Professional Documents
Culture Documents
M.E. Sharpe, Inc. is collaborating with JSTOR to digitize, preserve and extend access to Journal of
Management Information Systems.
http://www.jstor.org
TheDeLoneand McLeanModelof
Information
SystemsSuccess:
A Ten-YearUpdate
WILLIAM H. DeLONE AND EPHRAIM R. McLEAN
ofInformation
William DeLone is anAssociateProfessor
SystemsandChairofthe
at theKogodSchool of BusinessatAmerican
Information
Department
Technology
areasofresearch
include
DeLone's primary
inWashington,
DC. Professor
University
of information
and value,theimplementation
theassessment
systemseffectiveness
in smalland medium-sized
and use of information
businesses,and the
technology
He has been publishedin various
of information
technology.
global management
JourDATABASE,
Research,MIS Quarterly,
Systems
Information
journalsincluding
and Journalof Information
nal of Global Information
Technology
Management,
fromVillanovaUniProfessor
DeLone earneda B.S. in mathematics
Management.
anda
fromCarnegieMellonUniversity,
administration
an M.S. in industrial
versity,
ofCalifornia,
Los
andInformation
Ph.D. inComputers
SystemsfromtheUniversity
Angeles.
Scholar's
andGeorgeE. SmithEminent
EphraimR. McLean is a Regents'Professor
of
Business
at
in
the
Robinson
Chairin Information
GeorgiaState
College
Systems
in
he was on
State
to
to
Prior
Atlanta.
1987,
University
coming Georgia
University,
Los Angeles(UCLA) for18 years.Dr.
of California,
thefacultyof theUniversity
ofinformation
focuseson themanagement
McLean'sresearch
services,thevalueof
He has publishedover 125
and careerissuesforIS professionals.
IS investments,
Research,JournalofManagement
Systems
papersin suchjournalsas Information
MIS Quarterly,
Science,Communications
ofthe
Management
Systems,
Information
andothers;
SloanManagement
HarvardBusinessReview,
Review,
ACM,DATABASE,
nowin itsthird
andhiscoauthored
book,Information
Technology
forManagement,
in theworld.Dr. McLean
thesecondlargestsellingIS textbook
edition,is currently
andhis S.M.
fromCornellUniversity
earnedhisB.M.E. in mechanicalengineering
InstiattheMassachusetts
andPh.D. degreesfromtheSloanSchoolofManagement
for
the
Association
Director
of
Executive
is
also
the
He
tuteofTechnology
(MIT).
the
of
AIS.
in
was
a
Fellow
made
and
1999
Information
Systems(AIS)
theDeLone andMcLean Information
Abstract: Tenyearsago,we presented
Systhecomplexand modelformeasuring
tems(IS) Success Model as a framework
IS
In thispaper,we discussmanyoftheimportant
variableinIS research.
dependent
of thelast decade,focusingespeciallyon research
successresearchcontributions
toouroriginal
model.
andproposeenhancements
efforts
thatapply,validate,
challenge,
tothe
we proposeminorrefinements
ofthosecontributions,
Based on ourevaluation
modelandproposean updatedDeLone andMcLean IS SuccessModel.We discuss
e-commerce
oftheupdatedmodelformeasuring
theutility
systemsuccess.Finally,
of
measurement
current
andfuture
we makea seriesofrecommendations
regarding
IS success.
/ Spring2003,Vol. 19,No. 4, pp.9-30.
JournalofManagement
Systems
Information
2003 M.E. Sharpe,Inc.
0742-1222/2003 $9.50 + 0.00.
10
ofinformation
Key words and phrases: evaluation
impactofinformation
systems,
information
systemssuccess,servicequality,sysquality,information
technology,
usersatisfaction.
use ofinformation
temsquality,
systems,
11
of success
thesesix dimensions
Based on bothprocessandcausal considerations,
thanindependent.
Thishas important
rather
areproposedtobe interrelated
implicain
of
success
and
IS
tionsforthemeasurement,
empiricalstudies.
analysis, reporting
first
various
an
is
that
IS
model
A temporal,
created,containing
suggests
process
variousdegreesof systemand
as exhibiting
whichcan be characterized
features,
thesefeatures
users
and
information
Next,
byusingthe
experience
managers
quality.
or
its
information
the
or
with
dissatisfied
either
satisfied
and
are
prodsystem
system
thenimpactsor influences
ucts.The use ofthesystemanditsinformation
products
theindividualuserin theconductof his or herwork,and theseindividualimpacts
Theresultant
D&M IS SuccessModelis
resultinorganizational
impacts.
collectively
1
in
reproduced Figure [8, p. 87].
of
toa processmodel,a causal orvariancemodelstudiesthecovariance
Incontrast
ifthereexistsa causalrelationship
todetermine
thesuccessdimensions
amongthem.
and
user
satisfaction
to
to
lead
is
Forexample,highersystem
higher
quality expected
in
on
individual
to
resulting organizational
use,leading positiveimpacts
productivity,
with
thesuccesstaxonomy
The purposeof combining
improvements.
productivity
ofthepossiblecausalinterrelationthesuccessmodelwastoaidintheunderstanding
to
and
of
success
the
dimensions
exposiprovidea moreparsimonious
shipsamong
has
this
combination
critics
for
some
the
of
tion
proved
Unhappily,
relationships.
Thesewillbe disofreformulations.
troublesome,
leadingthemto suggesta number
cussedlaterinthispaper.
conclusionsoftheoriginalpaperwere:
The primary
natureofIS successrequirescareandinterdependent
1. The multidimensional
ofeach aspectofthisdepenandmeasurement
to thedefinition
fulattention
to measurethepossibleinteractions
dentvariable.It is important
amongthe
variof
various
effect
the
isolate
order
to
in
dimensions
success
independent
dimensions.
success
ableswithone ormoreofthesedependent
on the
and measuresshouldbe contingent
2. Selectionof successdimensions
tested
where
the
of
context
and
but,
possible,
investigation;
empirical
objectives
be
used.
should
measures
andproven
natureofIS success,an attempt
andcontingent
3. Despitethemultidimensional
measuresused
ofdifferent
thenumber
shouldbe madetoreducesignificantly
to measureIS successso thatresearchresultscan be comparedandfindings
validated.
and incorporate
4. Morefieldstudyresearchshouldinvestigate
organizational
measures.
impact
and valida5. Finally,"[t]hissuccessmodelclearlyneedsfurther
development
IS meaof
selection
for
the
a
basis
as
serve
it
could
tionbefore
appropriate
sures"[8, p. 88].
ModelAdoption
Research adoption of the D&M IS Success Model has exceededourexpectaof2002yielded285 refereed
searchinthesummer
tions.A citation
papersinjournals
12
System
Quality
' i
j |'
j
^
Information
Quality
Use
'
/
j 1/
j F
II
II
II
>x
User
j k
Jj'
i
:
*
Organizational
1 Individual
J
Impact
ft Impact
' 1/
Satisfaction j J
and proceedings
thathavereferenced
theD&M Model duringtheperiod1993 to
mid-2002.Manyofthesearticlespositioned
themeasurement
orthedevelopment
of
theirdependent
thecontext
oftheD&M IS Successframework.
variable(s)within
By
forreporting
andcomparing
researchwork
usingthemodelas a commonframework
IS successoreffectiveness,
we believeone oftheprimary
involving
purposesofthe
originalarticlehas beenachieved.
Although
manyofthecitedarticlestendedtojustifytheirempiricalmeasurement
ofIS successbycitingtheD&M IS SuccessModel,someofthemfailedtoheedour
cautions.Someresearchers
haveusedthemodeltosupport
theirchosensuccessvariable ratherthanto inform
thedevelopment
of a morecomprehensive
successcon- thatIS success is a
struct.Theyoverlookedthemainconclusionof thearticle
- andthatitis therefore
multidimensional
andinterdependent
construct
to
necessary
the
or
"Researchers
study interrelationships
among, tocontrolfor,thosedimensions.
shouldsystematically
combineindividual
measuresfromtheIS successcategories
to
createa comprehensive
measurement
instrument"
[8, pp. 87-88]. Althoughthese
authors
didnotchooseto measure(orcontrolfor)thevariousdimensions
ofIS suca
number
of
other
researchers
haveusedmultidimensional
cess,
measuresofIS success intheirempiricalstudiesandhaveanalyzedtheinterrelationships
amongthem.
Someofthesestudiesaresummarized
in thenextsection.
ModelValidation
Unlike a process model, whichmerelystatesthatB followsA, a causal model
thatA causesB; thatis,increasing
A willcauseB toincrease(ordecrease).
postulates
13
inour
In the1992articlewe proposedsuchinterrelationships
amongthedimensions
Since 1992,a numberof studieshave
model;butwe did nottestthemempirically.
ofthemultidimensional
undertaken
investigations
relationships
empirical
amongthe
measuresofIS success.
Empiricaltestingand validationoftheD&M IS Success Model was theprimary
purposeoftworesearchstudies[38,41]. SeddonandKiew [41] surveyed104users
relaofa recently
university
accounting
systemandfoundsignificant
implemented,
with
"user
satisfaction"
and
"individual
between
impact,"
"systemquality"
tionships
and"individual
and
between"information
impact,"
quality"with"usersatisfaction"
a
and "individualimpact."Rai et al. [38] performed
between"usersatisfaction"
Model
based
on
the
D&M
IS
Success
test
on
entire
survey
responses
goodness-of-fit
student
IS. The studyfoundthatsomegoodness-of-fit
from274 usersofa university
butotherswerenot.However,all ofthepathcoefficients
indicators
weresignificant
oftheD&M IS SuccessModelwerefoundtobe signifiamongsuccessdimensions
cant.
testedtheassociationsamongthemeasuresidenstudiesexplicitly
Otherempirical
tifiedin theD&M IS SuccessModel [10, 12, 14, 16, 22, 50]. Yet otherempirical
successdimensions
testedthemodelbyinvestigating
studieshaveimplicitly
multiple
and theirinterrelationships
54,
57,
58,
17,
48,
52,
60].
[11,
Figure2 displaysthe
inthe16
not
confirmed
and
the
Model
IS
Success
D&M
(or confirmed)
relationships
studiescitedabove.Thesestudieswereselectedbasedon thefactthatthey
empirical
andtheymeasuredtheassociationamong
successconstructs
usedmultidimensional
theseempiricalresultsaresumIn thefollowing
thesuccessconstructs.
paragraphs,
withthestrongest
Links
were
tested.
links
that
success
marizedby the
empirical
arediscussedfirst.
support
SystemUse- IndividualImpacts
Sevenofthe16 studiesin Figure2 [12, 14, 16,48, 52, 54, 60] testedtheassociation
between"systemuse" and "individual
impacts"andtheassociationwas foundto be
andwasmeasured
usewastypically
the
studies.
in
of
each
voluntary
System
significant
anddependency.
ofaccesses,usagepattern,
ofuse,timeofuse,number
as frequency
anddecision-making
Individual
impactsweremeasuredin termsofjob performance
performance.
- IndividualImpacts
SystemQuality
All fivestudies[10, 12,41, 50,57] thattestedthedirectassociationbetween"system
signifiimpacts"foundthoseassociationstobe statistically
quality"and"individual
cant.Systemqualitywas measuredin termsofease-of-use,
reliability,
functionality,
Individualimpacts
and importance.
dataquality,portability,
integration,
flexibility,
andjob performance.
weremeasuredas qualityofworkenvironment
14
ii
r^
O
T3
S
^
o
X l
J ^ c
II
C
C
si1!
i
' '
NO-.
If
/ /
/ /
s| S|a
/ /
/ /
g2 ^
'
g S "
Mwfvh'
/ i ^ >^
V'
L -J
Is
--
fe
|
-S
co
I l
a) /-^
^ 1^
^r
"3 rT
h ^ v L_l '
/ LjLT || ||
JL
Se ?
S?
s ^Q
'
Sil
!" fSSS
J_
5
111
H
'
' I I
SA"
ttC
1 /
!*
la
.S
li
i Sa!
'i -S-
c ^ 31
o S - o
o
'33 HM
s
3
c
D -o 2 - :
fi C 5
c
.S <tf13 ^
c U
n -8 -
SP^ g ^
k, C P C
15
- IndividualImpacts
Information
Quality
between"information
The fourstudies[10, 41, 50, 57] thattestedtherelationship
Information
impacts"foundtheassociationtobe significant.
quality"and"individual
and
in
terms
of
measured
was
relevance,
timeliness,
accuracy,
completeness,
quality
in
terms
was
measured
Individual
decision-making
performance,
impact
consistency.
andqualityofwork.
job effectiveness,
OtherLinks
all theotherlinksorassociationsintheD&M IS SuccessModel
Withoneexception,
validated.The one empiricalstudythatfoundtheassociationsnot
wereempirically
was a surveyofDutchmanagers[11],wheretheassociationbetweensyssignificant
was notstatistically
revenuesandprofitability
temuse andorganizational
significant.
tested
36 ofthe38 successfactorassociationsthatwereempirically
In conclusion,
Taken
as a
found
to
be
2
were
in
summarized
in the 16 studies
significant.
Figure
fortheproposedassociations
studiesgivestrong
among
whole,theseempirical
support
in themodel.
thecausal structure
andhelpto confirm
theIS successdimensions
ModelIssues
In addition to the manypapersthathavetestedandvalidatedtheD&M IS Sucor extend
cess Model,severalarticleshavebeenpublishedthatchallenge,critique,
toa better
themodelitself.On balance,thesearticleshavecontributed
understanding
ThesearticlesandtheissuestheyraisetotheD&M IS
ofsuccessanditsdimensions.
below.
SuccessModelaresummarized
ProcessVersusCausal Models
arebasedon a process
anditssix successcategories
TheD&M IS successtaxonomy
resultareinterrelated,
we arguethatthesixdimensions
modelofIS [43].In addition,
direction
as the
in
the
same
flows
that
ingina successmodelthatindicates causality
information
process.However,citingan earlierpaperbyNewmanandRobey[35],
Seddonarguesthat"theboxesandarrowsin variance-andprocess-model
diagrams
in one
be
combined
cannot
and
different
meaningfully
concepts
quite
represent
is
models
and
variance process
model.. . . Unfortunately,
exactlywhat
combining
todo" [40].Seddonfurther
arguesthatDeLone
[DeLoneandMcLeanhave]attempted
ofIS
tocombinebothprocessandcausalexplanations
andMcLeanhave"attempted
withthismodelforsomeyears,ithas become
successintheirmodel.Afterworking
intheirmodel
thattheinclusionofbothvarianceandprocessinterpretations
apparent
leads to so manypotentially
meanings"[40, p. 240]. Seddongoes on to
confusing
of
IS success.
model
variance
a
propose respecified
16
of processand variance
We agreewithSeddon's premisethatthecombination
of IS successin one modelcan be confusing.
However,we believe
interpretations
of theD&M Model intotwopartialvariancemodels
thatSeddon'sreformulation
oftheorigitheintent
thesuccessmodel,defeating
[40,p. 245] undulycomplicates
nal model.
Thecreation
oftheD&M IS SuccessModelwas drivenbya processunderstanding
of
thecreation
Thisprocessmodelhasjustthreecomponents:
ofIS andtheirimpacts.
a system,
theuse of thesystem,and theconsequencesof thissystemuse. Each of
fortheresultant
condition
butnotsufficient,
thesestepsis a necessary,
outcome(s).
orbenefits.
Forinstance,
without
However,
use,therecanbe noconsequences
system
there
or ill-informed,
withsystemuse, evenextensiveuse, whichis inappropriate
a
of
IS
the
dimensions
to
understand
be
no
benefits.
also
success,
Thus,
fully
may
the
out
himself
as
Seddon
variancemodelis also needed.Thus,
pointed [40, 42],
variance
applicationof ourmodelto empiricalresearchalso requiresa contextual
first
is
the
three
are
Here
too
there
the
model.
of
produccomponents:
specification
whether
is
The
is
net
benefits.
the
third
is
and
the
second
use,
tion,
onlyargument
intoonemodel.AlongwithSeddon,
canbe combined
dimensions
thesetwonecessary
aredifferent.
we believethattheycan; onlyourformulations
SystemUse as a SuccessMeasure
Seddon[40] further
arguesfortheremovalof "systemuse" as a successvariablein
forinclusionin
thecausalsuccessmodel,claimingthatuseis a behavior,
appropriate
a processmodelbutnotin a causalmodel.He arguesthatuse mustprecedeimpacts
andbenefits,
butitdoesnotcausethem.Wedisagree.Webelievethatsystem
usageis
an appropriate
measureofsuccessin mostcases.
ofthiscomplexvariable.
definition
The problemto datehas beena too simplistic
thenature
will
without
that
more
use
more
benefits,
considering
yield
Simplysaying
ofthisuse,is clearlyinsufficient.
mustalso considerthenature,
Researchers
extent,
could
be
of
use
of
use.
The
nature
system
quality,and appropriateness thesystem
ofa systemis beingusedfor
addressedbydetermining
whether
thefullfunctionality
theintended
purposes.
YoungandBenamati[59],forexample,suggestthatfullfunctionaluse ofan e-commerce
use,transactional
systemshouldincludeinformational
andBrancheau
and
customer
service
use.
With
to
the
extent
of
Lassila
use,
use,
regard
variousstatesofsystems
utilization
basedon theuse ornonuseofbasic
[27] identify
andadvancedsystemcapabilities.Simplymeasuring
theamountoftimea systemis
of
useddoes notproperly
the
between
usageandtherealization
capture relationship
be
an
results.
On
the
other
it
can
be
that
hand,
expected
argued decliningusagemay
indication
thattheanticipated
benefits
arenotbeingrealized.
important
The rejection
ofsystemuse as a successvariablewhensystemusageis mandatory
is also flawedforthereasonscitedabove.Even whenuse is required,
in
variability
on
the
realthequalityandintensity
ofthisuse is likelyto havea significant
impact
At
izationof thesystembenefits.
no systemuse is totallymandatory.
Furthermore,
someleveloftheorganization,
anexecutive
committee
haschosento
ormanagement
17
a system
andrequireemployees
touse it.Thus,whereasusageofa system
implement
one
the
be
at
continued
level,
may mandatory
adoptionanduse of thesystemitself
based
ata higherlevel.Manbe
may whollyvoluntary, uponmanagement
judgment,
the
of
a
that
the
option discontinuingsystem is notproviding
agementalwayshas
desiredresultsandbenefits.
tobe usedas a dependent
variableina number
ofempirical
Systemusagecontinues
studiesandcontinues
tobe developedandtestedbyIS researchers
[11, 12,14,16, 17,
in e-commerce
38, 47, 48, 52, 54, 60]. Systemuse has takenon new importance
andessentialto desiredoutwherecustomer
use is voluntary
successmeasurements
comes[7, 29, 36]. "WhilemoststudiesthatfollowD&M replacetheUse box with
to maintainUse as in theoriginalwork.In e-commerce
. . . , we prefer
Usefulness
and
[33,p. 6]. We agreewiththeseIS researchers
systemsUse is largelyvoluntary"
and effective
believethatuse,especiallyinformed
use,willcontinuetobe an imporofIS successformanysystems.
tantindication
Role ofContext
ofapplyingtheD&M IS Sucon thedifficulty
havecommented
Severalresearchers
IS successin specificresearchconcess Modelinordertodefineandoperationalize
"Thissuccessmodelclearlyneedsfurther
Thiswasnotunexpected:
texts.
development
IS
serveas a basis fortheselectionof appropriate
it
could
and validationbefore
success
measures"[8, p. 88]. Jiangand Klein [20] foundthatuserspreferdifferent
that
et
found
al.
of
evaluated.
on
the
measures,
Whyte
system
being
type
depending
variaand
from
differences
"thereareimportant
user,
systems
deriving organizational,
theviewas towhichattributes
tionswhichcanmodify
(successmeasures)areimpora
contribution
tant"[55,p. 65]. Seddonet al. [42] makean important
byproposing
measuresbasedon thetype
IS effectiveness
matrixforclassifying
two-dimensional
theIS is beingevaluated.
in whoseinterest
ofsystemstudiedandon thestakeholder
in
the
1992
As
stated
In thisregard,
we completely
article,"no singlevariable
agree.
so thechoiceofsuccessvariablesis oftena functhananother,
better
is intrinsically
. . . etc."[8,p. 80,emphacontext
theorganizational
tionoftheobjectiveofthestudy,
sis added].
Variables
VersusDependent
Independent
to theD&M IS Success Model flowfroma
Manyof thesuggestedimprovements
variableandwhatis partofthedependent
betweenwhatis an independent
confusion
arebuttwo
and
involvement"
"User
success.
variable,IS
support"
"topmanagement
examplesof suggestedadditionsto theD&M Model; yettheseareclearlyvariables
in ERP" may
thanbeinga partofsuccess."Investing
thatmaycause successrather
of
IS
butthe
"information
to
lead
success),
(an
quality" aspect
(ormaynot)
improved
variable.
of
the
is
latter
whereas
the
variable
is an independent
former
dependent
part
controlvaribetweenthemanagement
It is essentialthatIS researchers
distinguish
andimpacts.
use satisfaction,
ablesandthedesiredresultsintermsofquality,
18
ModelExtensions
ServiceQuality
The emergenceof end user computinginthemid-1980splacedIS organizations
inthedualroleofinformation
an information
provider(producing
product)andserviceprovider(providing
supportforend userdevelopers).Pittet al. observedthat
thanthe
rather
used measuresofIS effectiveness
focuson theproducts
"commonly
will
mismeasure
servicesoftheIS function.
there
is
a
that
IS
researchers
Thus,
danger
IS effectiveness
iftheydo notincludein theirassessment
packagea measureof IS
servicequality"[37,p. 173].Otherresearchers
haveagreedwiththis,citingtheneed
fora servicequalitymeasuretobe a partofIS success[25,28, 56].
Researchers
whohavearguedthatservicequalitybe addedto thesuccessmodel
frommarhaveappliedandtestedthe22-itemSERVQUAL measurement
instrument
of tangibles,
uses thedimensions
keting[25, 37] to an IS context.This instrument
tomeasureservicequality.Some
assurance,andempathy
reliability,
responsiveness,
instrument
items
include:
sampleSERVQUAL
19
NetBenefits
have
As the"impacts"of IS haveevolvedbeyondtheimmediate
user,researchers
suchas workgroupimpacts[18,34],interadditional
IS impactmeasures,
suggested
consumerimpacts[3, 15], and societal
and
[5,
6],
impacts
organizational industry
is
continuum
of
fromindividuthere
a
entities,
ever-increasing
impacts[40]. Clearly,
Thechoice
als tonationaleconomicaccounts,whichcouldbe affected
byIS activity.
orsystems
ofwheretheimpactsshouldbe measuredwilldependonthesystem
being
evaluatedand theirpurposes.Ratherthancomplicatethemodelwithmoresuccess
measures,we preferto movein theoppositedirectionand groupall the"impact"
called"netbenefits."
for
measuresintoa singleimpactorbenefit
category
Although,
we resistedsuchfurther
some studies,suchfinergranularity
maybe appropriate,
in theAnalysisand
Thisis discussedfurther
refinements
forthesakeofparsimony.
section.
Recommendations
Enhancements
Measurement
sucTables 1 through 6 in the 1992article [8, pp. 65-83] listedthenumerous
thathadbeenusedinpreviousempirical
eachsuccesscategory
cess measureswithin
variable
in thenumberof dependent
reduction
studies.We calledfor"a significant
Since
be
can
measuresso thatresearchresults
then,several
compared"[8, p. 80].
whichmeasureoneormoreof
studieshavedevelopedandtestedsurveyinstruments,
thesesix successconstructs.
literature
Based on a comprehensive
review,MiraniandLederer[32] developeda
Their
derivedfromIS projects.
benefits
tomeasureorganizational
instrument
33-item
straoforganizational
benefits:
ofthreecategories
consisted
framework
measurement
was
tested
instrument
The
transactional.
and
informational,
empirically
proposed
tegic,
in a surveyof 200 IS managersand systemsanalysts.The resultsshowedstrong
for
threesubdimensions
Further
evidenceofdiscriminant
analysisidentified
validity.
subdividedintocomwerefurther
benefits
each of thebenefit
categories.Strategic
Informational
benbenefits.
customer-relations
and
petitiveadvantage,alignment,
information
and
information
efitsincludedinformation
access,
flexibility
quality,
includedcommunication
benefits
transactional
andfinally,
subdimensions;
efficiency,
We believe
subdimensions.
andbusinessefficiency
efficiency,
development
systems
suchas this,thatmeasureIS benefitsare an important
thatvalidatedinstruments,
to IS successmeasurement.
contribution
oftheKaplanand
etal. [30] suggestanadaptation
In a conceptual
paper,Martinsons
oforganizaNorton"BalancedScorecard"(BSC) [23] approachforthemeasurement
thefinancial
The BSC consistsoffourperformance
tionalperformance.
perspectives:
and
businessprocessperspective,
theinternal
thecustomer
perspective,
perspective,
the
authors
to
an
IS
andgrowth
thelearning
context,
proposea
Applied
perspective.
a usermeasurement
balancedIS scorecardto includea business-value
dimension,
dimenfuture-readiness
and
a
an
orientation
dimension,
dimension, internal-process
thensuggestspecificmeasuresrelatedto each IS BSC dimension.
sion.The authors
20
on inandreturn
Forexample,costcontrol,
revenuegeneration,
alignment,
strategic
dimension.
forthebusiness-value
vestment
areamongthemeasuressuggested
12andDoll [52] developeda fourBased on a literature
Torkzadeh
factor,
review,
users
A
of
409
end
of
IS.
iteminstrument
formeasuring
theindividual
survey
impact
The
instrument.
from18 different
was used to testthemeasurement
organizations
evidencealso supports
overallreliability
ofthe12-item
scalewas 0.92.Theempirical
individual
of
instrument.
The
theconvergent
the
anddiscriminant
resulting
validity
dimensions
are:
impact
Taskproductivity
- theextentto whichan application
theuser'soutimproves
putperunitoftime;
Taskinnovation
- theextentto whichan application
helpsuserscreateandtry
outnewideasin theirwork;
- theextentto whichan application
Customer
satisfaction
helpstheusercreate
or
and
for
firm's
internal
external
value the
customers;
- theextent
Management
work
towhichtheapplication
control
helpstoregulate
and
processes performance.
instrument,
Jiangand Klein [20] surveyed
Usinga 24-itemimpactmeasurement
113 managers
across
three
different
system
types:transacimpacts
regarding
systems
tionprocessingsystems(TPS), information
systems(1RS), and decision
reporting
decisionqualityimpactsin
Their
found
that
users
value
(DSS).
systems
study
support
valuesystems
DSS, butotherwise,
impactsforTPS and1RS.Thesefindperformance
fordifferent
that
different
measures
are
typesofsysingssuggest
impact
appropriate
tems.
- oftheUserInformation
Therecontinues
to be muchuse- anddiscussion
Satisfaction(UIS) instrument
[2, 19].Saarinen[39] hasdevelopedanexpandedusersatisfactioninstrument
to the
thatadds development
processand IS impactdimensions
use and productqualitydimensionsof thetraditional
UIS instrument
[2, 19]. His
four52-itemusersatisfaction
was developedandtestedbasedon
instrument
factor,
48 completedIS development
and validitytestsfoundthefour
projects.Reliability
satisfaction
constructs
hadacceptablereliability
andsatisfactory
Theresultvalidity.
instrument
as
serve
a
more
forIS sucing
may
comprehensive
perceptual
surrogate
cess. Li [28] also proposedan extendedinformation
satisfaction
instrument
butdid
nottesttheinstrument.
Otherauthorshaveidentified
problemswithUIS [51] andhaveproposedan alternatesatisfaction
measure[44]. In spiteof thesecriticisms,
UIS continuesto be the
mostcommonly
usedanddevelopedsuccessmeasure;but,whenusedalone,itcannotfullymeasureIS success.
As indicated
usecontinues
tobe a popularsuccessmeasure[ 17, 26,
earlier,
systems
et
Strub
al.
studied
458 usersof a voice mail systemand
47, 48]. However,
[46]
foundthatself-reported
systems
usageandcomputer-recorded
usagewerenotcorrelated.Theirfindings
suggestthatself-reported
systemusageandcomputer-recorded
usageshouldbothbe measuredinempiricalstudiesbecausethetwodo notnecessarwithone another.
ilycorrelate
21
Based on a literature
Doll andTorkzadeh[9] developeda multidimensional
review,
measureof systemsusagebased on thenatureand purposeof a system.
A 30-item
instrument
was
tested
409
users
from
1
8
system
usage
using computer
organizations.
The 30 itemsmeasurethreeunderlying
use fordecision
systemsusage constructs:
anduse forcustomer
use forworkintegration,
service.Theempirical
results
support,
andgeneralapplicability.
reliability,
providedevidenceoftheinstrument's
validity,
Researchers
shouldconsideradoptingand applyingthismorecomprehensive
systemsusageinstrument.
offuture
AgarwalandPrasad[1] studiedbothinitialsystemusageandintentions
use andfoundthatdifferent
factors
affected
initialuse versusfuture
use oftheWorld
Karahannaet al. [24] founddifferent
factorswereassociated
WideWeb.Similarly,
withintention
tousewindowsbetweenpotential
and
users.These
adopters continuing
thatearlyuse andcontinued
use can differ.
twoempiricalstudiesdemonstrate
a
variable
in
use
is
IS
success;
but,toofrequently,
understanding
clearly
key
System
this
construct.
variables
are
used
to
measure
Moreresearch
complex
simpleusage
ofsystems
suchas citedaboveis neededto refinethemultidimensionality
usage.
associatedwithsystemuse and net
Information
qualityhas provento be strongly
benefitsin recentempiricalstudies[38, 54, 57] and especiallyin thecontextof
toMolla andLicker,"[Although
e-commerce
systems[7,29,33, 36,49]. According
as
an
asset to modernbusiness,
has long been considered
information
important
i.e. information
... to higherlevelsof signifihas elevatedcontent,
e-commerce
cance"[33,p. 7]. Information
qualitymeasuresthathavebeenusedinrecente-commercestudies[7,33,36] includeaccuracy,
relevance,
understandability,
completeness,
and variety.
Researchers
are strongly
encourcurrency,
personalization,
dynamism,
oftheirsuccess
qualitymeasuresas a criticaldimension
agedto includeinformation
construct.
measurement
OtherSuccessFrameworks
theD&M IS
Not all of the researchers have attemptedtocritiqueormodify
frameworks
formeaSuccessModel. Some havedevelopedandproposedalternate
basedperspecGroveretal. usedan alternative,
theoretically
suringIS effectiveness.
construct
"tobuilda theoretically-based
oforganizational
tive(theory
effectiveness)
&
and
extends
the
which
for
effectiveness
IS
[DeLone McLean]
complements
space
conandevaluation-type
IS SuccessModel" [13,p. 178].Based on unit-of-analysis
The sixeffectivecreatedsixIS effectiveness
theauthors
textdimensions,
categories.
ness classes are infusionmeasures(i.e., "organizational
impacts"in theD&M IS
measures(notcoveredintheD&M IS SuccessModel),ecoSuccessModel),market
nomicmeasures(i.e.,"organizational
usagemeasures(i.e.,"systemuse"),
impacts"),
measures(i.e.,"indiand
"user
measures
satisfaction"),
(i.e.,
productivity
perceptual
considers"systemquality"and"information
vidualimpact").Theirframework
qualwhereastheD&M IS SuccessModel
effectiveness
constructs,
ity"to be antecedent
theGrover
ofsuccessitself.In summary,
dimensions
considersthemtobe important
servestovalidatetheD&M IS SuccessModel
framework
etal. [13] IS effectiveness
22
froma theoretical
and suggestsan area forextension,
namely,market
perspective
orindustry
impacts.We includemarket
impactsinourupdatedmodeldescribedlater
inthispaper.
forIS evaluation
Smithson
andHirschheim
[45] proposeda conceptualframework
totheevaluainpracticebyapplying
theframework
anddemonstrated
itsusefulness
varioustheoretical
bases
tionofan outsourcing
Theirframework
situation.
presents
forIS evaluation
intothree"zones"ofevaluation:
effectiveness,
efficiency,
organized
andunderstanding.
constructs
ormetrics
couldbe drawnfromtheliteraAppropriate
orturestreamassociatedwitheachconceptualbase; forexample,software
metrics,
and so on. This framework
behavior,
ganizational
sociology,cognitive
psychology,
hardincludesevaluation
areasthatoverlaptheD&M successdimensions,
including
wareandsoftware
costmetrics("systemquality"),systemusage,usersatisfaction,
benefitanalysis,and so on, butalso suggestsmanyothertheoretical
sourcesof IS
evaluation
measures.
Theauthors
thatis a sourceforidentifying
providea framework
and developingIS evaluationmeasuresratherthana singleframework
of success
Their
dimensions
andtheir
framethe
D&M
IS
Success
Model).
(i.e.,
interrelationships
workdoesnotspecifyactualsuccessconstructs
Thismakesthe
andrelatedmeasures.
an
framework
difficult
to applyin practice.However,it does offertheresearcher
alternative
theoretical
framework
fordeveloping
IS evaluationschemes.
AnalysisandRecommendations
As DISCUSSEDEARLIER,IT NOWSEEMSAPPROPRIATE
to add a thirddimension,"service
tothetwooriginal
characteristics,
quality,"
system
"systems
quality"and"information
as discussedearlier,it appearsmoreparsimonious
to combine
quality."Conversely,
"individual"
and"organizational
into
a
"net
benefits."
variable,
impacts"
single
Thisnewvariable,"netbenefits,"
raisesthreeissuesthatmustbe taken
immediately
intoaccount:whatqualifiesas a "benefit"?
forwhom?andatwhatlevelofanalysis?
In theoriginalformulation
oftheD&M Model,theterm"impact"was used.Seddon
inhischaracterization
oftheoutcomes.
[40] used"consequences"and"netbenefits"
We havecometo prefer
theterm"netbenefits"
ourselvesbecausetheoriginalterm
"impacts"maybe positiveor negative,thusleadingto a possibleconfusionas to
whether
theresultsaregood orbad.Also,theinclusionof "net"in "netbenefits"
is
because
no
is
outcome
without
important
whollypositive,
anynegative
consequences.
of thefinalsuccess
Thus,"netbenefits"is probablythemostaccuratedescriptor
variable.
The secondissueofconcernis: benefits
forwhom- thedesigner,
thesponsor,
the
orstakeholders
have
different
user,orothers?Different
actors,players,
may
opinions
as towhatconstitutes
a benefit
tothem[42].Thus,itis impossible
todefinethese"net
benefits"
without
firstdefining
thecontextor frameof reference.
The factthatthe
D&M Model does notdefinethiscontextis a matter
ofdetail,notofoversight.
The
focusof anyproposedstudymustbe defined.Our modelmaybe usefulto both
Microsoft
andtheusercommunity,
buteach mayhavea verydifferent
definition
of
whatconstitutes
netbenefits
andthusIS success.
23
24
DeLONEAND McLEAN
INFORMATION L
QUALITY
^^
K
'
SYSTEMQUALITY
'
s.
INTENTION
TO USE
USE
^^'
/
/
SERVICE
QUALITY
'/
>*
y^
s'
USER
SATISFACTION
'
^^
^*s^*v>^
Z^
NET
BENEFITS
^^
'
3. Updated
D&M IS SuccessModel
Figure
The proposedassociationswould
moreusersatisfaction,
and positivenetbenefits.
would
moreuseofa poorqualitysystem
thenall be positive.In another
circumstance,
Theproposedassoandnegative
netbenefits.
be associatedwithmoredissatisfaction
ciationswouldthenbe negative.
E-Commerce
Success
in particular,
is havinga
Informationtechnology in general, and theInternet
in
dramatic
impacton businessoperations.
Companiesaremakinglargeinvestments
e-commerce
of
their
e-combut
are
to
evaluate
the
success
applications
hard-pressed
mercesystems.
and
IS researchers
haveturnedtheirattention
to developing,
testing,
successmeasures[7, 29, 33, 36, 49]. Molla and Licker[33]
applyinge-commerce
successmodelbasedon theD&M IS SuccessModel.This
proposedan e-commerce
sectiondemonstrates
howtheupdatedD&M IS SuccessModelcanbe adaptedtothe
measurement
world.
challengesofthenewe-commerce
As a powerful
communications
andcommerce
theInternet
is a communimedium,
cationand IS phenomenon
thatlendsitselfto a measurement
framework
(i.e., the
D&M IS SuccessModel) thatis builton communication
and
Shannon
theory(e.g.,
Weaver[43]). Withinthee-commerce
the
users
are
customcontext, primary
system
ersorsuppliers
rather
thaninternal
users.Customers
andsuppliers
use thesystemto
makebuyingorsellingdecisionsandexecutebusinesstransactions.
Theseelectronic
decisionsand transactions
will thenimpactindividualusers,organizations,
indusand
even
national
economies.
This
communications
and
commerce
tries,
processfits
nicelyintotheupdatedD&M IS SuccessModel anditssix successdimensions.
"Systemquality,"
intheInternet
measuresthedesiredcharacterisenvironment,
ticsof an e-commerce
system.Usability,availability,
reliability,
adaptability,
25
andresponsetime(e.g.,downloadtime)areexamplesofqualitiesthatarevaluedbyusersofan e-commerce
system.
"Information
the
content
issue.Webcontent
should
quality"captures e-commerce
be personalized,
to
and
relevant,
complete,
easy understand, secureifwe expect
toinitiate
transactions
viatheInternet
andreturn
prospective
buyersorsuppliers
tooursiteon a regularbasis.
theoverallsupport
"Servicequality,"
deliveredbytheserviceprovider,
applies
of
whether
this
is delivered
a neworgaregardless
support
bytheIS department,
nizationalunit,or outsourced
to an Internet
serviceprovider(ISP). Its importhanpreviously
tanceis mostlikelygreater
sincetheusersarenowourcustomwilltranslate
intolostcustomers
ersandpoorusersupport
andlostsales.
from
a
within
the
"Usage"measureseverything a visitto Website,tonavigation
toexecution
ofa transaction.
retrieval,
site,to information
remainsan important
meansof measuring
ourcustomers'
"Usersatisfaction"
our
and
should
cover
the
entire
of
e-commerce
customer
system
expeopinions
retrieval
and
riencecyclefrominformation
through
payment,
purchase,
receipt,
service.
arethemostimportant
successmeasuresas theycapturethebal"Netbenefits"
on our customers,
ance of positiveand negativeimpactsof thee-commerce
markets,
industries,
economies,and even
organizations,
employees,
suppliers,
our societies.Have Internet
purchasessaved individualconsumerstimeand
suchas largermarkets,
and
supplychainefficiencies,
money?Havethebenefits
for
an
net
benefits
Have
customer
organization?
yieldedpositive
responsiveness
ine-commerce
infrastructure
investments
countries'
yieldeda netpositivegrowth
in e-commerce
infraHave societalinvestments
in thegrossnationalproduct?
"Netbenefits"
measuresmustbe destructure
andeducationreducedpoverty?
investment.
foreache-commerce
andobjectives
termined
Thus,there
bycontext
but
"netbenefits"
willbe a variety
ofe-commerce
measures, manywillbe the
in general.
sameonesthathavebeendevelopedandtestedforIS investments
buttheycannotbe analyzed
successmeasuresaremostimportant,
"Netbenefits"
without
and understood
quality"measurements.
"systemquality"and "information
theimpactofa Websitedesignon
thee-commerce
Forexample,within
environment,
oftheusability
without
an evaluation
cannotbe fullyunderstood
customer
purchases
thatis
decisionsoftheinformation
oftheWebsiteandtherelevanceforpurchasing
purchaser.
providedto theprospective
how thesix dimensionsof theupdatedD&M IS Success
Table 1 demonstrates
framework
to organizethevarioussuccess
Model can be used as a parsimonious
literature.
intheIS ande-commerce
identified
metrics
andConclusions
Summary
the
Ten years ago we publishedan IS success frameworkinordertointegrate
researchworkon IS successthathad been done up to thatpointand to provide
26
Table 1. E-CommerceSuccessMetrics
Systemsquality
Adaptability
Availability
Reliability
Response time
Usability
Information
quality
Completeness
Ease ofunderstanding
Personalization
Relevance
Security
Servicequality
Assurance
Empathy
Responsiveness
Use
Natureofuse
Navigation
patterns
Numberofsitevisits
Numberoftransactions
executed
User satisfaction
Repeat purchases
Repeat visits
User surveys
Netbenefits
Cost savings
Expandedmarkets
Incremental
additionalsales
Reduced search costs
Timesavings
to
an IS successmodelthatattempted
directions
forfuture
research.We presented
natureofIS success.Thehundreds
andinterdependent
capturethemultidimensional
orvalidatedtheoriginal
ofresearchworksthathaveapplied,developed,challenged,
andtothe
modelspeaktotheneedfora commonapproachto successmeasurement
researchdesigns.The succeedingtenyearshaveseen
valueof a modelforframing
intermsoftheimpactsofIS on businessandsocietyas wellas
tremendous
progress
progressin IS research.In lightof thisprogressand change,we feltcompelledto
review,evaluate,andupdateoursuccessmodel.
ourmodelas
studiesthatbothvalidateandsupport
therecentresearch
Considering
wellas thosethatchallengeit,we concludethatouroriginalmodelandrelatedcon-
27
28
includmeasuresshouldcapturetherichnessofuse as a systemphenomenon
not
ofuse,andshould simplymealevel,andappropriateness
ingthenature,
ofuse.
surethefrequency
"Net
and incorporate
researchshouldinvestigate
8. Finally,morefield-study
this
conclusion:
and
Yuthas Youngsupport
measures.
Benefits"
"[Examining
to measuring
andusagemeasuresis notan acceptablealternative
satisfaction
arecorvariables
three
the
Net
Benefits]
[i.e.,
Although
directly.
performance
warrant
to
not
are
them
between
the
strong
sufficiently
related, relationships
has been
forone another"[60,p. 121].Good progress
theiruse as substitutes
on
measures theindiof"NetBenefits"
andtesting
madein thedevelopment
andnationallevels.
vidual,group,firm,
industry,
References
and perceived
1. Agarwal,R., and Prasad,J. The role of innovationcharacteristics
DecisionSciences,28, 3 (1997),
in theacceptanceofinformation
voluntariness
technologies.
557-580.
and analyzing
of a tool formeasuring
2. Bailey,J.E.,and Pearson,S.W. Development
530-545.
5
user
satisfaction.
Science,
29,
(1983),
Management
computer
welfare.
toconsumer
ofinformation
E. Thecontribution
3. Brynjolfsson,
Infortechnology
mationSystems
Research,7, 3 (1996), 281-300.
andindividual
andquantitative
4. Chan,Y.E. IT value:Thegreatdividebetweenqualitative
4
measures.
Journal
andorganizational
16,
2000),
(Spring
Systems,
Information
ofManagement
225-261.
information
coordination
5. Clemons,E.K., and Row,M.C. Limitsto interfirm
through
Journalof
Resultsof a fieldstudyin consumergoods packagingdistribution.
technology:
10, 1 (Summer1993),73-95.
Systems,
Management
Information
onthe
6. Clemons,E.K.; Reddi,S.P.; andRow,M.C. Theimpactofinformation
technology
JournalofManageofeconomicactivity:
The "movetothemiddle"hypothesis.
organization
mentInformation
10, 2 (Fall 1993),9-35.
Systems,
oftheWorldWideWeb.
factors
inuserevaluation
7. D'Ambra,J.,andRice R.E. Emerging
& Management,
38, 6 (2001), 373-384.
Information
success:Thequestforthedepen8. DeLone,W.H.,andMcLean,E.R. Information
systems
dentvariable.Information
Research,3, 1 (1992), 60-95.
Systems
use in
measureofsystems
9. Doll,W.J.,andTorkzadeh,
G. Developinga multidimensional
an organizational
context.
& Management,
33, 4 (1998), 171-185.
Information
modelof end usercomputing
10. Etezadi-Amoli,J.,and Farhoomand,
A.F. A structural
& Management,
satisfaction
anduserperformance.
30, 2 (1996), 65-73.
Information
11. Gelderman,
M. The relationbetweenusersatisfaction,
systems,
usage of information
andperformance.
& Management,
Information
34, 1 (1998), 11-18.
fitand individualperformance.
12. Goodhue,D.L., and Thompson,
R.L. Task-technology
MIS Quarterly,
19, 2 (1995), 213-233.
The con13. Grover,G.; Jeong,S.R.; and Segars,A.H. Information
systemseffectiveness:
struct
ofapplication.
& Management,
31, 4 (1996), 177-191.
space andpatterns
Information
14. Guimares,
thehumanside.
T.,andIgbana,M. Client/server
systemsuccess:Exploring
DecisionSciences,28, 4 (1997), 851-875.
15. Hitt,L., andBrynjolfsson,
E. The threefacesofIT value:Theoryandevidence.In J.I.
on
DeGross,S.L. Huff,andM.C. Munro(eds.),ProceedingsoftheInternational
Conference
Atlanta,GA: AssociationforInformation
Information
Systems.
Systems,1994,pp. 263-278.
16.Igbaria,M., andTan,M. Theconsequencesoftheinformation
acceptanceon
technology
individual
& Management,
32, 3 (1997), 113-121.
subsequent
performance.
Information
17. Igbaria,M.; Zinatelli,N.; Cragg,P.; and Cavaye,A. Personalcomputing
acceptance
factors
on smallfirms:
A structural
21, 3 (1997), 279-302.
equationmodel.MIS Quarterly,
29
30