep ic ted in F igure 12. F igure 11 represents a c ross , or comprom ise between the two structures. A l s o , note that when we have a situation a s set forward in F igure 12, the entit ies at the bottom (the N iche V S P s and F o c u s e d Solut ions) may behave in a spot market l ike fash ion , however from the cl ient 's perspect ive this is a hierarchy situation. W e shal l now examine why the client would make the dec is ion to use one mode l or another or not to use the A S P model at al l . Input Variables In this sect ion we def ine and present what we cons ider to be the main fac tors in the cl ient 's dec is ion to use one A S P model or another (implicitly the assumpt ion has been made that the client has dec ided to use an A S P , Chapte r 5 presents the reasons why a client would use an A S P ) . It is our theory that the dec is ion rests on these four var iab les : t ransact ional cos ts , specif ici ty, criticality and substitutability. W e borrow heavi ly f rom the ideas of Ol iver Wi l l i amson and [Malone, 1988], however we slightly redef ine their concep ts for the purposes of this thes is . Transactional Costs: For the purpose of this thes is , these are the cos ts (whether opportunity or real ized) of f inding, agree ing on an S L A , connect ing to, the cos ts of interacting with an A S P and the switching cos ts of chang ing f rom an in -house solut ion to a hos ted serv ice or swi tching be tween A S P s . No te that this wou ld inc lude the cos ts of customizat ion. T h e main argument behind the A S P model is that t hese cos ts a re lower over 60 the lifetime of the appl icat ion then the client buying, instal l ing, integrating and maintaining the appl icat ion for themse lves (this is d i s c u s s e d further in Chap te r 5). T h e cost of interacting with an A S P c a n include the subscr ipt ion fee (or whatever other pricing agreement is ar ranged) , a s well a s integrating the serv ices of that A S P with other A S P s or with any in-house appl icat ions that will remain in use (i.e. a coordinat ion cost) . Al l th ings being equa l , the number of A S P s and/or in -house solut ions the client is dea l ing with, the higher the t ransact ional cos ts . T h e reason for us ing numerous A S P s cou ld only be justif ied (from a purely t ransact ional cost point of view) if us ing those A S P s a s o p p o s e d to us ing one A S P lowers the cos ts of f inding, connect ing and interaction s o the ga ins in customizabi l i ty would offset the inc reased coordinat ion cos ts . Th is is not the tradit ional definit ion of t ransact ion cos ts a s def ined by [Wi l l iamson, 1979] but a variant of it we have created for the pu rposes of this thes is . Fur thermore we caut ion the reader that, a l though there are very strong paral le ls, we are not us ing this concept in the s a m e fash ion a s [Wi l l iamson, 1979; Ma lone , 1988]. Asset Specificity: In [Malone, 1988] an input used by a firm (or individual consumer ) is cons ide red to be highly asse t speci f ic (according to Ol iver Wi l l i amson 's definition) if it cannot readily be u s e d by other f i rms b e c a u s e of site specif ici ty, phys ica l asse t specif ic i ty, and/or human asse t specif ici ty. S o m e examp les of these types of specif ic i ty wou ld be : a natural resource s u c h a s an oil well is site speci f ic b e c a u s e it cannot be moved without great cost , a spec ia l i zed mach ine tool or a comp lex computer program des igned for a s ing le purpose would be cons ide red physical ly speci f ic . H u m a n asse t specif ic i ty wou ld include learning by do ing , or ski l ls p o s s e s s e d by a limited percentage of the populat ion. W e reuse this definit ion of specif ic i ty in the model c reated in this sec t ion . 61 Criticality: Criticality is defined as the extent to which a good or service is necessary for the continued operation of the firm without extraordinary cost or inconvenience. For example an ERP service provided by an ASP would be considered highly critical since disruption of that service would disrupt almost all business functions. However a service such as email can be readily replaced so it is not considered to be highly critical. Note that if email service could not be reinstated in a timely fashion it would result in a disruption of business functions (this is due to the time component implicit in our definition of criticality). Criticality and specificity are logically independent, yet highly correlated in a very interesting fashion. The reason for this is as follows: if a critical IT asset is highly specific then it cannot be readily replaced and that lag in replacement is what disrupts business function. If an IT asset is low specificity then it can be readily replaced; and regardless of its criticality that quick replacement means it will not have the opportunity to disrupt the function of that business unless the continuous operation of that asset is crucial. It is difficult to imagine a situation in which an IT good or service would be have low-criticality, yet be highly specific, so we feel justified simply considering this to be a highly specific asset. Substitutability: Substitutability is generally inversely related to specificity, however the two are logically independent. Substitutability is defined as the level of ease with which one set of functionalities can be replaced with another set that perform the same task. Email is an example of a highly substitutable service. The providers of this service are numerous and the specificity of this service is low so providers can be readily changed (unless there are contractual obligations preventing this). Substitutability is sometimes referred to by the inverse concept of 'lock-in'. We have chosen these four variables because they seem to include all of the questions the client would want answered with respect to the use of an ASP for providing a particular 6 2 appl icat ion. T h e t ransact ional cost var iable accoun ts for pr ice and cost conce rns . T h e amount of specif ic i ty relates to the cl ient 's conce rns about cus tomizat ion . A n y IT dec is ion 's poss ib le impact upon bus iness funct ions must be accoun ted for therefore we include criticality. A l s o , a s with all IT pu rchases , the client would have conce rns about how readily that IT asse t cou ld be rep laced and this is accoun ted for by the substitutabil i ty var iable. W e do not cons ider reliability at this s tage b e c a u s e it would be more of a conce rn in the cho ice to use an A S P or not, therefore we leave it for Chap te r 5. Note that per fo rmance and secur i ty wou ld a lso play a crucial role in the cl ient 's cho ice to pu rchase an IT asse t . However , a s with reliability it p lays less of a role in the cho ice of A S P type, s o we a lso leave the d i scuss ion of per fo rmance and securi ty to Chap te r 5, Sec t ion 5.1. W e will usual ly limit the granularity of our var iab les to a s imple d iscreet s p a c e of {Low, High} for the pu rposes of simplicity when mak ing predict ions. Howeve r when ana lys ing them in the next sect ion we shal l vary them over a cont inuous s p a c e of [Low, High]. Hav ing def ined our independent var iab les, let us now cons ider their interact ions to o n e another before we cons ider their impact upon the ou tcome var iab les. Interrelationships Of Input Variables A s d iscovered in a review of the current literature, many of the proponents of the A S P model a rgue A S P s lower t ransact ional cost in severa l ways . M a n y of their reasons for this asser t ion shal l be d i s c u s s e d in Chapte r 5. W e asser t that c l ients will opt to use only one A S P . If the cus tomer currently only has need to ou tsource part of its current IT infrastructure, whether or not it p lans on outsourc ing more in the future p lays a part in their cho ice of A S P type. For example , a c o m p a n y that p lans having its emai l serv ice hosted today and having its C R M hosted in the future would have two cho i ces : 63 Opt ion O n e : contract the serv ices of a provider with emai l se rv ices only, and then in the future, contract the serv ices of an A S P providing C R M and integrate the two or Opt ion Two : contract the serv ices of an A S P that prov ides both se rv i ces , pu rchase the emai l host ing se rv i ces today and then pu rchase the C R M serv ices o n c e they are n e e d e d . Wi th the first opt ion, a new A S P would have to be contacted and integrated with the first. Wi th the s e c o n d opt ion, one A S P would suff ice and would , a s the ana lys is of t ransact ional cost var iable shal l show, be a better si tuation. A n y cl ients who c h o s e the first option would likely switch to the s ingle A S P opt ion, and knowing this they would have incent ive to pick an A S P that prov ides a wide breath of serv ices , whether or not they currently need them, to avoid this si tuat ion. It is to capture this type of dynamic that we have separa ted switching cos ts out f rom transact ional cos ts . T h e appl icat ions themse lves are very high in site specif ici ty, a s mov ing them from platform to platform on the server s ide can be a difficult task. That c rea tes i m m e n s e pressure to leave them on the site where they were first c rea ted. Howeve r this contrasts against the situation for the hosted serv ices that are b a s e d upon these appl icat ions are very low in site specif ic i ty b e c a u s e they can be a c c e s s e d from a lmost anywhere in the wor ld . W e are not a s conce rned with the site specif ici ty of the appl icat ion a s we are the serv ice that is prov ided, part of the reason for raising this point is a s fol lows. App l ica t ions that operate between organizat ions (such a s procurement or C R M ) lend themse lves to the A S P mode l b e c a u s e of the universal accessib i l i ty of a web-hos ted serv ice . Th is is where the low site specif ic i ty of the serv ice is of advantage. Fur thermore, t ransact ion cos ts are reduced b e c a u s e of the e a s y a c c e s s to the Internet both part ies would have for connect ing to the A S P . T h e A S P can then act a s a buffer, protecting the client f rom the coordinat ion cos ts of connect ing to a new partner. 64 Hos ted se rv i ces c a n be high or low in physica l specif ici ty, for examp le emai l se rv i ces require relativity little customizat ion and hence are not very spec i f ic whe reas an E R P sys tem requiring a great dea l of customizat ion and client knowledge would be very spec i f ic . H u m a n asse t s that support the hosted serv ice are very speci f ic and s c a r c e . A s noted by [Malone, 1988] the specif ici ty and complexi ty of g o o d s or se rv ices are often corre lated; however in most genera l d i scuss ions the fact that they are logical ly independent m e a n s that their have to be treated a s separa te factors. S i te-spec i f ic a s s e t s c a n go against this t rend. Cons ide r , for examp le , an ar tes ian wel l , the wel l is very low in complexi ty , yet it is highly site speci f ic . Another examp le would be that of a comp lex good or serv ice that is low in specif ic i ty s u c h a s an automobi le . Most information techno logy g o o d s and serv ices are very comp lex when compa red to other good and serv ices ; however , relative to one another, they fol low a c lear t rend. L o w specif ic i ty g o o d s a re usual ly those that a re es tab l i shed a n d well unders tood by users and providers, require little customizat ion and h e n c e they are low in specif ic i ty when compa red against other IT goods and se rv i ces . H igh specif ic i ty goods and serv ices are those that require a great deal of customizat ion and are general ly the more comp lex i tems in the IT realm. Speci f ic i ty is a doub le -edged sword . T h e buyer b e c o m e s vu lnerable b e c a u s e a s the asse t b e c o m e s more and more speci f ic v ia customizat ion, it b e c o m e s harder and harder to f ind a rep lacement suppl ier . Converse ly , by customiz ing their good or serv ice , the sel ler has l imited the number of buyers who would be interested in that good or serv ice and is reliant on the cont inued pat ronage of the buyer to recoup that investment in cus tomizat ion . High specif ic i ty of a s s e t s tends to inc rease the lock-in of the cus tomer [Wi l l iamson, 1979; Ma lone , 1988; Eng lander 1988]. 65 Note that we are not talking about ' m a s s customizat ion ' but referring to further specif ic i ty required by each new client. Cus tomiza t ion eats into an A S P ' s margin on a contract [King, Sep t . 2000] . A S P s general ly try to build their appl icat ions ba lanc ing functionality and required customizat ion for each customer . Th is is where spec ia l izat ion is key for non-aggregators . C o n s i d e r the fol lowing quote: ' "What ' s different about us is that other A S P s have 10 to 12 appl icat ions [for rent],' s a y s Co reHarbo r C E O J a y Chaudh ry . 'If you do that, you b e c o m e a general is t . W e ' r e the heart su rgeons when it c o m e s to e -p rocurement . . . . W e do Ar iba day and night, s o we really know how to support it.'" [Aun, 2001] . Th is m e a n s that for the max imum eff ic iency non-aggregators shou ld min imise the cos ts of customizat ion by focus ing on the appl icat ion they are renting. A S P s are sel l ing a serv ice . That serv ice inc ludes the customizat ion of the software. A S P s make money w h e n they do a s little customizat ion a s needed and do what customizat ion is needed a s efficiently a s poss ib le . In the aggregator mode l , money is made purely on integration and customizat ion cost reduct ion abil i t ies. However , for the current p layers in the market that a re not adapt ing to the chang ing environment. N e w entrants have looked at what is work ing and built their own new appl icat ions with the web in mind (unlike many of the current of fer ings). C o n s i d e r the fol lowing: No matter how well they tune their engines, though, the first generation ASPs will have a tough time outperforming the newer entrants. The newbies establish one super-reliable Web site that all their customers hook into—plugging their information into simple templates. If the Young Turks get it right, the economies of scale could produce gross profit margins topping 90%-dramatically better than the 70% average among traditional software makers and the 15% to 20% margins that early ASPs have achieved, say analysts. [Kerstetter, 2001] 66 This is the m a s s customizat ion we referred to previously. Th is is important b e c a u s e these 'newbies ' a re lowering the switching cos ts between A S P s and h e n c e reduc ing specif ici ty. Th is force is a push toward a market style structure in the A S P industry. B e c a u s e cl ients are not going to be conce rned with brand n a m e if sof tware b e c o m e s a utility a s projected by many analysts , cus tomer loyalty gets p laced on the A S P and the functionality, loyalty to the underlying ISV e rodes . Exacerba t ing the eros ion of ISV impor tance, many A S P s are starting to create their own solut ions, h e n c e increas ing margin by cutting out the ISV. Th is would be inline with our t ransact ion cos t theory s ince removing the ISV would further reduce the cost per t ransact ion. T h e lower the amount of customizat ion required, the better for the A S P if cus tomers are usual ly short term. For an A S P , lower customizat ion m e a n s higher marg in in the short run, which can be t ranslated into further reducing t ransact ion cos ts if the relat ionship is a short-term relat ionship. However f rom a specif ici ty point of v iew long term relat ionships would be created with the lock- in of a high-specif ici ty serv ice ; and high up-front customizat ion cos ts cou ld be cons ide red an investment to gain a long s t ream of monthly payments genera ted by a long-term relat ionships. Se rv i ces that are not easi ly cus tomizab le (i.e. a re highly speci f ic before the client pu rchases them) face a ser ious prob lem b e c a u s e of the high cost assoc ia ted with cus tomiz ing them: bus i nesses don't want to c h a n g e their p r o c e s s e s to suit the appl icat ion. S A P exper ienced this prob lem o n c e cus tomers started to wonder about loss of compet i t ive advan tage . Many potential c l ients will not buy if the software is perce ived to require bus iness p rocess c h a n g e s . Th is m a k e s further customizat ion cost ly s ince the appl icat ion is a l ready highly physical ly speci f ic . However , the cus tomer often adap ts to the sof tware, and is effectively creat ing specif ic i ty by mak ing himself or herself spec i f ic to the appl icat ion, hence creat ing lock- in. 67 T h e net result of these interrelations is that many of these var iab les a re not or thogonal (i.e. if the ax is they would be represented by would not meet at a 90 degree ang le if d rawn in two d imens ions) . Th is m e a n s that a change in one var iable often may result in a c h a n g e of another (or others). B e c a u s e it is outs ide the s c o p e of this thes is we shal l not de lve into this in a much greater extent. W e summar i ze the interact ions of the four var iab les in the fol lowing table: T a b l e 8. S u m m a r y O f V a r i a b l e Interre lat ionships F a c t o r 1 F a c t o r 2 Impact of i n c r e a s i n g F a c t o r 1 u p o n F a c t o r 2 Transact iona l C o s t Speci f ic i ty N o Effect Transact iona l C o s t Crit icality No Effect Transact iona l C o s t Substitutabil i ty N o Effect Speci f ic i ty Transact iona l C o s t Increases Speci f ic i ty Crit icality Increases Speci f ic i ty Substitutabil i ty D e c r e a s e s Crit icality Transact iona l C o s t Increases Crit icality Speci f ic i ty N o Effect Crit icality Substitutabil i ty D e c r e a s e s Substitutabil i ty Transact iona l C o s t N o Effect Substitutabil i ty Speci f ic i ty D e c r e a s e s Substitutabil i ty Crit icality D e c r e a s e s W e now cons ider the impact each of these var iab les would have upon the output dec is ion var iable. 4.3. Framework Predictions In this sect ion we examine the impact each input var iable wou ld have upon the output dec is ion var iable. E a c h var iable is cons idered a lone, with the or thogonal var iab les being 68 held equa l and the result ing c h a n g e s predicted by our mode l is put forward. Note that in e a c h of the d iag rams in this sect ion the client is not necessar i l y mov ing f rom quadrant to quadrant , their cho ice is at the outset of dec id ing whether to use an A S P and if s o which A S P s to contract. Fur thermore we a s s u m e that the client is not currently us ing an A S P . T r a n s a c t i o n a l C o s t s A s ment ioned, t ransact ional cos ts (denoted T C in this sect ion) and swi tching cos ts are interrelated, however it is our opinion that they are related weak ly enough that the impact of varying T C over the s p a c e of [Low, High] can be ignored in the fol lowing ana lys is . It is our argument that a s T C is inc reased , speci f ical ly a s the cos ts of f inding, agree ing on an S L A , connect ing to, and interacting with an A S P (or A S P s ) i nc reases cl ients will move toward an A S P that suppor ts the widest number of bus iness funct ions. Graph ica l l y this c a n be represented a s the fol lowing: F i g u r e 13. C l i e n t C h o i c e O f A S P T y p e A s T r a n s a c t i o n a l C o s t s Increase CO sz CO a. (A o N O X 0 5 O > Niche Enterprise Span of Services In this diagram TC represents transactional costs 69 In this d iag ram the arrow represents the cho ice of the client, and how it c h a n g e s a s t ransact ional cos ts are inc reased . In this ana lys is , t ransact ional cos ts are indifferent to whether E V or E H are u s e d . A n argument cou ld be made that the client wou ld favour E V before E H ( remember E H is an E S S whereas E V is a Ful l V S P ) . However , it is our belief that this would heavi ly dependant upon what vert ical industry the client is in s ince any V S P only opera tes within one vert ical industry, and hence we do not include this at this juncture. A l s o , in this d iagram and in F igures 14 to 18, once the arrow ends the cl ient 's dec is ion b e c o m e s not to use an A S P . In [Malone, 1988] a c ross -compar i son of complexi ty of product descr ipt ion against asse t specif ic i ty is u s e d to determine the structure of intra and inter f irm interact ions. E v e n though the thes is of their work w a s that a s information technology lowered t ransact ion cos ts , f i rms will use markets structures more and more. UDDI and S O A P are techno log ies that lower the cost of f inding and connect ing to new A S P partners in N V and N H . H e n c e techno log ies such a s these would lower the T C of connect ing to var ious A S P s and push cl ients toward these quadrants . W e feel that the reason for this is the control the cl ient ga ins in the lower levels of our functionality hierarchy would then offset the cos t s of maintaining more relat ionships with A S P s . UDDI and S O A P cou ld topple aggregators in E H and E V s ince these techno log ies enab le e a s y l inkage between cl ients and N iche V S P s (NH) or F o c u s e d Solu t ions (NV) . However , it would s e e m that cl ients buying into the A S P model would benefit f rom totally outsourc ing rather then mixing A S P serv ices with in -house cl ient/server or main f rame sys tems . Full serv ice A S P s del iver the most benefit when they are in control of all the appl icat ions the client is us ing , b e c a u s e this min imises the number of f i rms and se rv i ces that must be 70 coord inated by the client. O n c e the T C of us ing an A S P b e c o m e s higher then us ing in-house solut ions the client will make dec is ion X X . Asset Specificity Like switching cos ts , we cannot cons ider asse t specif ic i ty to be or thogonal to the other four var iab les. Increasing specif ic i ty invariably m e a n s increas ing cus tomizat ion , h e n c e impact ing t ransact ional cos ts and by definition increas ing specif ic i ty i nc reases switching cos ts . Increases in specif ic i ty imply a dec rease in substitutabil ity. Crit ical i ty c a n be held constant in this ana lys is . However , letting specif ici ty dominate where p a c k a g e level se rv ices are required p roduces the fol lowing: Figure 14. Client Choice Of ASP Type For P(i) As Asset Specificity Dominates s 8. o CO O > Q2 High Specificity Q1 Q3 Low Specificity Q4 Niche Enterprise Span of Services Th is is ignoring the other factors, however we shal l v iew what we feel is the more realistic ou tcome when the other factors are taken into account . O u r rat ionale beh ind the fol lowing is this: If speci f ic i ty requirements are low, an enterpr ise wide serv ice m a k e s the most s e n s e a s 71 it k e e p s T C at a min imum and presumably an E S S would cou ld provide the most gener ic solut ion that would satisfy those needs . A s specif ic i ty b e c o m e s a greater conce rn a Full V S P would be somewha t "p re-cus tomized" for the industry that the client is in, hence sat isfying the inc reased need for specif ici ty. O n c e specif ic i ty requi rements inc rease further, the ability to 'mix and match ' appl icat ions (and at the ext reme functionali t ies) b e c o m e s the most important factor. If the client only requires an appl icat ion A(i) this p r o c e s s starts in N H and ends in N V , if the client only requires a functionality f(i) the p rocess would start and end in N V . However , this ana lys is ignores the impact that specif ic i ty has on T C , switching cos ts and substitutabil ity. W e feel that a more realistic ou tcome would be the fo l lowing: Figure 15. Client Choice Of ASP Type For P(i) As Asset Specificity Increases s (9 s. o C D O •e C U > Q2 Q Low Sf 3 jecificity Q1 \ High Sp Q / ecificity 4 Niche Enterprise Span of Services B e c a u s e the client is search ing for a p a c k a g e P(i) it would m a k e little s e n s e b a s e d on our prev ious T C and switching cost ana lys is to cons ider N V or N H se rv i ces . S o to simpli fy the ana lys is we a s s u m e the client is seek ing a p a c k a g e P(i) and then extend this ana lys is to A(i) 72 and f(i). A s the required specif ici ty inc reases the client would dec ide to use a V S P if one is within their vert ical market, s ince these serv ices wou ld p resumab ly require less customizat ion at the outset. However the c a s e where the client is sea rch ing for an A S P to provide A(i) or f(i) is rather interesting when the other factors are cons ide red : Figure 16. Client Choice Of ASP Type For f(i) Or A(i) As Specificity Increases Q2 Q3 I 1 I Low Specificity Q1 i High Specificity Q4 Niche Enterprise Span of Serv ices Desicion path starts in Q(1) iff(i) is desired, Q(2) if A(i) is desired Note that this is c l ose to being the comple te opposi te of the predict ion m a d e in F igure 15. T h e reason for this is a s fo l lows: a s s u m i n g f(i) is required a n d speci f ic i ty i n c reases s o would switching cos ts , to avoid the situation of coordinat ing with numerous F o c u s e d Solu t ions, or A S P s in genera l (i.e. in an attempt to min imise T C ) , the client wou ld dec ide upon an A S P in N H (from this point onward the argument for A(i) or f(i) is identical s o we shal l only refer to A(i)). A s specif ic i ty of the appl icat ion inc reases further the switching cos t and t ransact ional 73 cost a rguments aga in push the client toward an A S P support ing a wider number of bus iness p r o c e s s e s and f rom this point onward or argument paral le ls that for p a c k a g e P(i). In F igures 15 and 16 the cl ient 's dec is ion will b e c o m e X X a s specif ic i ty b e c o m e s extremely high. Th is argument is strongly paral le led by research in the f ield of genera l outsourc ing [Butler, 2000] . Howeve r the cl ient 's dec is ion in the c a s e is st rongly re lated to criticality a n d substitutabil ity s o before we ana lyze criticality and substitutabil ity in the fash ion we did the previous three var iab les we must first s tep back, and cons ider the cho i ces the client is mak ing before they have dec ided to enter into a contract with an A S P or build the des i red p a c k a g e , appl icat ion or functionality themse lves . Th is is exp la ined in the fol lowing, rather lengthy, as ide . T h e perce ived level of criticality (to the client) would s e e m to be inversely related to the wi l l ingness of cl ients to buy into an A S P ' s serv ices (i.e. high criticality impl ies the client will favour dec is ion X X ) . Cl ients are aware that if the provider of a high criticality, highly speci f ic serv ice were to go out of bus iness would halt their operat ions for an undetermined amount of t ime. C l ients are a lso aware that this is a new market and Gar tner is giving any particular A S P a 2 in 5 c h a n c e of surviving this year [Bushaus , Sep t 2000 , Pr ing F e b . 2001] . Th is has lead to an interesting set of pa radoxes within the industry. It would s e e m cl ients would f lock to the first A S P that w a s profitable, and in response many A S P s did not build data cent res (consider B reakaway and Cor io in Chap te r 2). However the A S P with the largest cl ient b a s e is the only A S P that built data cent res : U S i . U S i even g o e s s o far a s to use its ownersh ip of its data cent res a s a market ing tool . Th is may be b e c a u s e cl ients v iew this a s a more " rea l " c o m p a n y b e c a u s e it has real , tangible a s s e t s whe reas the others may s e e m too much like the dot corns, where once they go out of bus i ness nothing remains . 74 Cl ients are will ing to let A S P s handle ' headache ' appl icat ions s u c h a s emai l and m e s s a g i n g . Th is due of the fact it is not hard to switch A S P s if need be with these appl icat ions s ince the specif ic i ty is low and they are highly subst i tutable. B e c a u s e crit ical appl icat ions are usual ly not easi ly subst i tuted, cl ients have two reasons not to buy in. Firstly, substitutabil i ty is low and specif ic i ty is often high. T h e client cannot easi ly buy these se rv i ces somewhe re e lse once they have moved to an A S P solut ion that has been highly cus tom ized . C o m p o u n d i n g this lock- in is that A S P solut ions often m a k e in-house IT staff redundant, hence the IT department is often empt ied and a new staff cou ld be hard to get aga in . S e c o n d l y the d e p e n d e n c e on the A S P is extremely high a s wel l . E v e n if the A S P gave a copy of the software to e a c h client when they went out of bus iness , it is unc lear if many of the c o m p a n i e s would be ab le to maintain or run it on their own and most wou ld sure ly lose the t ransact ion cos t benef i ts that drove them to A S P s to begin with. A high criticality and a high substitutabil ity situation would be dominated by the fact that new providers are e a s y to c o m e by. But this logic only holds to a limit. If the sys tem is crit ical in the ext reme and virtually zero downt ime is a l lowed, con f idence in the provider may dominate. If a serv ice is of low criticality and not easi ly subst i tuted, then the cl ient has little concern if the serv ice is of fered or not s ince it is not crit ical, however it is unc lear if this situation would ever exist. If a serv ice is low criticality and easi ly subst i tuted then the c o m p a n y has no downs ide for go ing with an A S P solut ion, and lock-in is low. S o , f rom the cl ients point of v iew, the fol lowing conc lus ions are m a d e : 75 T a b l e 9. C l i e n t A c c e p t a n c e A n d L o c k - I n B y Cr i t i ca l i ty A n d Subs t i tu tab i l i ty Cr i t i ca l i ty S u b s t i t u t a b M 1 l y \ ^ L o w H i g h L o w In this c a s e the amount of lock- in is high and criticality is low. It is unknown how the client would accept these serv ices . In this c a s e the lock- in is high and the criticality is high a s wel l . H e n c e , the client would be less wil l ing to pu rchase these se rv i ces . H i g h In this c a s e lock-in is low. A n d s ince the serv ice is not cri t ical, the client would be very will ing to purchase these serv ices . A l though lock- in is low, the criticality is h igh. H e n c e the cl ient 's a c c e p t a n c e would depend upon which factor domina tes . Intuitively this table would s e e m correct s ince lock-in d e c r e a s e s with substitutabil ity and accep tance inc reases with criticality. In terms of our ou tcome var iab le and specif ic i ty we arrive at the fol lowing conc lus ions : high criticality and low substitutabil i ty dr ive cl ients toward an X X dec is ion . H igh substitutabil ity and low criticality m a k e cl ients c h o o s e f rom the set {NH, N V , E H , EV}, however in this c a s e specif ici ty is low. T h e ou t comes of the mixed c a s e s are unclear . A s we shal l s e e momentar i ly , criticality and substitutabil ity are weak var iab les with regard to the cho ice of A S P type in our model when compared to the other three input var iab les. However , they are powerful var iab les for explaining the dec is ion on the cl ient 's part to use an A S P or not. Wi th this in mind we cont inue in a similar fash ion a s before and a s s u m e the dec is ion to use an A S P has been made . Cri t i ca l i ty A s the astute reader has probably d e d u c e d , our a rguments regarding criticality has strong paral le ls with the arguments regarding specif ici ty. Crit ical i ty 's relat ionship to the other var iab les (that is to say : t ransact ional cos ts , specif ic i ty and substitutabil i ty), is identical to 76 that of specif ic i ty, except with respect to substitutabil ity where the inverse relat ionship is not a s powerful . However criticality, a s we shal l s e e in the next chapter , is one of the major fo rces s topping cl ients f rom purchas ing the serv ices of an A S P in the market today. R e g a r d l e s s of this, we shal l a s s u m e that the dec is ion to contract an A S P has been made . Simi lar to specif ici ty, let us first cons ider the c a s e where criticality is a l lowed to dominate: Figure 17. Client Choice Of ASP Type For P(i) As Criticality Dominates Q2 Q3 r - -Low Criticality 1 High Criticality 1 Low Criticality Q1 Q4 Niche Enterprise Span of Services Note that we cou ld have used a f igure similar to F igures 13 or 14 to indicate that criticality is indifferent to E H or E V A S P , however we have used this f igure to e m p h a s i z e the paral lels between criticality and specif ici ty. In this c a s e , b e c a u s e the client is looking for a p a c k a g e P(i) we would a s s u m e that the other factors would drive them to contract an E S S or Ful l V S P . However a s criticality inc reases , ignoring other var iab les , it wou ld m a k e s e n s e for the client to dec ide upon a col lect ion of A S P s f rom N H b e c a u s e they are no longer a s reliant upon the cont inue operat ion of one c o m p a n y (so they are, in a fash ion , diversi fy ing their 77 portfolio in the A S P market by diluting the d e p e n d e n c e on any one A S P ) and at the ext reme the highest number of A S P s would be involved in del iver ing a p a c k a g e if only F o c u s e d Solut ions were u s e d . A g a i n , a s with specif ici ty, this argument starts with a N iche V S P for an appl icat ion A(i) and starts a F o c u s e d Solut ion for a functionality f(i). Th is is highly unreal ist ic o n c e the other var iab les are cons ide red . However we cannot m a k e a s c lear a set of predict ions with regard to criticality. It is our belief that the dec is ion path a s criticality i nc reases would mimic the dec is ion path of specif ic i ty in F igure 17; however there is no driving reason f rom a criticality point of v iew to pick a A S P of the E S S or Full V S P type or v ice ve rsa . Ult imately, cl ient cho ice with regard to this var iable would be b a s e d on an empir ical study of A S P s in the market that s e e m to be the most f inancial ly healthy and the reputat ions of the A S P s in the market for reliability of serv ices a s d i s c u s s e d in Chap te r 5. S o , unfortunately, our analyt ical mode l fal ls short in fully captur ing the client behaviour in regard to criticality and further research would be required to m a k e any predict ions on this var iable when the other var iab les are taken into account . Substitutability A s exp lored earl ier substitutabil ity has a signif icant bear ing on the cho i ce of A S P type by the cus tomer . Substitutabil i ty is inversely related to switching cos t s and specif ic i ty and is or thogonal to neither. A s we have done with the previous two var iab les , let us vary the level of substitutabil i ty whi le ignoring the other var iab les. N o f igure is required to descr ibe what occu rs in this c a s e , b e c a u s e nothing happens a s substitutabil ity is i nc reased or d e c r e a s e d if the other var iab les are ignored. Wha teve r forces c a u s e the client to dec ide upon one A S P or another are still dominant b e c a u s e substitutability, in our op in ion, has no direct logical impact on A S P type cho ice when cons idered a lone. S imi lar to criticality, substitutabil ity 78 b a s e d dec is ions would likely be the result of bus iness ana lys is of the current A S P s and plays an important role in the cl ients dec is ion a s to whether or not to use an A S P to begin with. However when specif ic i ty is a l lowed to vary a s our past ana lys is has s tated they would the fol lowing dec is ion path is fo l lowed: Figure 18. Client Choice Of ASP Type For P(i) As Substitutability Increases High Sub. Q2 Q3 Low Sub. Q1 Q4 Niche Enterprise Span of Services Not surpr is ingly this is the exact opposi te of F igure 15 b e c a u s e substitutabil i ty is inversely related to specif ic i ty. A n d with regard to our output var iable, the argument for this var iable is the exact oppos i te of our ana lys is of specif ici ty. However , a s ment ioned, criticality and substitutabil ity play a profound role in the dec is ion a s to whether or not to use an A S P . 79 Summary Of Framework Predictions W e s u m m a r i z e the predict ions made in this sect ion and in the next sect ion we apply them in order to determine the strategies A S P s shou ld fol low depend ing upon what quadrant they fall into. In the fol lowing table we vary e a c h var iable over the range [Low, High] and al low the non-or thogonal var iab les to c h a n g e in va lue a s we have in F igures 13, 14, 17 and 19. T a b l e 10. F r a m e w o r k P r e d i c t i o n s V a r i a b l e D e c i s i o n Pa th a s v a r i a b l e i n c r e a s e s f r o m low to h i g h Transac t iona l C o s t s N V , N H , {EH or EV}, X X Speci f ic i ty N V , N H , E H , E V , X X Crit icality {NV, N H , E H , EV}, X X Substitutabil i ty X X , E V , E H , N H , N V T h e s e predict ions are made in light of the non-or thogonal relat ionships be tween the input var iab les a s we have descr ibed them. E v e n with a s low a granulari ty a s Q = {Low, High} these four var iab les would represent a vector s p a c e of 32 va lues , s o our ana lys is is far f rom exhaust ive on this s p a c e . A l s o , note that each of these predict ions form a testable hypo theses with regard to the cho ices of the client. Empi r ica l test ing wou ld be required to conf i rm or d ispel these hypotheses Th is conc ludes Chap te r 4 . In Chap te r 5 we focus more on the fo rces push ing a client toward, or away f rom, buying serv ices f rom an A S P then we have in this chapter . O n c e this is comple ted we shal l have mode led the full dec is ion p rocess on the part of the client. 80 Chapter 5 . A S P Framework: Dec id ing Whether To Use A n A S P A s s e e n in the prev ious chapters , the A S P model is one that is b a c k e d by many p layers in the IT wor ld, including the current big n a m e s in software. Wi l l this sof tware del ivery model s u c c e e d ? In this chapter we set forth to examine the fo rces driving cl ients toward adopt ing the A S P mode l and the fo rces impeding the growth of the A S P mode l . T h e s e fo rces are the result of an extens ive literature review or have been created by the author. W e reuse the var iab les of Chap te r 4 in our ana lys is of these fo rces wherever poss ib le , however a s the reader shal l s e e shortly, they are not enough to fully expla in the dec is ion to use an A S P . H e n c e we shal l add var iab les to this part of the mode l wh ich are strong in predict ing if a X X dec is ion will be made . T h u s we hope to capture the entire dec is ion p rocess of the client in its dec is ion to use an A S P or not, and if y e s , what type of A S P . O n c e we have mode led this we then attempt to determine the st rategies that the A S P shou ld fol low to max imize the l ikel ihood of long-term s u c c e s s in the next chapter . For the majority of this chapter we shal l use X X to indicate a cho ice not to use an A S P or Z Z to indicate that the client has c h o s e n to use an A S P . Most t imes we c a n only state that a cho ice of Z Z only impl ies that Z Z e {NH, N V , E V , EH}, without be ing ab le to s a y which e lement Z Z is equa l to in most c a s e s . S o our ou tcome var iable rema ins the s a m e , except for the fact that we are unable to state clear ly what the A S P type is in the fol lowing ana lys is . A l s o unl ike our prev ious ana lys is , we have many more var iab les , wh ich we def ine in the fol lowing sec t ion . 81 5.1. Forces That Are Causing Or Hindering ASP Growth This sect ion examines the fo rces that are driving the growth and d e m a n d for A S P serv ices . A s we have s e e n in Chap te r 3, a l though decl in ing, the investment in A S P s is still fairly large. W h a t is driving that investment? A l s o , why are the research c o m p a n i e s , desp i te their obv ious con fus ion , projecting such large revenue for A S P s ? T h e IT industry has severa l p rob lems that A S P s purport to so lve. A thorough review of exist ing literature, and ana lys is of the var ious factors the author has found and created a s a result of this review, are presented in this sec t ion . Fur thermore the author agg rega tes the factors a client would use in its dec is ion to use an A S P into the fol lowing categor ies : • Pure information technology conce rns • C o s t conce rns • Managemen t conce rns • Inertia to stay with in-house sys tems 'Pu re information techno logy concerns ' include i ssues with regard to a c c e s s , secur i ty and integration of appl icat ions. A n A S P is s u p p o s e d to bring a long all the knowledge required to set up and maintain an appl icat ion. 'Cos t concerns ' a re s imply the a n s w e r s to the quest ions "Wi l l an A S P s a v e my bus iness money, a n d how will an A S P s a v e m e m o n e y ? " A l though the express ion "time is money" d o e s hold true, most t ime conce rns are inc luded in 'managemen t concerns ' . In 'management concerns ' , we have inc luded i ssues that general ly are assoc ia ted with IT, other than technical cons iderat ions and cost . T h e s e i ssues inc lude: managemen t distract ion f rom running the bus iness , t roubles in account ing for IT cos ts , project managemen t conce rns and predict ing cost . W e find that it is in this a rea that A S P s 82 gain the most tract ion, and these are what cl ients focus on when cons ider ing a hosted appl icat ion. Let us examine each of these categor ies in detai l and look at their underly ing factors. Theor ies about e a c h individual factor are put forward below. T h e reader is caut ioned and reminded of the fol lowing p iece of symbo l logic. That if the statement "If A then B" is true, it is not a s s u m e d that "not A then not B" is a l so true un less it is explicit ly s tated. Pure information technology concerns Pure information techno logy conce rns include i ssues with regard to a c c e s s , secur i ty and integration of appl icat ions. S o m e of the i ssues that end -use rs a re c o n c e r n e d about in this regard are : 1. Geograph ica l l y Distr ibuted and/or Mobi le Work fo rce 2. Secur i ty Issues 3. Pe r fo rmance Issues 4 . N e e d for Internal and External Col laborat ion 5. Integration Issues Let us examine each of these in detail and look at how these factors wou ld impact the dec is ion to use an A S P . Geographically Distributed and/or Mobile Workforce Por tera f rames the need of modern compan ies with emp loyees who are widely distr ibuted or highly mobi le , to a c c e s s c o m p a n y appl icat ions: Distributed organizations and those with highly mobile workforces are prime candidates for a hosted applications solution. Good communication is critical to any 83 organization, but especially in environments where people don't have the opportunity to personally interact with all of their peers on a regular basis. Hosted applications enable better communication within distributed firms by providing full-time universal access to information regardless of location - bypassing the barriers and occasional outages often associated with remote access procedures. Firms with centralized employees who all work together in the same building will not enjoy many of the benefits of a hosted software solution. [Portera, 2000] Clear ly this is t ied to the concept of low site specif ic i ty of the serv ice prov ided by the A S P a s ment ioned in Chap te r 4 . S o if a c o m p a n y has a geograph ica l ly distr ibuted or mobi le workforce they have an incentive to take advantage of this property of the serv ices of an A S P . Th is argument could a lso be f ramed in terms of t ransact ional cos ts , where the cost per t ransact ion of support ing this wide sp read or mobi le workforce is more expens ive if it is done in-house than by an A S P . Either way the fol lowing theory is put forward: If the workforce of a client is highly mobi le or geographica l ly sp read out then the client is incl ined to use an A S P , otherwise they would dec ide not to use an A S P . Security Issues S i n c e cl ients often b e c o m e more securi ty and data protect ion aware , s tandard protect ions like backups b e c o m e more strictly en forced. Ironically, the secur i ty p rob lems created by deal ing with an A S P are prob lems that a l ready exist in most organ izat ions today. Most non-IT personne l and managemen t bel ieve that the t ransmiss ion of sensi t ive da ta over the Internet is unsafe , however cons ider the fol lowing: "The risk of being intercepted is not even in the top 1,000 concerns for companies today. Web sites make a point of using encryption to guard against a problem we do not have....A case in point: ICSA.net has verified with all major credit card companies, security firms, numerous banks, and law-enforcement agencies the number of cases in which credit card information was intercepted over the Internet. The answer: none. Ever. The real security issue when dealing with an ASP is much closer to home. It involves locking down security inside your own company and 84 ensuring that your ASP does the same. When problems occur, it's at either end of the data transmission, and it's clear that on that score, both companies and the ASPs they use have a lot to think about." [Gilster, 2000] A s with their own organizat ions, cl ients must insure the A S P they 've dec ided to use is taking necessa ry secur i ty precaut ions. Th is is the crux of the s tatement that A S P s improve securi ty: end users often over look securi ty gaps in their own enterpr ise, however o n c e an A S P is involved they b e c o m e hypersensi t ive to securi ty i ssues . Fur thermore, A S P s can hire ded ica ted secur i ty exper ts that none of their cl ients a lone cou ld afford. A l though securi ty is someth ing holding back s o m e cus tomers , prel iminary ana lys is of the mode l shows that secur i ty actual ly improves. S o we propose the fol lowing: If the client perce ives that A S P s improve secur i ty then they are incl ined to c h o o s e to use an A S P , otherwise they would dec ide not to use an A S P . Note that this is b a s e d upon what the client be l ieves to be true about secur i ty i ssues with regard to A S P s . A s t ime g o e s on and the A S P model b e c o m e s more famil iar that percept ion shou ld b e c o m e more accura te . Performance Issues It is a lmost imposs ib le to guarantee response t imes over the Internet b e c a u s e there are too many factors outs ide the A S P ' s control . However , for s o m e potential c l ients, a guaranteed response t ime is a must and may prevent s o m e cl ients f rom using an A S P . Packe t labell ing will enab le priority ass ignment , and thus al low I S P s to provide s o m e level of serv ice guaran tees , which cou ld be p a s s e d on to cus tomers v ia the A S P . 85 A l s o , a s ment ioned, U D D I / S O A P and packet labell ing will s p e e d the market forward. UDDI will a l low potential c l ients to link up with A S P s quickly and easi ly further reduce coordinat ion cos ts by lowering up front investment. A s stated in Chap te r 2, packet label l ing with al low a guaran teed quality of serv ice level. W e asser t that a s U D D I / S O A P mature, and packet labell ing b e c o m e s avai lab le, appl icat ions that are speci f ical ly built for the onl ine market c o m e into be ing ; hosted serv ices will gain stature a s their per fo rmance and e a s e of implementat ion app roaches that of in -house sys tems . Fur thermore, barr iers to A S P growth, such a s network response t imes, will be el iminated with packet label l ing thus reducing uncertainty about per formance. Essent ia l ly , we are stat ing the fol lowing: A s per fo rmance d i f ferences between the appl icat ion a s provided in-house or by an A S P b e c o m e smal le r the cl ient is incl ined to dec ide to use an A S P , otherwise they would dec ide not to use an A S P . Need for Internal and External Collaboration C o n s i d e r the fol lowing: Cross-organizational collaboration has become a necessity in today's business environment. However, the overhead and security issues associated with enabling collaboration via the Internet has created a dilemma in many organizations between the need for network security and the need to share information with partners, clients and other third parties. Hosted applications resolve this issue by providing access to all customer-authorized parties, with higher security levels than most individual firms can attain. [Portera, 2000] C o n s i d e r F igure 2 on page 11; in addit ion to the cus tomer of the client, the A S P cou ld a lso interact with any party that w ished to contact the client electronical ly. Tradi t ional connect ion methods s u c h a s EDI do not al low the s a m e f reedom of readily connect ing to new partners 86 or third part ies in the way that the Internet (and addit ional ly us ing S O A P and/or UDDI) . Th is is one of the pr imary advan tages of A S P s over traditional outsourc ing. For these reasons we make the fol lowing c la im: A s A S P s a n d techno log ies support ing A S P s enab le the quick and cheape r creat ion of e lectronic relat ionships comb ined with the cl ient 's increas ing need for these relat ionships, the client will be incl ined to use an A S P . Integration Issues W e cons ider sof tware integration prob lems to be the pr imary reason that c l ients would move toward us ing an E S S or Ful l V S P b e c a u s e they are very difficult to m a n a g e , a s we have ment ioned in Chap te r 4 . For example , if a client needed an account ing p a c k a g e and a H R p a c k a g e and the client u s e s the serv ices of two N iche V S P s to provide that, then the client would have to integrate these p a c k a g e s . If an E S S or Ful l V S P is u s e d then the A S P integrates those p a c k a g e s . By using an E S S or Ful l V S P the client is benefi t ing those A S P ' s exper t ise in integration. A l though it cou ld be a rgued that this is an ' ease of managemen t ' i ssue , it has been ca tegor ized with IT prob lems b e c a u s e al though integration p rob lems are usual ly difficult to m a n a g e , they are a prob lem that or ig inates with the IT itself. S i n c e the A S P and the client will have an S L A , if the client orders multiple serv ices form the A S P it is the A S P ' s prob lem to get those appl icat ions to work together. T h e only c a s e were this would not be by default is when the client takes a mixed approach where a hosted appl icat ion is to be t ied in with a local non-hosted appl icat ion. In this c a s e it is not immediate ly c lear who would be respons ib le for integrat ion, however this would be an i ssue worked out in the S L A . 87 For these reasons we m a k e the fol lowing c la im: A s A S P s inc rease the cl ients ability to integration appl icat ions easi ly , the client will be incl ined to use an A S P . Cost concerns In this sect ion the cost sav ings of A S P s are ana l ysed . T h e s e sav ings c a n be signif icant, espec ia l l y for larger c o m p a n i e s a n d c o m p a n i e s in the start-up p h a s e . O f the m a n y potential sav ings , fol lowing are s o m e of the most c o m m o n found in a review of the literature: 1. Lower up front cos ts 2. C o s t of appl icat ion can be expensed outright similar to a l ease payment 3 . C o s t of appl icat ions are sp read over many users 4. Lower Total C o s t of Ownersh ip 5. Appl icat ions that are needed infrequently can be u s e d with out having to buy the appl icat ion outright Many of these benef i ts are obv ious c o n s e q u e n c e s of the A S P mode l and billing sys tem. Hypo theses are not made on the first four points in the list above . Th is is b e c a u s e they are inherent propert ies of the A S P model and individual A S P s wou ld not c reate strategies explicit ly upon these factors. However they are part of the cl ient 's dec is ion p rocess s o they a re inc luded for the s a k e of comp le teness . Let us exp lore e a c h in greater detai l . Lower up front costs Lower up front cos ts occur for the users of a hosted serv ice s ince cos ts are deferred over the cou rse of yea rs . Th is is an obv ious advan tage over buying sof tware. Of ten compan ies make large upfront investments in software p a c k a g e s only to f ind the p a c k a g e d o e s not meet their IT requi rements. Th is is of benefit to compan ies that do not have a c c e s s to large 88 sou rces of capital and would not have been able to afford it o therwise. T h e s lowing e c o n o m y would s e e m helpful to A S P s . Uncertainty is driving many m a n a g e r s to avoid long-term commi tments [Wainewright, March 2001 ; Koblentz , F e b . 2001] . Further reducing the cos ts of us ing an A S P is the fact that s ince A S P s have t remendous exper ience customiz ing their sof tware, hosted appl icat ions are usual ly up and running much faster than in-house IT projects. Issues regarding hardware and networking are a lso great ly simpl i f ied under the A S P mode l , driving the cost of creat ing this relat ionship lower. However , the pr ice of starting a relat ionship, whi le still lower, is not a s inexpens ive a s it once was . Up until recently A S P s fo l lowed a pure "pay per play" mode l . However , Cor io and U S i (two of the industry leaders) swi tched to a down-payment mode l [Kerstetter, March 2001] . W h a t impact this will have on client accep tance , and whether other A S P s will fol low this lead , is yet to be s e e n . Rega rd l ess , the econom ics behind the rent ve rsus buy argument still ho lds true. W e feel that this factor is not a driving force toward the use of an A S P until a new functionality, appl icat ion or p a c k a g e is needed b e c a u s e cl ients that a re sat is f ied with their current IT infrastructure would not buy the serv ices of an A S P s imply b e c a u s e it is inexpens ive to start. However , once there is a need for a new set of funct ional i t ies, this b e c o m e s a powerful motivator for purchas ing the serv ices of an A S P . Cost of application can be expensed outright similar to a lease payment C o s t of appl icat ion can be e x p e n s e d outright (like a lease payment) without having to deprec ia te the cost of sof tware. Under the traditional sof tware a s a product mode l , the huge upfront cos t had benefit over many reporting per iods (whether those per iods are months or years) . For this reason , the C a n a d a C u s t o m s and R e v e n u e A g e n c y d o e s not al low the 89 c o m p a n y to e x p e n s e that outlay of c a s h , but treats it a s the pu rchase cost of capi ta l . Th is capital investment cou ld never be fully e x p e n s e d , therefore the tax benef i ts are only on the amount the product d e c r e a s e s in va lue accord ing to capital cos t a l l owances a s set forth by the government . Th is is a pr imary advantage of us ing an A S P over purchas ing sof tware. Th is is a cos t factor that would drive cl ients to dec ide to pu rchase an A S P when new functional i t ies are required. Cost of applications are spread over many users It has been est imated that the cost of maintaining a ded ica ted server c o m p a r e d to sha red server resources c a n be a s much a s three t imes higher in a 15 month t ime f rame [ S C N , 2000] . Th is d o e s not take into account maintaining the sof tware that res ides on those servers . S o a s a result the cus tomer of an A S P s a v e s on infrastructure cos ts . More importantly, the cus tomer s a v e s b e c a u s e they do not have to pay for the entire appl icat ion; they only pay for the part they are us ing (if per seat or per t ransact ion billing is used) . Th is has the s ide effect of al lowing smal l users to a c c e s s heavy-duty appl icat ions that were previously unaffordable, the effect of which cannot be unders ta ted. Fur thermore this is one of the pr imary advan tages of using an A S P over purchas ing sof tware. Lower Total Cost of Ownership Lower Total C o s t of Ownersh ip ( T O C ) exist s ince ma in tenance of the appl icat ion and availabil i ty are the concern of the A S P . Fur thermore, cost sav ings are rea l ized b e c a u s e the cos t of maintaining a desk top client, under the cl ient/server mode l , is est imated at $10 ,000 dol lars per year in s o m e organizat ions [ S C N , 2000] . Under a full A S P sys tem, low-end desk tops with a web browser can replace most of these . T h e sav ings on installation and 90 main tenance of desk top software could be in the mil l ions for larger corporat ions. Lower t ransact ion cos ts a lso play a role b e c a u s e the cost of the IT infrastructure to support the appl icat ion is sp read over many users . A l s o comput ing power, espec ia l l y on the server s ide, is better uti l ized. Applications that are needed infrequently can be used without having to buy the application outright W a r e h o u s e layout p lanning and W a r e h o u s e locat ing software would be a good examp le of such sof tware. Many compan ies use wa rehouses but few spec ia l i ze in opt imizing their layout, hence buying floor layout software m a k e s little s e n s e for t hese compan ies . T h e s e 'one of f appl icat ions would have to be pu rchased , or a c c e s s wou ld be ga ined through consul t ing c o m p a n i e s that would be us ing the software. Howeve r in the A S P model cl ients pay on a per use bas is , s o there is potential for signif icant cost sav ings if the appl icat ion is a one t ime use or infrequently used p iece of software. H e n c e the fol lowing c la im is made : If a client intends to use an appl icat ion rarely they are more incl ined to have an A S P provide that appl icat ion (in other words m a k e the dec is ion to use an A S P ) . Management concerns M a n a g e m e n t conce rns include but are not l imited to the fo l lowing: 1. N e e d to quickly sca le up or down based on work loads and growth rates 2. Yea r l y cos ts of appl icat ion can be predicted more accurate ly 3. A S P s c a n be held more accountab le for shor tcomings than an internal IT depar tment . 91 4. Upgrad ing i ssues Let us explore these in detai l . Need to Quickly Scale Up or Down Based on Workloads and Growth Rates Portera m a k e s the fol lowing suggest ion with regard to this i ssue Fast-growing companies and those with highly variable workloads may be well-advised not to make the capital outlay associated with a client/server application with limited future capacity, and instead opt to adopt the "pay as you go" model offered by the ASP model - with virtually unlimited scalability. If, however, a company's workload and growth rate is stable and projected to remain so, then a client/server application may make more sense. [Portera, 2000] Obvious ly what Por tera has sa id about c l ient/server can be ex tended to main f rame sys tems or any in -house sys tems a s wel l . Th is is b e c a u s e in-house sys tems are built to handle a part icular range of work load. If the workforce or work load is curtai led the investment in the resultant e x c e s s capac i ty is was ted s ince in-house sys tems are often very site speci f ic or physical ly spec i f ic . If the work load or workforce is rapidly inc reas ing , an in -house sys tem cou ld be overwhe lmed with d e m a n d levels it w a s not des igned to handle . The re is not a s bad an e x c e s s capac i ty situation for the client if an A S P is u s e d s ince the client is only paying for the part of the appl icat ion they are us ing. Conve rse l y , A S P s tend to sca le much better than in-house solut ions [Portera, 2000] . Th is leads to the fol lowing hypothesis : If a client is expect ing a volati le growth pattern the client will tend to c h o o s e to use an A S P . 92 ASPs can be held more accountable for shortcomings than an internal IT department. Serv i ce Leve l Ag reemen ts ( S L A s ) structure is very important in this regard. If an in-house project g o e s over budget in terms of cost or t ime, the c o m p a n y has little recourse . Th is is often the c a s e with consu l tanc ies that make no guaran tees (or often break promises) with regard to cost and dead l ines . However S L A s form a legal and en forceab le document between two independent entit ies. It is reassur ing to the client to have s o m e protection against these all too c o m m o n events in the IT wor ld. Th is concept is d i s c u s s e d extensively in the literature on outsourc ing. Th is leads to the fol lowing theory: the requirement of a set of funct ional i t ies that traditionally involve cos t overruns, de lays or extens ive consul tant activity will dr ive client cho ice toward us ing an A S P . Upgrading Issues A S P s have an inherent advantage with regard to tradit ional E R P and cl ient/server p a c k a g e s . M a n y compan ies that bought into E R P s are not sat is f ied and are probably hesitant to upgrade their E R P p a c k a g e s . Large IT projects tend to take years , and once they start running wel l , an upgrade to the p a c k a g e is on the hor izon (this is often referred to a s ' technology indigest ion' in the literature on A S P s ) . Th is is not go ing to change , in fact: "the rapid p a c e at which technology is chang ing is only go ing to inc rease even more" [Bushaus , Sep t 2000] . T h e prob lem of software becoming obso le te is no longer a prob lem that the cl ient has to worry about if an A S P is used b e c a u s e upgrad ing the appl icat ions is now the conce rn of the A S P . 93 This leads to the fol lowing hypothesis : If the client requires a set of funct ional i t ies then A S P s lower the cos ts of us ing that appl icat ion by el iminating or reducing the cos ts of upgrading thus increas ing the l ikel ihood that the use of an A S P will be c h o s e n . Inertia to stay with current In-House systems If a client dec ided to use an A S P they may have to walk away form exist ing IT investments, IT human resources and traditional methods of deal ing with IT i s sues . W h a t prob lems is this caus ing? Let us break down this i ssue into four major componen ts . 1. R e s i s t a n c e f rom internal IT staff 2 . S u n k cos ts of exist ing sys tems 3. T h e knowledge that is lost with IT staff is very hard to regain 4. Cl ient uncertainty about the future for A S P s T h e s e are factors that, a s they increase, encourage the retention of in -house IT sys tems . Let us examine e a c h . Resistance from internal IT staff R e s i s t a n c e f rom internal IT staff c a n b e signif icant. A s in [Phair, 2000] the prev ious 'IT champ ion ' may not be will ing to let go of current sys tems they had advoca ted and worked hard to get. Th is IT champ ion would a lso feel somewhat fool ish admitt ing that sys tems they had advoca ted are a l ready obsole te and may not endorse an A S P solut ion for that reason . Fur thermore, IT staff wou ld be aware of the fact that A S P solut ions wou ld m a k e many of their jobs redundant, and that "knowledge empi res" they have built up a round exist ing sys tems would fall by the ways ide [Phair, 2000] . 94 Th is leads us to the fol lowing hypothesis : Internal IT staff res is tance c a n reduce the l ikel ihood of the client choos ing to use an A S P . Sunk costs of existing systems S u n k cos ts of exist ing E R P , C l ien t /Server or Main f rame sys tem are hard to walk away f rom for obv ious reasons . [Char ley, 2000] refers to this a s the sunk cost effect, when the bias is towards stay ing with the incumbent IT technology, even when the prof i t -maximizing dec is ion for a f irm would be to c h o o s e the A S P . Th is is of great conce rn to A S P s that rep lace the most expens ive sys tems s ince potential c l ients may use this logic to justify not us ing those A S P ' s solut ions. Th is leads to the fol lowing theory: Large or recent sunk cos ts reduce the l ikel ihood of the client choos ing to use an A S P . The knowledge lost with IT staff is very hard to regain it if need be O n c e an A S P is adopted and internal IT staff is lost, the knowledge that is lost with them is very hard to regain and is d i s c u s s e d widely in the literature on outsourc ing [Butler, 2000] . A n interesting account of the c o n s e q u e n c e s of this effect is found in [King, 1992]. Th is is c lose ly related to the fear of A S P fai lure. IT staffing i ssues represent a doub le -edged sword for A S P s , such that it is a reason to buy and not to buy A S P se rv i ces . Th is is c lose ly related to the substitutabil i ty var iable in the previous chapter . Th is impl ies that the lock-in that results b e c a u s e the client has reduced staff is a deterrent to purchas ing the serv ices of an A S P . W e shal l not b a s e a hypothes is about how the substitutabil i ty of the serv ice impacts the cho ice of the cl ient. However this idea is expanded further in the next factor. 95 Uncertainty About the prospects for ASPs W e have n a m e d the fol lowing cyc le the "Ca tch 22 syndrome" it cons is ts of: a) C l ients are not sure which A S P mode ls will fail s o they do not buy in. b) A S P s cannot get cl ients b e c a u s e they are all wait ing to s e e which A S P s will s u c c e e d c) A S P s fail without cl ients d) S e e a . Th is is where s o m e of the pall of the dot corns hangs over the h e a d s of A S P s ; cus tomers have little faith in many IT compan ies and ideas. However this fear many not be unfounded, here 's why: The Gartner Group predicts that 60% of them will fail in the next year [2001]. And by the way, that is a lot more serious than an online pet store suddenly running out of cat chow After all, the demise of a company that runs your critical software would be a little like the power company's cutting off your service for good. [Lashinksy, 2000] A g a i n , Mark Ches tnu t ' s quote c o m e s back to us: " C u s t o m e r s are de lay ing pu rchase dec is ions b e c a u s e they don't know whether the A S P s will s u r v i v e . . . A S P s aren't getting enough cus tomers to surv ive." Mission-critical applications such as ERP and C R M were among the first to be offered by ASPs, but they can be a tough sell to customers concerned about turning over control and security of their data to unproven service providers. Even a successful sale results in a small number of seat sales because the majority of a customer's employees do not need access to those applications, analysts say. [Hagendorf, 2000]. If an A S P who prov ides miss ion critical appl icat ion fai ls, results cou ld be devasta t ing, and it is for this reason many analys ts project 'commodi ty appl icat ions ' may be the earl iest growth market s p a c e . "Whi le ana lys ts say the A S P model has been s low to ca tch on , they bel ieve 96 a ubiqui tous appl icat ion such a s messag ing cou ld be the ticket to ma ins t ream s u c c e s s . " [Hagendorf, 2000]. Fur thermore "Cus tomers a l ready think of e-mai l a s a utility," s a y s Laur ie M c C a b e , v ice pres ident a n d serv ice director at consul t ing firm Summi t S t ra teg ies . "In every survey I've s e e n , m e s s a g i n g and col laborat ion were in the top half of those appl icat ions cus tomers would cons ider host ing." [Hagendorf, 2000] Cons ide r ing the A S P Industry Survey quoted earl ier in Chapte r 3 and the ana lys is of Chap te r 4, this wou ld s e e m to hold true. T h e literature about A S P s f ocuses strongly upon this prob lem. C u s t o m e r s are not moving to the mode l a n d that is caus ing A S P s to fai l . Th is is c lose ly re lated to the ana lys is in Chap te r 4 that lead to Tab le 8. Th is is the c a s e with any new market and c a n only be ove rcome in t ime. W e now state the previous theor ies explicit ly in the terminology of this chapter : A s substitutabil ity of an A S P ' s serv ice d e c r e a s e s the l ikel ihood of the cl ient choos ing to use an A S P d e c r e a s e s . Fur thermore, a s the substitutability i nc reases s o d o e s the l ikel ihood of a dec is ion to use an A S P . T h e s a m e set of s ta tements hold true for criticality of the serv ices in the reverse. A s criticality of an A S P ' s serv ice d e c r e a s e s the l ikel ihood of the client choos ing to use an A S P d e c r e a s e s . A s the criticality i nc reases s o d o e s the l ikel ihood of a dec is ion to use an A S P . Low criticality, high substitutabil ity and low specif ici ty appl icat ions cou ld be cons ide red the most ' A S P friendly' a s these appl icat ions are the eas ies t to se l l , and cheapes t to cus tomize . High criticality, low substitutabil ity and highly speci f ic appl icat ions are the least ' A S P friendly' s ince they are the hardest to sel l and most expens ive to cus tomize . T h e next sect ion s u m m a r i z e s the f indings of this chapter . 97 5.2. Summary Of Predictions And Findings Of ASP Framework Throughout the prev ious sect ion we made hypothes is and theor ized about the affects of severa l factors in the cl ients cho ice of whether to use an A S P or not. W e bring each of these hypo theses together to form the s e c o n d part of our A S P f ramework, which is summar i zed in the fol lowing set of tab les. Note that s o m e of these f indings are A S P speci f ic and s o m e can be appl ied to outsourc ing in genera l , we indicate those factor that are A S P speci f ic . Table 11. Factors Increasing The Likelihood Of ASP Usage Highly mobi le or geographica l ly d iverse workforce ( A S P speci f ic) Cl ient perce ives that A S P u s a g e improves securi ty Pe r fo rmance of appl icat ions s a m e onl ine a s in-house Increased need for external electronic relat ionships ( A S P speci f ic) Cl ient des i re to integrate appl icat ions (Advantage over purchas ing software) C o s t and t ime reduct ions in creat ing external electronic re lat ionships ( A S P speci f ic) Cl ient needs an appl icat ion infrequently (Advantage over purchas ing software) Cl ient is expect ing volati le growth pattern in work load or workforce ( A S P speci f ic) Cl ient expectat ions of cost overruns, de lays or extens ive consul tant activity (Advantage over purchas ing software) High substitutabil i ty of serv ices Low criticality of se rv ices T h e next table prov ides the set of our hypo theses with regard to factors that c a n inc rease the l ikel ihood of the cl ient 's dec id ing not to use an A S P . Th is is different f rom the above 98 b e c a u s e we have found that not all of the above factor can d e c r e a s e the l ikel ihood of us ing an A S P a s they d e c r e a s e . Table 12. Factors Increasing The Likelihood Of Not Using An ASP Low mobility or geographica l ly central workforce Cl ient perce ives that A S P u s a g e d e c r e a s e s secur i ty ( A S P speci f ic) Pe r fo rmance of appl icat ions worse onl ine a s compa red to in -house ( A S P speci f ic) Low substitutabil i ty of se rv ices High criticality of se rv ices T h e s e tab les comb ined with Tab le 9 of Chap te r 4 d isp lay the results of the f ramework the A S P f ramework this thes is set out to create. T h e s e three tab les attempt to model the cus tomers dec is ion p rocess a s to whether or not to use an A S P and if s o , which of the four types c rea ted in this thes is . Contrast Between ASPs And Outsourcing A s d i s c u s s e d in Chap te r 2, the fact that A S P s are access ib l e over the Internet is one of their pr imary advan tages over traditional outsourc ing of IT funct ions. Th is has severa l important impl icat ions: • B e c a u s e the Internet can be connec ted v ia s tandard networking equ ipment and protocols, A S P s reduce the need for infrastructure investment. • A S P se rv i ces can go 'l ive' in a much shorter t ime f rame then other outsourc ing methods b e c a u s e the serv ices of an A S P are a l ready running. • Prev ious ly many smal ler cl ients cou ld not afford the investment required for larger appl icat ions and serv ices with traditional outsourc ing. However , b e c a u s e the 99 serv ices of an A S P can have a low starting cost these cl ients c a n now have a c c e s s to these appl icat ions. • T h e Internet a l lows third part ies who wish to interact with the cl ient to do s o with greater e a s e . Other forms of outsourc ing do not a l low the t ransparency to third par t ies, such a s cus tomers , a s an A S P does . T ranspa rency is where the third party be l ieves that they are interacting with the cl ient 's IT sys tems , when in reality they are interacting with the A S P ' s IT sys tems . Th is is depic ted in F igure 2 of Chap te r 2. • A S P s c a n suppor t the rapid creat ion bus iness - to -bus iness re lat ionships be tween the client and part ies such a s suppl iers by act ing a s a buffer on the Internet. Th is can be attributed to the fact that A S P s not only can integrate appl icat ions internal to organizat ions, but can integrate cross-organ izat iona l appl icat ions a s wel l . Fur thermore, A S P s can act provide the facil i t ies for ' t ranslat ion' between these appl icat ions. • B e c a u s e the serv ices of an A S P are often pu rchased on a 'per user ' bas i s , A S P s are better sui ted to highly volati le work loads. • A S P s support a highly mobi le or geographica l ly d iverse work force better then tradit ional outsourc ing b e c a u s e of the universal accessib i l i ty of the Internet. N o w that we have shown why a client would opt to use an A S P a s o p p o s e d to buying, bui lding or outsourc ing in the traditional ways , we now move on to the st rategies A S P s shou ld adopt in light of this. 5.3. ASP Strategies As Dictated By ASP Framework Th is sect ion out l ines the strategies an A S P shou ld adopt. Th is is b a s e d on what type of A S P max im izes the l ikel ihood of s u c c e s s and what type of market structure it shou ld expect . 100 This is b a s e d on the f indings of this and the prev ious chapter . W e shal l use the A S P f ramework c reated in Chap te r 4 and 5 a s wel l a s the f indings in the other chapters to outl ine these st rategies. W e start by reviewing the f indings of Chap te r 4 . In Chap te r 4 we found that t ransact ion cost theory and our own f indings would indicate that cl ients would tend to use a Enterpr ise Suppor t Se rv i ce provider or a Ful l V S P , un less they are conf ident that they will not outsource more than a limited amount of functionality or if t ransact ional cos ts are s o extremely low that coordinat ion cos ts incurred are cons idered worthwhile when compa red to the benefit of being ab le to mix and match the se rv i ces of any set of A S P s and ISVs . W e do not feel that t ransact ional cos ts are low enough in this market b e c a u s e of integration related cos ts [King, 1992], s o it is our theory that cl ients will tend to use either Enterpr ise Suppor t Serv ice providers or a Ful l V S P s . Th is impl ies that the A S P will take on the form shown in F igure 12. In that f igure the client interacts with one A S P for a p a c k a g e P(i). Th is has the benefit for the client of offsetting all the cos t s of coordinat ion and integration to the A S P . Under this market structure, N iche V S P ' s serv ice would be resold though an E V or E H A S P b e c a u s e t ransact ion cos ts are min imised by having all IT funct ions hand led by one contact. Th is a lso min imises t ransact ion cos t s for the client with regard to connect ing to new appl icat ion serv ices b e c a u s e an E V or E H A S P would be respons ib le for integrating that new serv ice into the cl ient 's p a c k a g e of se rv i ces . W e predict the two main retail mode ls will be Enterpr ise Suppor t Se rv i ce providers and Full V S P s , and that both shou ld both have lower specif ic i ty se rv i ces (which have less lock-in accord ing to our theory) thus al lowing them to gain cus tomers . O n c e cl ients are us ing their se rv i ces their high specif ic i ty would be used to max imize lock- in. N iche V S P s must focus on lowering customizat ion cos ts , and both N iche V S P s and F o c u s e d Solu t ions must improve interoperabil i ty by us ing techno log ies and s tandards that 101 reduce the cost of connect ion , thus lowering t ransact ional cos ts . However , for N iche V S P s a n d F o c u s e d So lu t ions a s a group, this will i nc rease compet i t ion a s swi tch ing cos ts a re lowered. Fur thermore, N iche V S P s must maintain their focus on a core serv ice . T h e y may gain the highest margin by sel l ing serv ices to the E S S and Ful l V S P mode ls , hence cutting down on interaction cos ts with cus tomers . S o m e N iche V S P s that own se rv i ces with broad appea l , such a s account ing p a c k a g e s , may be able to operate without depend ing heavi ly upon an Full V S P or E S S for resa le . However most N iche V S P s will require an aggregator or resel ler to provide them with an aud ience of poss ib le cl ients, espec ia l l y the providers of infrequent u s e appl icat ions. F rom a specif ic i ty perspect ive, the client would prefer to be ab le to dea l with N iche V S P s or F o c u s e d Solu t ions. Th is would al low the client the highest deg ree of customisabi l i ty. Techno log ies s u c h a s U D D I , S O A P a n d X M L a re at tempts to p u s h the market in that direct ion by lowering the t ransact ional cos ts of these relat ionships. However , the integration p rob lems the client would face are great, and a s a result the t ransact ional cos ts of maintaining severa l relat ionships would be s o high, that we conc lude that the client would move toward an E S S or Full V S P . S o m e research in other a r e a s of IT outsourc ing arr ives at the s a m e conc lus ions [King, 1992; Loh , 1992]. Next generat ion deve lopment tools and techn iques shou ld further i nc rease the t ransact ional cost sav ings by reducing the cos ts of increasing specif ici ty. A S P s have c h a n g e d the ISV market and how I S V s interact with the end user of their products [Bushaus , Sept . 2000 ; S C N , 2000] . It is unc lear what impact this will have on ISVs , however there is the potential for them to b e d i sp laced . Network ing improvements that m a k e Q o S more predictable will instil greater client con f idence and a lso spur A S P growth. 102 Enterpr ise Suppor t Se rv i ce providers or Full V S P s will opt imal ly have both low crit icality/high substitutabil ity offer ings to gain cl ients and high crit icality/low substitutabil i ty offering to lock-in cus tomers a s found in Chap te rs 4 and 5. A s they sca le cus tomers f rom low criticality solut ions to high criticality solut ions, lock-in is i nc reased . However , cus tomers are not blind to the concept of lock-in or substitutability, and A S P s may have to p roceed more caut iously than E R P s did b e c a u s e cus tomers may feel they were taken advan tage of in previous enterpr ise implementat ions, and A S P s are in a f inancial ly weak posi t ion. E V A S P s gain the most traction in markets that have ' spec ia l ' n e e d s . Hea l thcare and governments fee l , maybe rightly s o , that they are s o different f rom other organizat ions that they require solut ions a imed at them a lone [Brown, Ju ly 2000 ; M c C a b e , F e b . 2001] . Th is is b e c a u s e most software that has been des igned for bus iness in genera l requires an extra amount of speci f icat ion for them. Th is is due to the fact that they are different f rom other organizat ions and have a different set of criterion for s u c c e s s than most for profit b u s i n e s s e s . T h e Enterpr ise Suppor t Serv ice providers and Ful l V S P s will c o m e into direct compet i t ion in these vert ical markets , and it is unclear who will prevai l . I S V s have the most to lose in this market. If they don't move to an A S P mode l they will lose direct contact with most of their current cus tomers . A s cus tomer loyalty and lock- in swi tches to the A S P away f rom the ISV, the future role for the 'brand n a m e s ' in sof tware starts to b e c o m e unclear. Ana lys ts are a l ready compla in ing about m y S A P and Orac le ' s B u s i n e s s Onl ine offering only its own software. ISVs are go ing to have a hard t ime compet ing in the utility mode l that s e e m s to be on the hor izon a s brand n a m e of sof tware d imin ishes in va lue relative to the va lue of the A S P ' s name. 103 5.4. Summary Th is conc ludes Chap te r 5. In this chapter we examined the factors in the cl ient 's dec is ion to use an A S P or not. In sect ion 5.2 we const ructed two tab les which represent the s e c o n d half of our A S P f ramework and in sect ion 5.3 we stated the st rategies A S P s shou ld adopt a s a c o n s e q u e n c e of the predict ions in our mode l . W e now move on to Chap te r 6. 104 Chapter 6 . C o n c l u s i o n T h e A S P model is still in its early s tages of deve lopment and many of the current bus iness mode ls are most l ikely f lawed [Lashinksy, 2000] . However its s u c c e s s is b a c k e d by most of the big n a m e s in software and cus tomers are starting to accept the idea of outsourc ing their appl icat ions in this fash ion . Arguab ly , current apprehens ion about A S P s is, in part, a funct ion of larger conce rns assoc ia ted with many Internet software compan ies , namely the dot corns. A s t ime g o e s on and s o m e A S P s are success fu l , these conce rns will be reduced and cus tomers may be more will ing to e m b r a c e A S P s . Del l , Microsoft , S u n , Hewle t t -Packard and Orac le have entered the market lending their credibil ity to f ledging A S P c o m p a n i e s though partnerships and cert i f ication p rograms [Fonseca , Nov. 2000 ; Trager , F e b 2 0 0 1 ; Tillett, Mar 2 0 0 1 ; Burke , Nov. 2000] . A S P s will start gaining cus tomers a s dot c o m fears start to sl ip away and network improvements will el iminate many cus tomer conce rns [Mears , F e b . 2001] . Th is thes is set out answer the fol lowing quest ion: W h a t are the determinants of A S P s u c c e s s ? T o answer this quest ion we first set about def ining and f raming the concep ts and i ssues surrounding the A S P doma in . W e then introduced a new stratif ication of A S P s b a s e d on target market ve rsus the span of the p r o c e s s e s they support . Hav ing done this, we created an analyt ic f ramework in an attempt to predict what bus iness mode ls and strategies will be success fu l in the appl icat ion serv ice provider market s p a c e and what s h a p e that market will have . T o do this we mode led the dec is ion p rocess which the cl ient u s e s to dec ide whether or not to use an A S P and if s o which A S P type to use . W e conc luded the fol lowing st rategies max imize the l ikel ihood of long-term s u c c e s s depend ing on whether 105 they are of type Full V S P or Enterpr ise Suppor t Se rv i ces ve rsus types N iche V S P or F o c u s e d Solut ion: Table 13. Summary Of Successful Strategies EV or EH NH or NV K e e p customizat ion cos ts low. K e e p customizat ion cos ts low by us ing m a s s customizat ion and building expert ise in the serv ice of fered. Offer both high and low subst i tutable serv ices Foster relat ions with aggregators that E V and E H are predicted to b e c o m e Mix and match best of b reed software S tay f ocused on core knowledge F o c u s on integration and functionality de l ivered, not of the n a m e s of the ISVs that c reated solut ions Uti l ise UDDI and S O A P (or s imi lar technologies) for their connect iv i ty s tandards to drive down the cos t s of connect ing to E V and E H A l s o a s a result of our ana lys is , we stated that t ransact ional cos t s and asse t specif ic i ty in this market are very high. H e n c e the A S P market will take on a h ierarchy structure in which cl ients will opt to interact with one A S P that sat isf ies the m a x i m u m number of IT requi rements a s depic ted in F igure 12. S i n c e this thes is at tempts to predict c h a n g e s that have not occur red yet and market structures in a still forming market s p a c e a s imple conceptua l analyt ic f ramework w a s const ructed instead of us ing a systemat ic empir ical study. A conc lus ive test of this model and the predict ions m a d e a s a result of it will, therefore, require further empir ical work. Empi r ica l s tud ies of current A S P s and A S P cus tomers to determine reasons for buying serv ice a s wel l a s benef i ts real ized could a lso further research in this f ield greatly. Fur thermore, there is a great dea l of ana lys is that c a n be done on the interact ions of the var iab les used within the A S P f ramework that were cons ide red beyond the s c o p e of this document . 106 Bib l iog raphy [I] A lexander , M ichae l (Jan . 2001) "Microsoft R e v e r s e s C o u r s e O n B iz A p p s , " Internetweek, Issue 843 , p9-10 . [2] Ap ice l la , Mar io (June 2000) "Al ternat ives to the traditional A S P mode l , " InfoWorld, V o l . 22 Issue 26 , p65-66. [3] A r iba Inc.; IBM Corpora t ion , Microsoft Co rp . (Sept. 2000) "UDDI Execut ive Whi te Pape r , " U D D I . O R G . [4] A r iba Inc.; IBM Corporat ion, Microsoft Co rp . (Sept. 2000) "UDDI Techn ica l Wh i te Paper , " U D D I . O R G . [5] A u n , F red (Feb . 2001) " A S P in retail war," ZDNet eWeek (www.zdnet .com). [6] Ben jamin , Rober t I.; Ma lone , T h o m a s W . ; Y a t e s , J o a n n e (June 1987) "Elect ron ic Marke ts and Electronic Hierarch ies, " Communications of the ACM, V o l . 30 Number 6, P 4 8 4 - 4 9 7 . [7] Bernard , A l len (Jan. 2001) , "Is Th is How A S P s Wi l l Surv ive in 2 0 0 1 ? " ASPnews.com, (www.aspnews.com) . [8] Bernard , A l len (Feb . 2001) , "Host ing M o v e s up the Va lue C h a i n , " ASPnews.com, (www.aspnews.com) . [9] B lum, Jona than (Sept. 2000) , " M a k e no Mis take: A c c e s s Is In," The Net Economy, V o l . 1 N u m b e r 1, p82-83 . [10] B rown, Jus t ine (July 2000) , " P a y T o Play , " Government Technology, V o l . 13 Issue 9, p32-36. [ I I ] Burke , S teven ; Hagendor f Follett, Jenni fer (Nov. 2000)" lndustry big gun back A S P s , " Computer Resellers News, Issue 921 , p3-4. [12] B u s h a u s , D a w n (Oct. 2000) , "An A S P Real i ty C h e c k , " The Net Economy, Vo l . 1 N u m b e r 2, p72-73 . [13] B u s h a u s , D a w n (Dec . 2000) , " A S P s C o u l d get Swa l l owed W h o l e , " The Net Economy, V o l . 1 N u m b e r 4, p34. [14] B u s h a u s , D a w n (Feb. 2001) , " A S P s G o Into Surv ivor Mode , " The Net Economy, Vo l . 2 Number 4 , p60-62 . [15] B u s h a u s , D a w n (Sept . 2000) , " A S P s get supe rs i zed , " The Net Economy, V o l . 1 N u m b e r 1, p51-53 . 107 [16] B u s h a u s , Dawn (Nov. 2000) , "F inance : Appl icat ion Serv i ce Prov iders , " The Net Economy, V o l . 1 Number 3, p82. [17] But ler (Ed.) , Jane t (2000), "Winn ing the Outsourc ing G a m e , " Auerbach. [18] Char ley , J o a n n e (2000) "Appl icat ion Serv i ce Prov iders : an industry ana lys is , " Simon Fraser University Master's Project. [19] Chr is topher , Al istair (Oct. 2000) "The IT Hot List: Lead ing V C s S a y W h e r e the Smar t M o n e y is H e a d e d , " Venture Capita/Journal, p1-6. [20] C o o k e , J a m e s (Sep . 2000) "Logis t ics e x c h a n g e s and A S P s : O n the evolut ionary path," Logistics Management and Distribution Report, p E 8 - 1 3 . [21] Cornf ie ld , G e n e (Nov. 2000) " A S P Partner F ramework and Serv i ce Del ivery Mode l , " Microsoft Corporation. [22] C r u z , M ike ; To rode , Cr is t ina (Feb. 2001) " H P g ives A S P program a boost ," Computer Resellers News, Issue 933 , p12. [23] Der ing , Bill (Jan 2001) " A S P s : W h i c h Wi l l S u r v i v e ? " ASPnews.com, (www.aspnews.com) . [24] Doug lass , Michel le (Aug. 2000) "Is A S P model right for y o u ? " Computer Dealer News, Vo l . 16 Issue 16, p16-19 . [25] Dubie , Den i se (Nov. 2000) "Obl icore to give A S P s a little help meet ing S L A s , " NetworkWorld, V o l . 17, p43-44. [26] Eng lander , Ernest J . (Nov. 1988) "Techno logy and Ol iver W i l l i amson ' s Transac t ion C o s t E c o n o m i c s " Journa l of E c o n o m i c Behav iour and Organ iza t ion , N u m b e r 10, p341-353 . [27] F o n s e c a , Br ian (Nov. 2000) "Microsoft , U S i swing dea l for . N E T host ing," InfoWorld, V o l . 22 Issue 47 , p16 . [28] Gi ls ter , Pau l A . (Sept. 2000) "Safety First," Workforce, V o l . 79 Issue 9, p52-57. [29] G r e e n b a u m , J o s h u a (Feb. 2000) "ASP- i r i ng to G r e a t n e s s , " Software Magazine, Vo l . 20 Issue 1, p72-74. [30] G r u h n , Marty (June 2000) "Unravel ing the A S P P u z z l e : T h e Four R u l e s for S u c c e s s , " Market Strategy Report, Summi t St rateg ies. [31] Hagendor f Follett, Jenni fer (Aug 2000) " A S P s starting to get the m e s s a g e , " Computer Reseller News, Issue 907, p63-66. [32] Harr ington, L i sa H. (Aug 2000) "Po in t -and-Cl ick to anyp lace on Ear th , " Transportation & Distribution, V o l . 41 Issue 8, p93-96. [33] Hein le in , Scot t (Feb. 2001) " C r a s h C o u r s e for A S P s , " ASPnews.com, (www.aspnews.com) . 108 [34] Hende rson , L i za (Jan 2001) "F inding the Market for D i sposab le A p p s " ASPnews.com, (www.aspnews.com) . [35] Jas t row, Dav id (Aug 2000) "Gett ing the A S P model right," Computer Reseller News, Issue 908 , p119-122 . [36] Kap lan , S teven ; Mohanbi r , Sawhney , (Feb 2000) "The E-Hub Hubbub , " WorldLink, Vo l . 13 Issue 1, p73- 77 . [37] Kap lan , S teven ; Mohanbi r , Sawhney , (June 2000) " E - H u b s : T h e N e w B 2 B Marke tp laces , " Harvard Business Review, p97-103 . [38] Kerstet ter, J i m (March 2001) "Sof tware Shakeou t : App l ica t ion se rv ice prov iders p romised to t ransform the way bus iness is done. W h a t h a p p e n e d ? " Business Week, Issue 3722 , p72-78. [39] K ing , R a c h a e l (Sept. 2000) , "F ive S teps to A S P S u c c e s s , " The Net Economy, V o l . 1 Number 1, p56-62 . [40] K ing , W . R . (1994), "Strategic Outsourc ing Dec is ions , " Information Systems Management, 11 (4), pp. 58 -61 . [41] Kob lentz , E v a n (Feb. 2001) " A S P ' s w o e s mirror those of the industry," ZDNet eWeek (www.zdnet .com). [42] Kob lentz , E v a n (Feb. 2001) " A T & T exec to A S P s : 'It's Jung le warfare ' , " ZDNet eWeek (www.zdnet .com). [43] Ko lbasuk M c G e e , Mar ianne (Jan 2001) "Out look for 2 0 0 1 , " InformationWeek, Issue 819 , p42 -61 . [44] Kucharvy , T o m (May 2000) "Hos ted Appl icat ions: W h a t C o m e s Nex t? " ASPnews.com, (www.aspnews.com) [45] Lash inksy , A d a m (Oct. 2000) "Don't get bitten by the A S P s , " Fortune Magazine, V o l . 142 Issue 7, p 2 5 4 - 2 5 7 . [46] Longwel l , J o h n (Oct. 2000) "Debat ing the A S P Mode l , " Computer Reseller News, Issue 914, p12-13 . [47] Mase l l i , Jenni fer (Feb . 2001) "Cor io st icks to A S P strategy," InformationWeek, Issue 8 5 1 , p14. [48] Mase l l i , Jenn i fe r (Oct. 2000) "Orac le overhau ls A S P pr ic ing, S L A s , " InformationWeek, Issue 810 , p90 . [49] M c C a b e , Laur ie (Nov. 2000) "Death of an A S P P ioneer " ASPnews.com, (www.aspnews.com) . 109 [50] M c C a b e , Laur ie (Feb . 2001) " V S P s B loom in the A S P Microc l imate" ASPnews.com, (www.aspnews.com) . [51] M e a r s , Jenni fer (Feb. 2001) "Repor t s a y s 2001 to be turning point for A S P s , " Network World, Vo l . 18 Issue 6, p18. [52] Microsoft C o r p . (Feb. 2001) "A Y o u n g Pe rson ' s G u i d e to the S imp le Object A c c e s s Protoco l , " M S D N Onl ine . [53] M o c h , C h r i s s y (June 2000) "Instant Grat i f icat ion," Telephony, V o l . 238 Issue 2 3 , p294 -295 . [54] N e w c o m b , Kev in (Jan. 2001) " A S P Industry R a t e s We l l on the IBSi Su rvey " ASP News, (www. lnternetnews.com). [55] N e w c o m b , Kev in (Feb . 2001) "Wh ich A p p s are A S P s Se rv i ng? " ASP News, (www. lnternetnews.com). [56] N e w c o m b e , T o d d (July 2000) "The lease we c a n do," Gove rnmen t Techno logy , V o l . 13 Issue 9, p38 . [57] Phai r , Ma thew and R o e , And rew (Dec. 2000) "Third Party Prov iders F ind In-House S y s t e m s Still Ru le . " E N R . [58] Por tera S y s t e m s (2000) " W h e n D o e s Us ing a Hos ted Appl icat ion M a k e S e n s e ? " Portera Whitepaper. [59] Pr ing , B e n (May 2000) " A S P - Quicks i lver Rather than G o l d ? " ASPnews.com, (www.aspnews.com) . [60] Pr ing , B e n (Aug. 2000) "Instant A n s w e r s f rom the A S P T ip Tab le , " ASPnews.com, (www.aspnews.com) . [61] S C N Educat ion (Ed.) (2000) " A S P — a p p l i c a t i o n serv ice provid ing: the ult imate guide to hiring rather than buying appl icat ions," Academic Press. [62] S k e m e r , Dav id ; T a n , Patr ick (Sept. 2000) "How to m a n a g e o n - d e m a n d serv ices and technica l requi rements," Communications News, Vo l . 37 Issue 9, p76-77 . [63] S lav id , Peter ( Jan . 2001) "A L e s s o n f rom Serv ice Prov ider History," ASPnews.com, (www.aspnews.com) . [64] Sper l ing , E d (Jan . 2001) " S p e e d could be the enemy, " ZDNet eWeek (www.zdnet .com). [65] Taylor , Bart; Taylor , D a n (Jan 2001) "2001 : A n A S P O d y s s e y , " Communications News, p78-79 . [66] Tillett, Scot t (Mar. 2001) "Orac le P lans Integrated W e b A p p s , " Internetweek, Issue 8 5 1 , p14-17. 110 [67] Trager , Lou is (Dec 2000) "The Incredible Shr ink ing A S P Future," The Net Economy, Vo l 1 Number 4, p31 . [68] Trager , Lou is (Feb . 2001) " C o m m o n C a u s e , " The Net Economy, V o l . 2 Number 2 , p30-37. [69] Va i l , Mark (Jan . 2000) "The solut ion for IP serv ice quality," Telephony, V o l . 238 Issue 4, p54-59 . [70] Va lent ine, Br ian (Nov. 2000) "Execut ive C o l u m n , " MCSP Insider, Microsoft . [71] Venka t raman , N ; Loh , Lawrence (1992), "Determinants of Information Techno logy Outsourc ing : A Cross -sec t i ona l ana lys is , " Journal of Management Information Systems, vo l . 9, p7-24. [72] Wainewr ight , Phi l (March 2001) " E c o n o m i c Downturn a Boos t for A S P s " ASPnews.com, (www.aspnews.com) [73] Wainewr ight , Phi l (July 2000) " P a n d e s i c G i v e s up the St rugg le" ASPnews.com, (www.aspnews.com) [74] Wainewr ight , Phi l (Feb 2001) "Week l y A S P Industry Rev iew" ASPnews.com, (www.aspnews.com) [75] Wi l l i ams, O a k i e ( 1998 ) , " Outsourc ing : A C l O ' s Perspec t ive , " St. Lucie Press. [76] W i l l i amson , Ol iver (1975), "Marke ts and h ierarchies: Ana l ys i s and antitrust impl icat ions" T h e F ree P r e s s , N e w York . [77] W i l l i amson , Ol iver (1979), "Transact ion-cost economics : T h e gove rnance of construal relat ions" Journal of Law and Economics, V o l . 22 , p233-262 . [78] Wi l l is , Cl int (Feb . 2001) , " R a m p C h a m p s , " Forbes, Spec ia l A S P Issue, p94-102 . 111 A p p e n d i x A Lis t Of C o m p a n i e s Accen tu re : A provider of consul t ing serv ices and sys tems integration serv ices A d W a r e S y s t e m s : A Ful l V S P (Vertical Se rv i ce Provider) f o c u s e d on the adver t isement and market ing industry Agil i t i : A S P of type E S S (Enterpr ise Suppor t Serv ices ) and an appl icat ion aggregator App l icast Ar iba : A n E S S type A S P ISV (Independent Sof tware Vendor) f o c u s e d on providing supply cha in bus iness solut ions A T & T : Bluetrain: B reakaway : La rge te lecommunicat ions c o m p a n y A S P of type E S S and an appl icat ion aggregator an E S S type A S P C a p Gemin i Ernst & Y o u n g : A provider of managemen t consul t ing se rv i ces , s ys tems integration, and technology deve lopment Citrix: Concent r i c : Cor io : Da ta return: Digex: e B a y : E x o d u s : FutureLink: G o v H o s t . c o m : Grea t P la ins : IBM: Prov ider of software for porting exist ing sof tware to the Internet an E S S type A S P A n E S S type A S P . Prov ider of data centre serv ices Prov ider of data centre se rv ices , owned by W o r l d C o m Onl ine auct ion company . Recent ly a n n o u n c e d adopt ion of . N E T platform thus making it a F o c u s e d Solut ion A S P Network managemen t serv ice provider A n E S S type A S P Full V S P targeting governments ISV of bus iness software recently acqu i red by Microsof t Hardware manufacturer and ISV. 112 ICSA.ne t : Intacct: Intel: Interliant: International Compu te r Secur i ty Assoc ia t i on . R e s e a r c h e s and consul ts on securi ty i ssues related to comput ing N iche V S P providing account ing appl icat ion se rv i ces W o r l d s largest manufacturer of computer p rocesso r ch ips A n E S S type A S P International Da ta Corporat ion (IDC): A provider of consul t ing se rv i ces and IT industry research Interpath: J . D . Edwards : J a m c r a c k e r : K P M G : M ic roage : Microsoft : M I R U S : M y S A P . c o m : Ob l icore : Orac le : P a n d e s i c : Peop lesof t : Peop lesof t On l ine : Por tera : A n aggregator A S P of type E S S Consu l t ing company and ISV A S P of type E S S and an appl icat ion aggregator Consu l t ing and integration serv ices c o m p a n y A S P serv ice resel ler Large ISV that prov ides software for the persona l computer market. Recent ly re leased . N E T f ramework for the deve lopment of Internet native appl icat ions Full V S P targeting multi-unit restaurant cha ins Onl ine extension of S A P , E S S type A S P A S P Offer ing S L A monitoring serv ices La rge ISV of da tabase managemen t sof tware of the s a m e name. H a s A S P divis ion ca l led 'Orac le B u s i n e s s Onl ine ' , type E S S Joint venture between Intel and S A P . Prov ided e - C o m m e r c e serv ices , went out of bus iness in Ju ly of 2000 . N iche V S P type A S P La rge provider of E R P sys tems Onl ine extension of Peoplesof t , E S S type A S P Deve loper and A S P for subscr ipt ion b a s e d sof tware for pro fess ional se rv ices industr ies (consult ing f i rms, creat ive a g e n c i e s and other profess ional serv ices organizat ions) , N iche V S P type A S P Qwes t commun ica t ions : Large te lecommunicat ions c o m p a n y 113 Qwes t Cyber .So lu t ions : O w n e d by K P M G and Qwest . Qwes t Cyber .So lu t ions is an E S S A S P S A P : Largest provider of E R P sys tems Sprint: La rge te lecommunicat ions c o m p a n y and ISP S u n S y s t e m s : Manufacturer of S u n computers , and an ISV for U N I X b a s e d computer sys tems and competi tor to Microsoft and the persona l computer market USinternetwork ing (US i ) : A n ear ly entrant in the A S P market . Qua l i f ies a s a both a F o c u s e d Solut ion (in the a reas of f inance, account ing and supp ly chain) b e c a u s e of its use of . N E T , a s a N iche V S P and E S S U U N e t : Prov ider of data centre serv ices , owned by W o r l d C o m Viasof t : A S P serv ice resel ler W W W Consor t ium ( W 3 C ) : Deve lops interoperable techno log ies (speci f icat ions, gu idel ines, software, and tools) for the Internet W o r l d C o m : La rge te lecommunicat ions c o m p a n y and I S P 114 A p p e n d i x B Lis t Of Abbrev ia t ions U s e d (Brief descr ip t ions are given where necessary ) C L E C : Compet i t ive Loca l E x c h a n g e C o m p a n i e s ( common term in te lecommunica t ions industry appl ied to local te lephone compan ies that resul ted f rom the deregulat ion of the te lephone industry c o m m o n in the 1990s) C O M : C o m p o n e n t Object Mode l (programming mode l that Microsoft uses for development) C R M : C u s t o m e r Relat ionship Managemen t D B M S : D a t a B a s e Managemen t S y s t e m D C O M : Distr ibuted C O M (see C O M above , a distr ibuted vers ion of the componen t object model u s e d by Microsoft . D C O M is mainly used in distr ibuted, multi-tier appl icat ions and da tabases ) D H T M L : Dynamic H T M L (an extended vers ion of H T M L that a l lows for the use of the C P U of the cl ient computer . Us ing D H T M L , the client computer is u s e d for more than just presentat ion, which is the c a s e with H T M L ) E R P : Enterpr ise R e s o u r c e P lann ing E S P : Emerg ing Serv i ce Provider (term co ined by Eas te rn M a n a g e m e n t G r o u p and appl ied to Compet i t ive Loca l E x c h a n g e C o m p a n i e s ( C L E C s ) , providing mult iservice a c c e s s serv ices (data, vo ice and wi re less hardware instal lation and serc ive) and integrated providers such a s in-Bui lding serv ice providers) E S S : Enterpr ise Suppor t Se rv i ce H T M L : HyperText Mark -up L a n g u a g e H T T P : HyperText Transpor t Protocol 115 I E T F : Internet Eng ineer ing T a s k Force I P O : Initial Pub l ic Offer ing ISP : Internet Se rv i ce Prov ider ISV : Independent Sof tware Vendor IT : Information Techno logy M L P S : Mul t i -Label Protocol Swi tching (Method p roposed for ass ign ing priorities to da ta packe ts . M L P S would al low for var ious Q o S on the Internet) M R P : Mater ial R e s o u r c e P lann ing N P V : Net P resen t V a l u e O L A P : O n L i n e Analyt ica l P rocess i ng Q o S : Qual i ty of Se rv i ce R P C : R e m o t e P rocedu re Ca l l S L A : Se rv i ce Leve l Ag reemen t S O A P : S imp le Object A c c e s s Protocol (New connect iv i ty s tandard for se rv ices v ia the Internet, p roposed by Ar iba , Microsoft and I B M . ) T C : T ransac t iona l C o s t s T C O : Total C o s t of Ownersh ip UDDI : Un iversa l Descr ipt ion D iscovery and Interface (New connect iv i ty s tandard p roposed by A r iba , Microsoft and I B M . ) V A R : Va lue A d d e d Rese l le r V C : Venture Capi ta l is t 116 V S P : Ver t ica l Se rv i ce Prov ider W 3 C : W W W Consor t ium X M L : ex tens ib le Mark-up L a n g u a g e (a vers ion of H T M L that a l lows for the creat ion of da ta tags by the user . Of ten p roposed a s a rep lacement for e lectronic da ta interchange) 117 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0089933/manifest

Comment

Related Items