Beruflich Dokumente
Kultur Dokumente
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.
Management Information Systems Research Center, University of Minnesota is collaborating with JSTOR to
digitize, preserve and extend access to MIS Quarterly.
http://www.jstor.org
MISQuarterly/September1988 445
Defining Data Classes Data are grouped into categories called data
classes based on theirrelationshipsto the business
processes identifiedabove. Chartsare builtto re-
flectthose relationships.
446 MISQuarterly/September1988
To carry out SISP (especially in the alignment studies. Three popular methodologies include
mode),an organizationusuallyselects an existing Business Systems Planning(IBM,1975; Lederer
methodologyandthen embarkson a major,inten- and Putnam,1986), StrategicSystems Planning
sive study.The organizationformscommitteesof (HollandSystems, 1986), and InformationEngi-
users withISspecialists as membersor advisors. neering(Martin,1982). Theyare describedbriefly
Itmost likelyuses the SISP vendor'seducational methodologiesandwillbe alludedto
as illustrative
supportto trainthe committeemembers and the in the research findings.These three were se-
vendor's consulting support to guide the study lected because, together,they accountedforhalf
and auditits results.A multi-stepprocedureis car- of the responses to the survey.
riedout over several weeks or months.The dura-
tiongenerallydepends on the scope of the study. Business Systems Planning (BSP), developed
In additionto identifyingthe portfolioof applica- by IBM,involvestop-downplanningwithbottom-
tions, the organizationprioritizesthem. Itdefines up implementation.In this methodology,a firm
databases, data elements, and a networkof com- recognizes its business mission, objectives and
putersandcommunicationsequipmentto support functions,and how these determineits business
the applications.Italso prepares a schedule for processes. The processes are analyzedfortheir
developmentand installation. data needs, and data classes are then identified.
Databases are developed by combiningsimilar
data classes. The final BSP plan describes an
Frequentlyappliedmethodologies overallinformation system architectureas well as
Organizationsgenerallyapplyone of a numberof the installationschedule of individualsystems.
methodologies in order to performthese SISP Table 1 detailsthe steps inthe study.
Table 1. (continued)
MISQuarterly/September1988 447
BSP places heavy emphasis on top management by helping identifyfuture management control
commitmentand executive involvement.Top ex- systems. (Recentversionsof BSP also use CSF.)
ecutive sponsorshipis perceivedas critical.Infor-
mationsystems analysts mightserve primarilyin IEprovidesseveral softwarepackages for facili-
an advisorycapacity.The study producessuch a tating the SISP effort. However, IE differs from
other methodologies by providing automated
large volume of informationthat IBMhas begun tools to linkits outputto subsequent systems de-
marketingan automatedversion called Informa-
tionQualityAnalysis(Vacca, 1984). velopmentefforts.An applicationgeneratoris in-
tegrated with IE and produces systems with
Strategic Systems Planning (SSP), developed COBOLcode.
by Robert Holland,defines a business function
modelby analyzingmajorfunctionalareas. Adata
architectureis derivedfromthe business function Othermethodologies
model by combining informationrequirements Besides BSP, SSP and IE, firms mightchoose
intogeneric data entities and subjectdatabases. Method/1(Arthur Andersenand Co., 1982), infor-
An informationsystems architecturethen identi- mationQualityAnalysis(Vacca, 1984), Business
fies new systems andtheirimplementation sched- InformationAnalysis and IntegrationTechnique
ule. Althoughthe language differs slightly, the (Carlson,1979), Business Information
Character-
steps inthe SSP procedureare similarto those in ization Study (Kerner, 1979), CSF (Rockart,
BSP. 1979), Ends/Means Analysis (Wetherbe and
A majordifferencefromBSP is SSP's automated Davis, 1982), Nolan Norton Methodology
(Moskowitz,1986), PortfolioManagement(Mc-
storage, manipulation,and presentation of the Farlan,1981), StrategySet Transformation
(King,
data collected duringthe SISP process. Software
1978), ValueChainAnalysis(Porter,1985), orthe
produces reportsin a wide range of formatsand Customer Resource Life Cycle (Ives and Lear-
withvarious levels of detail. Forexample, "affin-
month,1984).
ity"reportsshow the frequencies of accesses to
data, while "clustering"reportsgive guidancefor firmsoften select features of these
Alternatively,
database design. Menus guide the user through methodologies and then, possibly with outside
online data collection and maintenance. A data consulting assistance, develop their own in-
dictionaryinterfacefacilitatessharing SSP data house approach(Arthur Andersenand Co, 1985;
with an existing data dictionaryor other auto- Sullivan,1987).
mateddesign tools.
Table 2 presents four majordistinctionsamong
Inadditionto SSP, HollandSystems Corporation some methodologies.Itclassifies them as align-
offersTacticalSystems Planning(TSP)and Logi- mentor impactapproaches.The table also distin-
cal Database Design (LDD).TSP is a methodol- guishes them by their primaryfocus. Finally,it
ogy for guiding the implementation of the shows whetherthey define a data architecture,
informationsystem architecture.LDDis used to and whetherthey provideautomatedsupport.
develop data structures for modules from the
study or fromother systems, and then is used to
mapthe structuresto the SSP data architecture. Problemswiththemethodologies
Information Engineering (IE), developed by Ithas long been recognizedthat SISP is an intri-
James Martin,provides techniques for building cate and complex activityfraughtwith problems
enterprise models, data models, and process (McFarlan, 1971). Several authors have de-
models.These forma comprehensiveknowledge scribed these problems.Theirworkis based on
base whichthen creates and maintainsinforma- fieldsurveys,cases, and conceptualstudies, and
tion systems. IE is considered by some to be a investigatesmostof the methodologiesdescribed
more technically oriented approach than other previously.A reviewof the most significantof their
SISP methodologies. articlesserved as the basis to create a compre-
hensive listof the problems(see Table3).
InconjunctionwithIE,Martinadvocates the use of
CriticalSuccess Factors(CSF)(Rockart,1979), a Inorderto organizeandsummarizethe problems,
technique for identifyingissues considered by this research used three categories-resources,
business executives as the most vitalforthe suc- process, and output.Resource-relatedproblems
cess of their organization.Martinsuggests that addressed the issues of time-requirements,
each general manager should use CSF. The re- money, personnel,and top managementsupport
sultingfactorswillthen guide the SISP endeavor for the initiationof the study. Process-related
448 MISQuarterly/September1988
MISQuarterly/September1988 449
450 MISQuarterly/September1988
Table 3. (continued)
Outputof the Planning Methodology
01 SISP outputfailsto providea statementof organizational McLeanand Soden,
objectivesforthe IS department. 1977
02 SISP outputfails to designate specificnew steeringcommittees.
03 SISP outputfailsto identifyspecificnew projects. McLeanand Soden,
1977
04 SISP outputfails to determinea uniformbasis forprioritizing King,1978
projects.
05 SISP outputfailsto determinean overalldata architecture Zachman,1982
forthe organization.
06 SISP outputfailsto provideprioritiesfordevelopingspecific Zachman,1982
databases.
07 SISP outputfails to sufficientlyaddress the need forData Sullivan,1985
Administration inthe organization.
08 SISP outputfailsto includean overallorganizationalhardware McLeanand Soden,
plan. 1977
09 SISP outputfails to includean overallorganizationaldata Sullivan,1985
communicationsplan.
010 SISP outputfails to outlinechanges inthe reportingrelationships
inthe IS department.
011 SISP outputfails to includean overallpersonneland training McLeanand Soden,
planforthe IS department. 1977
012 SISP outputfails to includean overallfinancialplanforthe McLeanand Soden,
IS department. 1977
013 SISP outputfailsto sufficientlyaddress the roleof a King,1984
permanentIS planninggroup.
014 The outputplans are notflexibleenough to take intoaccount McLeanand Soden,
unanticipatedchanges inthe organizationand its environment. 1977
015 The outputis not in accordancewiththe expectationsof top Gill,1981
management.
016 Implementingthe projectsand the data architectureidentified Zachman,1982
inthe SISP outputrequiressubstantialfurtheranalysis.
017 Itis difficultto secure top managementcommitmentfor Gill,1981
implementingthe plan.
018 The experiences fromimplementingthe methodologyare not Zachman,1982
sufficientlytransferableacross divisions.
019 The finaloutputdocumentis notveryuseful. King,1984
020 The SISP outputdoes notcaptureallthe information that Gill,1981
was developed duringthe study.
MISQuarterly/September1988 451
452 MISQuarterly/September1988
454 MISQuarterly/September1988
Table 6. OverallSatisfaction
Average Satisfied Neutral Dissatisfied
The Methodology 3.55 54% 23% 23%
The Resources Required 3.02 38% 24% 38%
The Methodology'sProcess 3.68 48% 17% 25%
The Methodology'sOutput 3.38 55% 17% 28%
CarryingOutthe Plan 2.53 32% 15% 53%
At the same time, organizations had begun ning was financial/tacticalratherthan strategic.
projectswhich were not partof their SISP plan. Analysisof variancetested the differencein the
These constituted about 38% of all projects ini- mean scores underalternativesfor each factor.
tiated after the study. Finally, in organizations Thus,Table7 shows the levels of statisticalsignif-
where the SISP had recommended changes in icance forthe alternativewithmore severe prob-
the IS department, only 50% of these changes lems. (The relativelylow mean scores reflectthat
had been carriedout. some of the problems are considerably less
severe thanothers.)
These data suggest thatthe respondentsdid not
execute theirfinalplans very scrupulously.They The followingsubsections discuss the problem
raise questions about the resemblance between factors.The headings reflectthe study'sfindings
the systems envisionedbythe planninggroupand based strictlyon the alternativewith the lower
theirfinalimplementations.One mightalso spec- mean score. The subsequent discussion further
ulatethatthe methodologieshave failedto gener- considersthe strengthof the findings;cautiousin-
ate useful ideas which organizationscould then terpretationof the non-significantdifferences is
translateinto implementablecomputersystems. suggested.
Giventheirgreat expense and timeconsumption,
such findingsseriouslychallenge the utilityof the Factor 1: Organizations with less sophistica-
planning methodologies represented in this tion in business planning had more
study. severe problems than more sophis-
ticated organizations.
However,the findingsare not necessarilysurpris-
ing. They confirmthe workof Runge (1985) who Organizationsthat characterizedtheir business
studiedsuccessfully implementedstrategicinfor- planningas financialor tacticalhad significantly
mation systems. In 80% of his cases, existing more severe problems than organizations that
SISP procedures were either purposelycircum- characterizedtheirbusiness planningas strate-
vented or ignored.Runge attributedthe success- gic. The effectwas significantforratingsinallfour
ful implementationof these systems, not to SISP categories. It is not surprisingthat a general so-
methodologies, but largely to "productchampi- phisticationinsettinggoals andobjectivesperme-
ons," i.e., top general business executives who ates the SISP activities. In such sophisticated
secured the necessary resources, overcame re- organizations, IS executives have less trouble
sistance to approval and development, and justifyingresources,carryingoutthe process, and
actively promoted the systems during analyzingthe output.
implementation.
Factor 2: Organizations with less participa-
tion by the IS department in busi-
Potentialcausal factorsaffecting ness planning had more severe
extentof problems problems than organizations with
greater participation.
Table7 identifiesthe previouslydiscussed organi-
zational and managerial factors potentiallyre- Althoughthe differenceswere notstatisticallysig-
lated to the severity of the SISP problems. It nificantin any category,this effect held in allfour.
shows the mean ratingsof the problemsfromthe The directionof the effect suggests the impor-
resources, process, and outputcategoriesandfor tance of such participation.Participationin strat-
all 49 (i.e., overall) problems. For example, the egy formulationenables the IS department to
2.38 in row 1, column 1 refers to the average betterunderstandtop management'sobjectives
severityof the 15 resource problems(inTable3) and thus,to ensure thatthe SISP outputssupport
for subjects who stated that theirbusiness plan- theirgoals.
MISQuarterly/September1988 455
Factor 3: Organizations where the top IS ex- of the top informationexecutives surveyed re-
ecutive reported to a controller had portedto a senior-levelfinancialofficer.
more severe problems than organi-
zations where the top IS executive Factor 4: Organizations where top manage-
reported to a president or vice ment initiated the study had more
president. severe problems than organiza-
tions where IS management
This effect was statisticallysignificantfor all four initiated it.
categories. Itparallelsthe findingsof Benjamin,et
al. (1985) who observedthatchiefinformation offi- Thisfindingwas surprising.Althoughdifferences
cers in leading-edgecompanies frequentlyreport in ratingswere not statisticallysignificant,the ef-
to an area otherthanfinance. Morehighlyplaced fect was uniformacross all fourcategories. The
IS executives can more easily initiate,carryout, findingsuggests thatalthoughISexecutives seek
and analyze SISP exercises. Also,the primarily fi- top managementinvolvement,they stillpreferto
nancial orientationof the controllermay not en- maintaincontrol.Topmanagement-initiated SISP
hance the IS Department'spositionor contribute studies maylikelybe the resultof displeasurewith
to its SISP skills.The reportingrelationshipmight the performanceof IS management.IS manage-
merelyreflectmore archaicorganizations. ment-initiatedSISP studies probablypermitIS
managementto exercise moreinfluenceover the
importantbecause top IS
Thisresultis particularly SISP study.However,the findingmightsimplybe
executives frequentlyreportto a controller.Arthur to the fact thatthe respondentswere
attributable
Andersenand Co. (1986) recentlyfoundthat32% IS executives.
456 MISQuarterly/September1988
Factor 5: SISP studies with a division or func- et al., 1983; Zachman,1982), BSP's top problem
tion as their scope had more severe is thatits documentationdoes not adequatelyde-
problems than studies with the en- scribethe steps to follow.The top problemof SSP
tire enterprise as their scope. and IE is the difficultyof obtainingtop manage-
mentcommitmentforimplementingthe plan,per-
Again,the finding,althoughveryweak, is surpris- haps because these methodologies (and their
ing. Studies of divisionsand functionshad signifi- vendors)are less well-knownto top management
cantly more severe resource problems than than IBM.Itmay also be due to theirmore recent
studies of entireenterprises.The same effect, al-
origin.The majorproblemof in-house-developed
thoughnotstatisticallysignificant,was trueforthe methodologiesis theirlackof sufficientcomputer
process and overall categories. For the output support;this is notsurprisingwhen one considers
category,the ratingswere nearlyequal (although the expense of developingsuch supportand the
inthe opposite directionas the others).The impli- likelihoodthata firmwoulddo so.
cationof this findingis thatthe broadand general
recommendations of the methodologies might Table 8 also shows some other potentialdiffer-
simplybe bettersuited to the definitionof data ar- ences amongthe methodologies.Lackof a train-
chitecturesof broaderscopes.
ing plan and the lengthydurationof the planning
Factor 6: When the SISP study failed to spec- exercise are two problemsin SSP's and IE'stop
ten thatare notinBSP'stopten. Inadequatedocu-
ify a planning horizon, problems
were more severe than when it did mentation,lackof computersupport,and depen-
dence on a team leader are three problems in
specify a planning horizon.
BSP's top ten that are not in SSP's and IE'stop
The effect of thisfindingwas consistentacross all ten.
four categories and was significant for the re-
sources and overallcategories. Its implicationis Amongthe top ten problemsof the fourmethod-
fairlystraightforward- a planninghorizonis a ologies, three are common to all four.These in-
controlmechanism. Itdemands that a schedule clude the difficultyin obtainingtop management
be drawnup and followed.Itforces planningpar- commitmentforimplementingthe outputs,the re-
ticipantsto confrontand resolveproblemsinorder quirementforsubstantialfurtheranalysis,and the
to meet their milestones. This findingsuggests difficultyfindinga good team leader. Infact, most
thatthe importanceof a planninghorizonhas not of the top ten problemsof each methodologyare
diminishedeven though increasingenvironmen- relatedto carryingout the plan and the planning
tal volatilityhas made its use more difficult(Led- team;this findingaccents the underlyingsimilari-
ererand Mendelow,1986c; Sullivan,1987). ties amongthe methodologies.
Factor 7: Publicly-owned organizations had These similarities and differences might offer
more severe problems than pri- some preliminaryguidance to firmsselecting or
vately-owned organizations. developingan SISP methodology.However,due
to smallsample sizes (17, 11, 12, and 11, respec-
Althoughthe effect was consistentforallfourcat- tively,forthe fourmethodologiesinTable8), cau-
egories, none of the differenceswere significant. of thistable is suggested.
tious interpretation
Still,the implicationmightbe that publicly-owned
firmsare generally more bureaucraticand more
subject to externalpressures than privately-held
organizations, possibly because it is easier to Summaryand Conclusion
controlplanningin a privately-heldfirm.Thus, IS ImprovedSISP is a majorchallengefacing IS ex-
departmentsin publiccompanieswouldfindmore ecutives today. Effectiveplanningis essential to
difficultyin obtaining resources, executing the the realizationof the potentialstrategicimpactof
process, and analyzingthe output. computer-basedinformation systems. Thisarticle
has examined the difficultiesof implementinga
methodologyto performSISP.
Theproblemsof specific SISP
The results suggest that IS planners are not
methodologies particularlysatisfied with their methodologies.
The top ten problemsof the fourmost frequently Planningrequirestoo manyresources.Top man-
used SISP methodologiesappearinTable8. De- agement commitment is not easily obtained.
spite the common belief that one of BSP's major Whenthe SISP exercise is complete,furtheranal-
strengthsis its detaileddocumentation(Bowman, ysis is requiredbefore the execution of the plan
MISQuarterly/September1988 457
can take place. Consequently, carryingout the Therefore,ifthe objectiveof the SISP exercise is
planis often not very extensive. to alignISobjectiveswithbusiness goals (as is the
primaryobjective of most of the methodologies
The final plan might be a good plan. However, used by participantsin this study),then detailed,
management commitmentto the plan might be lengthy and complex SISP may be of limited
missing or the means of controllingits execution value. Alternatively,
the objectiveof an SISP ex-
mightbe ineffective. ercise can be to use informationtechnology to
458 MISQuarterly/September1988
MISQuarterly/September1988 459
460 MISQuarterly/September1988
MISQuarterly/September1988 461