One of the most enduring research topics in the � eld of information systems (IS) is that of systemsuccess (Lyytinen and Hirschheim, 1987; deLone andMcLean, 1992; Ballantine et al., 1996). Prior researchhas addressed the measurement of success, theantecedents of success and explanations of success orfailure. Yet for each new type of information tech-nology (IT) or application the question of successcomes up again. In the case of enterprise resourceplanning (ERP) systems success takes on a specialurgency, since the costs and risks of these massive tech-nology investments rival their potential pay-offs.Failures of ERP system implementation projects havebeen known to lead to organizational bankruptcy(Bulkeley, 1996; Davenport, 1998; Markus and Tanis,2000).
Brie� y, ERP systems are commercial softwarepackages that enable the integration of transactions-oriented data and business processes throughout an organization. From a base in manufacturing and� nancial systems, ERP systems may eventually allowfor integration of interorganizational supply chains(Davenport, 1998; Markus and Tanis, 2000). Becausethese systems touch so many aspects of a company’s
internal and external operations, their successfuldeployment and use are critical to organizationalperformance and survival.
This paper describes the results of a study of prob-lems and outcomes in ERP projects. The study wasconducted under the sponsorship of an ERP vendorwho was interested in helping its customers be moresuccessful in ERP implementation. Two basic researchquestions are addressed: First, how successful arecompanies at different points in time in their ERPexperiences, and how are different measures of successrelated? (That is, can early success be followed byfailure and vice versa?) Second, what problems do ERPadopters encounter as they implement and deployERP, and how are these problems related to outcomes?
Success with ERP and how it happens
The de� nition and measurement of success are thornymatters. First, success depends on the point of view from which you measure it. It became clear earlyon in our research that people often mean different things when talking about ERP success. For example,people whose job it was to implement ERP systems (e.g.project managers and implementation consultants)
Journal of Information Technology (2000) 15, 245–265
Learning from adopters’ experiences with ERP:problems encountered and success achievedM. LYNNE MARKUS, SHERYL AXLINE*, DAVID PETRIE‡ AND CORNELIS TANIS§City University of Hong Kong, Hong Kong, PR China,* School of Behavioral and Organizational Sciences and ‡School of Information Science, Claremont Graduate University, 1021 North Dartmouth Avenue, Claremont, CA 91711, USA, §Coach and Commitments, Utrecht, The Netherlands
Enterprise resource planning (ERP) packages touch many aspects of a company’s internal and externaloperations. Consequently, successful deployment and use of ERP systems are critical to organizational perfor-mance and survival. This paper presents the results of a study of the problems and outcomes in ERP projectswhich was conducted under the sponsorship of an ERP systems vendor. Two basic research questions wereaddressed. First, how successful are companies at different points in time in their ERP experiences and howare different measures of success related? (That is, can early success be followed by failure and vice versa?)Second, what problems do ERP adopters encounter as they implement and deploy ERP and how are theseproblems related to outcomes? The � ndings showed that the success of ERP systems depends on when itis measured and that success at one point in time may only be loosely related to success at another pointin time. Companies experience problems at all phases of the ERP system life cycle and many of the problemsexperienced in later phases originated earlier but remained unnoticed or uncorrected. These � ndings suggestthat researchers and companies will do well to adopt broad de� nitions and multiple measures of successand pay particular attention to the early identi� cation and correction of problems.
often de� ned success in terms of completing the projectplan on time and within budget. However, peoplewhose job it was to adopt ERP systems and use themin achieving business results tended to emphasize hav-ing a smooth transition to stable operations with thenew system, thereby achieving intended businessimprovements such as inventory reductions and gain-ing improved decision support capabilities.
In this paper we adopt an inclusive perspective thatfocuses on the organizations that adopt ERP systemsand the individuals within these organizations (ratherthan on ERP vendors and external implementationconsultants). We recognize that our ‘etic’ perspective(non-interpretive, outside looking in) may not havecorresponded with that of any particular actor(s) inthe organizations we studied, but it allowed us toinclude many different dimensions in our assessmentof success, including the following.
(1) Success viewed in technical terms.(2) Success viewed in economic, � nancial or
strategic business terms.(3) Success viewed in terms of the smooth running
of business operations.(4) Success as viewed by the ERP-adopting organ-
ization’s managers and employees.(5) Success as viewed by the ERP-adopting organ-
ization’s customers, suppliers, and investors.
A second important issue in the measurement ofsuccess concerns when one measures it. Some yearsago, Peters and Waterman (1982) attracted muchattention with their study of ‘excellent companies’. Afew years later, a sizeable number of their excellentcompanies were no longer star performers. Projectmanagers and implementers can afford to declaresuccess in the short run, but executives and investorsare in it for the long haul. The organizations that adoptERP systems need to be concerned with success, notjust at the point of adoption but also further down theroad. The importance of considering ERP success atmultiple points in time was made clear in a case studyby Larsen and Myers (1997) in which a successfullyinstalled ERP system was later terminated when thecompany merged with another.
In this study we were concerned with the assess-ment of success at three different points in time duringthe adopting organization’s experience with an ERPsystem. We can conceptually differentiate three distinctphases in the ‘ERP experience cycle’ (Markus andTanis, 2000): (1) the project phase during which ERPsoftware is con� gured and rolled out to the organiza-tion, (2) the shakedown phase during which thecompany makes the transition from ‘go live’ to ‘normaloperations’ and (3) the onward and upward phaseduring which the company captures the majority of
business bene� ts (if any) from the ERP system andplans the next steps for technology implementation and business improvement. A number of successmetrics can be de� ned for each of these phases.
Success in the project phase
(1) Project cost relative to budget.(2) Project completion time relative to schedule.(3) Completed and installed system functionality
relative to original project scope.
Success in the shakedown phase
(1) Short-term changes occurring after system ‘go-live’ in key business performance indicators suchas operating labour costs.
(2) Length of time before key performance indica-tors achieve ‘normal’ or expected levels.
(3) Short-term impacts on the organization’sadopters, suppliers and customers such as aver-age time on hold when placing a telephoneorder.
Success in the onward and upward phase
(1) Achievement of business results expected for theERP project, such as reduced IT operating costsand reduced inventory carrying costs.
(2) Ongoing improvements in business results afterthe expected results have been achieved.
(3) Ease in adopting new ERP releases, other newITs, improved business practices, improveddecision making, etc., after the ERP system hasachieved stable operations.
These success metrics include indicators of humanand organizational learning. It is important not justhow well the ERP system itself performs (e.g. accu-racy, reliability and response time), but how wellpeople in the organization know how to use, maintainand upgrade the ERP system and how well the busi-ness improves its performance with the ERP system.
An unresolved question is the relationship betweenthe measures of success at different points in time.Larsen and Myers (1997) found that an ERP experi-ence could be an early success and a later failure. Butcan an ERP experience be an early failure yet a latersuccess? How important is it for organizations to besuccessful at all three phases of the ERP experiencecycle? And how often do organizations push throughinitial failure to achieve an ultimate measure of success?These are empirical questions.
246 Markus et al.
A third important issue in the measurement ofsuccess is the yardstick or criterion against which tocompare an actual level of achievement. It is quitecommon in systems evaluation, technology assessmentand impact studies to use the adopters’ objectives,expectations and perceptions as the standard forde� ning and measuring success. Naturally, these sub-jective judgements of success can be quite importantin understanding how organizations behave. If acompany stops using an ERP system because cor-porate objectives have not been met, it does not matterthat an outside observer might have assessed theimplementation project and system operation assuccessful.
However, there are serious disadvantages to usingperceptions, objectives and expectations as the solemeasures of success. In the � rst place, it is hard tonormalize them across individuals and organizations,thus making comparisons dif� cult. Second, their rela-tionship with so-called ‘objective’ measures of success(such as whether or not a project is terminated priorto completion), (cf. Sauer, 1993) is unclear. People’sobjectives for and expectations of ERP systems maybe overly ambitious so that they are unrealizable no matter what people do. Alternatively, or they maybe insuf� ciently ambitious so that people do not takefull advantage of the capabilities ‘in’ the technologywhich are available for them to use (Markus and Tanis,2000).
If one wants to compare the outcomes achieved by the organizations that have adopted ERP systems,it is useful to have an external yardstick of success in addition to internal perceptual measures and localobjectives and expectations. For this purpose, weproposed using optimal success, as de� ned by (Markusand Tanis, 2000), as our crieterion as follows:
Optimal success refers to the best outcomes theorganization could achieve with enterprise systems,given its business situation, measured against a port-folio of project, early operational, and longer termbusiness results metrics. Optimal success can be far more or less than the organization’s goals for anenterprise system. Further, optimal success can bedynamic; what is possible for an organization toachieve may change over time as business conditionschange (p. 186–7).
Naturally, the concept of optimal success de� nedthus is dif� cult to operationalize. However, the advan-tage of attempting to assess outcomes by using suchan external, non-interpretive yardstick is that it helpsus compare the results achieved in different organiza-tions and explore the interesting relationships between‘objective’ outcomes and people’s perceptions ofresults.
The phrase optimal success suggests that mostorganizations experience outcomes that fall somewhatshort of what a ‘best in class’ organization mightachieve. This observation directs attention at the prob-lems companies experience when they adopt, deployand use ERP systems and how they respond whenproblems arise. This is not a focus on ‘success factors’per se, but on aspects of the ‘lived experience’ of organ-izations’ ERP journeys. One wants to know how (theprocess by which) some companies realize better orworse outcomes than other companies do and whatthey do that makes the difference. Put differently, onewants to know whether all companies experience thesame types of problems with ERP systems, whetherthey respond similarly to the problems and whetherthe problems and responses are related to the outcomesthey experience.
It should be clear that we believe that the outcomescompanies achieve with ERP systems (varying degreesof suboptimality, relative to what they could achieve,if all went perfectly well) are non-deterministic.Problems such as a lack of resources or turnover ofpersonnel can arise in each phase of the ERP experi-ence cycle. They may or may not be perceived as prob-lems by the people in the organization and, even whenpeople perceive the problems as problems, they mayor may not take appropriate actions for resolving them.As a result, the outcomes in a particular phase maybe optimal or less than optimal and the problems mayor may not remain unresolved, thereby affectingoutcomes later. See Markus and Tanis (2000) for amore detailed treatment of this ‘theory’ of ERPsuccess.
In practice, it can be extremely dif� cult to differen-tiate between problems, symptoms of problems andoutcomes (that is, the consequences of problems).Nevertheless, the importance and complexity of theERP experience suggests the need to try. Therefore,this study addresses two related questions about theERP experience. First, how successful are companiesat different points in time in their ERP experiencesand how are different measures of success related?(That is, can early success be followed by failure andvice versa?) Second, what problems do ERP adoptersencounter as they implement and deploy ERP and howare these problems related to outcomes?
Table 1 helps to frame the � ndings of this researchby providing a more complete description of the issuesinvolved in assessing the success of ERP projects.Table 1 outlines (1) the activities that characterize eachphase of the ERP life-cycle, (2) how and why the activ-ities in each phase may affect the outcomes an adopterachieves, not just in the phase but also downstream,(3) some of the problems an ERP adopter may expe-rience during the phase (which can also be understood
Learning from adopters’ experiences with ERP 247
248 Markus et al.
Tab
le 1
Ass
essi
ng
ach
ieve
d s
ucc
ess
in t
he
ER
P e
xper
ien
ce
Maj
or a
ctiv
itie
s of
th
e p
has
eIm
plic
atio
ns
of p
has
e ac
tivi
ties
for
C
omm
on
pro
ble
ms
Ph
ase
succ
ess
mea
sure
ssu
cces
s in
ph
ase
and
lat
er
Pro
ject
ph
ase
Pro
ject
tea
m f
orm
atio
n a
nd
tra
inin
gT
he
maj
orit
y of
ER
P e
xpen
dit
ure
pla
ns
Inab
ility
to
acq
uir
e/re
tain
em
plo
yees
an
d
Pro
ject
cos
t re
lati
ve t
o D
evel
op e
nte
rpri
se m
odel
fo
r co
n�
gura
tion
ar
e m
ade
du
rin
g th
is p
has
eex
tern
al a
dvi
sers
wit
h r
equ
isit
e ex
per
tise
in
b
ud
get
and
dev
elop
an
d v
alid
ate
kern
el i
n
Few
ben
e�ts
are
exp
erie
nce
d d
uri
ng
this
E
RP
, p
roje
ct m
anag
emen
t an
d s
up
por
tin
g P
roje
ct c
omp
leti
on t
ime
mu
ltip
le i
mp
lem
enta
tion
sp
has
e u
nle
ss t
he
orga
niz
atio
n p
urs
ues
a
tech
nol
ogie
sre
lati
ve t
o sc
hed
ule
Con
�gu
re E
RP
sof
twar
e to
re�
ect
eith
er
‘qu
ick
win
s’ o
r ‘l
ow h
angi
ng
fru
it’
stra
tegy
Tu
rnov
er o
f p
roje
ct s
pon
sor
or p
roje
ctC
om
ple
ted
sys
tem
cu
rren
t op
erat
ion
s o
r p
lan
ned
new
o
f id
enti
fyin
g an
d i
mp
lem
enti
ng
bu
sin
ess
man
ager
fun
ctio
nal
ity
rela
tive
to
bu
sin
ess
pro
cess
esp
roce
ss i
mp
rove
men
ts w
hile
ER
P p
lan
nin
g E
xces
sive
tu
rno
ver
(an
d/o
r st
ress
-rel
ated
o
rigi
nal
pro
ject
sco
pe
Des
ign
an
d e
xecu
te c
han
ges
(if
any
are
and
con
�gu
rati
on i
s u
nd
er w
ayh
ealt
h p
rob
lem
s) o
n p
roje
ct t
eam
pla
nn
ed)
in t
he o
rgan
izat
ion
’s b
usi
nes
s T
he
lon
ger
the
pro
ject
ph
ase,
th
e lo
wer
U
nw
illin
gnes
s of
bu
sin
ess
man
ager
s an
d
pro
cess
es a
nd
rel
ated
org
aniz
atio
nal
th
e ov
eral
l �
nan
cial
ben
e�ts
fro
m t
he
key
use
rs t
o m
ake
tim
e fo
r p
roje
ct
elem
ents
(or
gan
izat
ion
str
uct
ure
, jo
bs,
sy
stem
on
a d
isco
un
ted
cas
h �
ow b
asis
acti
viti
esco
mp
ensa
tion
, et
c.)
If t
he
pro
ject
goe
s ve
ry b
adly
, d
ecis
ion
M
ajor
ch
ange
s in
pro
ject
sco
pe
afte
r st
art
Imp
lem
ent
add
-on
s, m
odi�
cati
ons
and
m
aker
s m
ay t
erm
inat
e it
of
pro
ject
inte
rfac
es w
ith
oth
er e
nte
rpri
se s
yste
ms
Wh
en t
he
sch
edu
le g
ets
tigh
t, t
eam
may
P
oor
qu
alit
y so
ftw
are,
doc
um
enta
tio
n a
nd
an
d l
egac
y sy
stem
sd
ecid
e to
cu
t sc
ope,
so
that
str
ateg
ical
ly
trai
nin
g m
ater
ials
Acq
uir
e IT
in
fras
tru
ctu
re r
esou
rces
an
d
esse
nti
al p
roce
sses
are
not
su
pp
orte
dM
odi�
cati
ons
that
do
not
wor
k an
d d
elay
s in
tegr
ate
ER
P s
yste
m w
ith
in
fras
tru
ctu
re
in d
evel
opm
ent
of
mod
i�ca
tion
s an
d
and
leg
acy
syst
ems
(if
any)
inte
rfac
esD
ocu
men
t co
n�
gura
tio
n d
ecis
ion
s an
d
Con
�ic
ts w
ith
im
ple
men
tati
on
con
sult
ants
ra
tion
ale
ove
r p
roje
ct p
lan
s an
d m
anag
emen
tD
ecid
e ho
w t
o sa
tisf
y d
ecis
ion
su
pp
ort/
Cu
ttin
g te
stin
g an
d/o
r tr
ain
ing
wh
en
rep
orti
ng
nee
ds
sch
edu
le g
ets
tigh
tC
omm
un
icat
ion
an
d c
han
ge m
anag
emen
tP
ress
ure
to
term
inat
e p
roje
ct i
f co
st a
nd
C
lean
up
dat
a an
d c
onve
rt d
ata
to n
ew
sch
edu
le o
verr
un
s oc
cur
ER
P s
yste
mT
est
the
new
sys
tem
Tra
in u
sers
Learning from adopters’ experiences with ERP 249
Tab
le 1
con
tin
ued
Maj
or a
ctiv
itie
s of
th
e p
has
eIm
plic
atio
ns
of p
has
e ac
tivi
ties
for
C
omm
on p
rob
lem
s P
has
e su
cces
s m
easu
res
succ
ess
in p
has
e an
d l
ater
Sh
aked
own
ph
ase
Mak
e th
e tr
ansi
tion
to
‘nor
mal
op
erat
ion
’ T
he
orga
niz
atio
n m
ay n
ot
be
able
to
Ext
rem
ely
poo
r sy
stem
per
form
ance
Sh
ort-
term
det
erio
rati
on i
n
of t
he
new
sys
tem
an
d t
he
new
bu
sin
ess
real
ize
pla
nn
ed i
mp
rove
men
ts i
n I
T c
osts
E
xces
sive
str
ess
and
/or
turn
ove
r of
key
ke
y (b
usi
nes
s) p
erfo
rman
ce
pro
cess
esan
d/o
r b
usi
nes
s p
roce
ss e
f�ci
ency
un
til (
1)
use
rs a
nd
/or
key
syst
em s
up
por
t p
erso
nn
elin
dic
ator
s (K
PIs
) (e
.g.
pro
cess
‘R
ewor
k’ (
mis
take
cor
rect
ing)
act
ivit
ies
the
new
sys
tem
sta
bili
zes,
(2
) th
e ol
d
Exc
essi
ve d
epen
den
ce o
n ‘
key
use
rs’
cycl
e ti
mes
, in
ven
tory
lev
els
may
in
clu
de
chan
gin
g co
n�
gura
tion
sy
stem
s ar
e in
terf
aced
or
turn
ed o
ff a
nd
(p
roje
ct t
eam
mem
ber
s) a
nd
/or
IT
and
op
erat
ing
lab
our
cost
s)se
ttin
gs,
up
grad
ing
IT i
nfr
astr
uct
ure
, ol
der
IT
res
ourc
es a
re r
emov
ed f
rom
sp
ecia
list
sL
engt
h o
f ti
me
bef
ore
KP
Is
revi
sin
g b
usi
nes
s p
ract
ices
an
d p
roce
du
res
mai
nte
nan
ce a
gree
men
ts a
nd
(3)
use
rs
Mai
nte
nan
ce o
f o
ld p
roce
du
res
or m
anu
al
and
bu
sin
ess
imp
acts
ret
urn
an
d r
etra
inin
g u
sers
ach
ieve
fu
ll p
ro�
cien
cy w
ith
th
e n
ew
wor
karo
un
ds
in l
ieu
of
lear
nin
g th
e to
no
rmal
syst
emre
leva
nt
syst
em c
apab
ilit
ies
Sh
ort-
term
neg
ativ
e im
pac
ts
In a
dd
itio
n,
the
orga
niz
atio
n m
ay h
ave
to
Dat
a in
pu
t er
rors
on o
rgan
izat
ion
’s s
up
plie
rs
mak
e si
gni�
can
t n
ew e
xpen
dit
ure
s fo
r In
abili
ty t
o d
iagn
ose
and
rem
edy
syst
em
and
cu
stom
ers
(e.g
. av
erag
e te
mp
orar
y an
d o
vert
ime
lab
ou
r, c
onsu
ltin
g an
d/o
r b
usi
nes
s p
roce
ss p
erfo
rman
ce
tim
e on
hol
d,
lost
cal
ls,
lost
h
elp
an
d a
dd
itio
nal
IT
res
ourc
es i
n o
rder
p
rob
lem
ssa
les
and
cu
stom
er
to c
om
ple
te t
he t
ran
siti
on f
rom
‘go
liv
e’
Ext
rem
ely
neg
ativ
e re
acti
on
s fr
om
sati
sfac
tion
lev
els)
to ‘
nor
mal
op
erat
ion
s’cu
stom
ers
and
su
pp
liers
(e.
g. l
arge
los
ses
Th
e lo
nge
r th
e sh
ake
dow
n p
has
e, t
he
of b
usi
nes
s)lo
wer
th
e ov
eral
l b
usi
nes
s be
ne�
ts o
n a
A
bse
nce
of
shar
p a
nd
fas
t im
pro
vem
ents
d
isco
un
ted
cas
h �
ow b
asis
du
rin
g sh
ake
do
wn
If s
hak
e d
own
goe
s ve
ry b
adly
, th
e sy
stem
A
bse
nce
of
sati
sfac
tory
man
agem
ent
may
be
rem
oved
or
the
orga
niz
atio
n m
ay
info
rmat
ion
/an
alys
is a
nd
rep
orti
ng
bec
ome
un
wil
ling
to u
nd
erta
ke f
utu
re
Pre
ssu
re t
o d
e-in
stal
l sy
stem
syst
em i
mp
rove
men
ts (
e.g.
, u
pgr
ades
)
250 Markus et al.
Tab
le 1
con
tin
ued
Maj
or a
ctiv
itie
s of
th
e p
has
eIm
plic
atio
ns
of p
has
e ac
tivi
ties
for
C
omm
on p
rob
lem
s P
has
e su
cces
s m
easu
res
succ
ess
in p
has
e an
d l
ater
On
war
d a
nd
up
war
d p
has
e
On
goin
g op
erat
ion
an
d u
se o
f sy
stem
an
d
Th
e m
ajor
ity
of b
usi
nes
s b
ene�
ts (
if a
ny)
‘N
orm
al o
per
atio
n’
nev
er m
ater
iali
zes
Ach
ieve
men
t of
pla
nn
ed
bu
sin
ess
pro
cess
aft
er t
he
shak
edo
wn
ar
e ac
hie
ved
aft
er s
hak
e d
own
Non
-im
pro
vem
ent
in u
sers
’ E
RP
ski
ll b
usi
nes
s re
sult
s (e
.g.
IT
pha
seM
any
des
ired
bu
sin
ess
ben
e�ts
may
no
t le
vels
(e.
g. m
any
pot
enti
al u
sers
rem
ain
op
erat
ing
cost
s, i
nve
nto
ry
Pla
nn
ing
for
up
grad
es a
nd
mig
rati
on
to
be
pos
sibl
e w
ith
th
e cu
rren
t re
leas
e, b
ut
un
trai
ned
an
d u
sers
rou
tin
ely
rely
on
ca
rryi
ng
cost
s, b
usi
nes
s la
ter
rele
ases
/ver
sion
s of
har
dw
are
and
m
ay r
equ
ire
the
orga
niz
atio
n t
o u
nd
erta
ke
pro
ject
tea
m m
emb
ers
and
tec
hn
ical
p
roce
ss c
ost
and
cyc
le t
ime)
ER
P s
oftw
are
a se
ries
of
up
grad
es (
e.g.
red
uct
ion
s in
an
su
pp
ort
per
son
nel
to
per
form
‘n
orm
al’
job
U
se o
f d
ata
and
dec
isio
n
Ad
opti
on o
f ad
dit
ion
al m
odu
les/
pac
kage
s or
gan
izat
ion
’s I
T p
erso
nn
el e
xpen
dit
ure
s ac
tivi
ties
)an
alys
es p
rod
uce
d b
y th
e an
d i
nte
grat
ion
wit
h E
RP
may
not
be
real
izab
le d
uri
ng
the
init
ial
Fai
lure
to
reta
in p
eop
le w
ho
un
der
stan
d
syst
emB
usi
nes
s d
ecis
ion
mak
ing
bas
ed o
n d
ata
ER
P e
xper
ien
ce c
ycle
an
d a
chie
vin
g th
e im
plem
enta
tion
an
d u
se o
f E
RP
O
ngo
ing
imp
rove
men
ts i
n
pro
vid
ed b
y th
e E
RP
sys
tem
bu
sin
ess
pro
cess
‘vi
sib
ility
’ ac
ross
sit
es
syst
ems
bu
sin
ess
resu
lts
(aft
er
Con
tin
uou
s im
pro
vem
ent
of u
sers
’ oc
curs
aft
er a
ll si
tes
hav
e b
een
N
o d
ocu
men
tati
on o
f ra
tion
ale
for
pla
nn
ed r
esu
lts
hav
e b
een
IT
ski
llsim
ple
men
ted
)b
usi
nes
s ru
les
and
con
�gu
rati
on d
ecis
ion
sac
hie
ved
)C
onti
nu
ous
bu
sin
ess
pro
cess
im
pro
vem
ent
Man
y b
ene�
ts c
ann
ot o
ccu
r u
nti
l: (
1)
Dif
�cu
lty
in o
pti
miz
ing
syst
em
Eas
e in
dev
elop
ing/
adop
tin
g/in
ord
er t
o ac
hie
ve b
ette
r b
usi
nes
s re
sult
su
sers
hav
e le
arn
ed h
ow t
o u
se t
he
syst
em
per
form
ance
an
d i
n r
econ
�gu
rin
g th
e im
ple
men
tin
g ad
dit
ion
al
Rec
on
�gu
rati
on o
f cu
rren
t re
leas
e/ve
rsio
nw
ell,
(2
) m
anag
ers
hav
e u
sed
th
e d
ata
syst
em t
o s
up
por
t bu
sin
ess
inn
ovat
ion
sin
nov
atio
ns
in t
ech
nol
ogy
, co
llec
ted
by
the
syst
em i
n o
rder
to
mak
e U
nw
illin
gnes
s of
org
aniz
atio
n t
o ad
opt
bu
sin
ess
pra
ctic
es a
nd
b
usi
nes
s d
ecis
ion
s an
d p
lan
im
pro
vem
ents
ad
dit
ion
al c
han
ges
in b
usi
nes
s p
roce
sses
, m
anag
eria
l d
ecis
ion
mak
ing
in b
usi
nes
s p
roce
sses
an
d (
3)
add
itio
nal
sy
stem
co
n�
gura
tion
s or
IT
in
fras
tru
ctu
reO
rigi
nal
dec
isio
n t
o ch
ange
s ar
e m
ade
in b
usi
nes
s p
roce
sses
, P
ress
ure
to
de-
inst
all
syst
emim
ple
men
t E
RP
sti
ll m
akes
p
ract
ices
, so
ftw
are
con
�gu
rati
on,
etc.
sen
se i
n l
igh
t of
su
bse
qu
ent
bu
sin
ess
dec
isio
ns
and
ev
ents
(e.
g. m
erge
rs a
nd
ac
qu
isit
ion
)(O
ver
tim
e) d
ecre
ases
in
le
ngt
h o
f p
roje
ct p
lan
nin
g an
d s
hak
e d
own
ph
ases
for
su
bse
qu
ent
ER
P
imp
lem
enta
tion
s
as indicators that the experience may be heading for suboptimal success) and (4) the success measuresrelevant to the phase.
Approach
This research study combined several methods: (1) reviews of published and in-process research studies and teaching cases of ERP implementations,(2) in-depth case studies of the ERP experience in � ve ERP-adopting organizations following the proce-dures prescribed by Yin (1994), 3) interviews with 11additional ERP-adopting organizations and (4) ap-proximately 20 interviews with ERP implementationconsultants and members of the ERP vendor companysponsoring this study. Table 2 describes each of the16 ERP-adopting organizations that directly partici-pated in this research. At the same time, the analysisand interpretation of the results presented in this reportre� ect the experiences of a much larger number ofcompanies (approximately 40 in total), including thosedescribed in teaching cases, other research reports andthe trade press.
The 11 ERP adopter interviews were conducted byphone or in person: one or more members of theresearch team discussed the ERP experience with oneor more members of the adopting organization. Theinterviews ranged from 1 to 3 h in length. The casestudies involved a much more signi� cant level of effort.Two to four members of the research team visited thecase site for 2–4 days, interviewing 12–25 people.Documents describing the company and its imple-mentation effort were collected and analysed. Noteswere transcribed and reviewed by project teammembers and summaries were written. The detail andthoroughness of the case study method meant that itwas not necessary to examine a large number of casesin order to gain the bene� ts of this research strategyin analysing ‘how and why’ research questions (Yin,1994). For such scienti� c purposes, four to 12 casestudies are considered perfectly adequate.
Several criteria were used in selecting the compa-nies for this study. First, we selected companies thatwere interested in learning about how to improve theirERP experiences from the research. These companieswere recruited at public presentations where wedescribed our research project.
Second, we studied companies at different stages ofthe ERP experience. Studying projects in processprovides useful knowledge about how the ERP expe-rience unfolds over time. This is particularly useful inidentifying why companies act the way they do. Afterthe project is over, people forget many details andreconstruct the past in order to be consistent with
known outcomes. Studying completed projects allowsresearchers to identify the key causal factors in successor failure. Thus, we aimed for a mix of both completedand in-process projects. Table 3 shows the stage ofcompletion reached by each company at the time wecollected data. Because some of the cases we studiedwere in process at the time of data collection, we donot have complete outcome data for all companies.
Third, we went out of our way to select projectsthat had experienced problems rather than projects thatwere unquali� ed successes. A major goal of the studywas understanding the problems adopters experiencewith ERP systems, why these problems occur and whatcould be done about them. Therefore, we skewed oursample towards companies with problems and subop-timal success. This means that the companies exam-ined in this study may not be a representative sampleof all companies using ERP systems. It would not bevalid to draw conclusions from this study about howfrequently ERP adopters experience certain problemsor how frequently they achieve success (or lack of it)on different measures.
Two additional factors may limit the potential statis-tical generalizability (Yin, 1994) of the results. First,all 16 of the adopter companies we studied were basedin North America or Europe. Second, all 16 compa-nies used the ERP products of a single software vendor. However, we do not believe that these factorsmaterially affected our � ndings about the kinds ofproblems and outcomes companies experience withERP systems. Our � ndings closely tracked reports byother academics and journalists. Further, these factorswere not likely to affect the analytical generalizability(Yin, 1994) of our results. Although the current studydesign did not provide reliable data about frequencies,it could provide reliable insights into how and whyproblems and outcomes occur when they do occur.
Findings
Table 4 presents a summary of the problems andoutcomes reported by the companies participating inthis research. Immediately below, we present someinteresting generalizations about the nature of successacross the ERP life-cycle. In a subsequent section, wediscuss the problems companies experienced.
Findings about adopters’ achieved success withERP systems
First, none of the ERP adopters we studied was anunquali� ed success at all of the stages of the experi-ence cycle completed at the time of our data collec-tion. This is to be expected given the nature of our
Learning from adopters’ experiences with ERP 251
252 Markus et al.
Tab
le 2
Ove
rvie
w o
f co
mp
anie
s in
th
e st
ud
y
Co
mp
any
Com
pan
y d
escr
ipti
onE
RP
im
plem
enta
tion
des
crip
tio
n –
D
ata
sou
rces
iden
ti�
erst
atu
s at
tim
e of
dat
a co
llect
ion
Co
mp
any
A$
250
mil
lion
US
co
mp
any
Pro
ject
was
ju
sti�
ed t
o an
d a
pp
rove
d b
y co
mp
any
Fo
ur
rese
arch
ers
per
form
ing
Com
pan
y h
ad g
row
n t
hro
ugh
pas
t m
erge
rs
boa
rdap
pro
xim
atel
y 1
2 h
of
inte
rvie
ws
wit
h
that
wer
e n
ever
fu
lly i
nte
grat
edS
ingl
e-si
te i
mp
lem
enta
tion
wit
h a
pp
roxi
mat
ely
one
key
info
rman
t p
lus
3 h
in
terv
iew
S
ingl
e m
anu
fact
uri
ng
loca
tio
n,
fun
ctio
nal
ly
400
use
rsw
ith
ch
ief
info
rmat
ion
of�
cer
(CIO
) or
gan
ized
All
maj
or
bu
sin
ess
fun
ctio
ns
incl
ud
ed i
n p
roje
ct,
(on
e re
sear
cher
)A
ssem
ble
to
ord
er m
anu
fact
uri
ng
plu
s so
me
incl
ud
ing
man
ufa
ctu
rin
g, �
nan
ce a
nd
dis
trib
uti
on
Dat
a co
llect
ed i
n U
S i
n M
arch
th
rou
gh
cust
om b
usi
nes
s p
roce
sses
(bu
t ex
clu
din
g h
um
an r
esou
rces
).to
Sep
tem
ber
199
8In
du
stry
det
ails
om
itte
d a
t co
mp
any
req
ues
tS
yste
m w
ent
live
Jun
e 19
97
Tot
al p
roje
ct c
ost
app
roxi
mat
ely
$1
7 m
illio
n i
ncl
ud
ing
har
dw
are
and
con
sult
ing
serv
ices
Co
mp
any
B$
1.2
bill
ion
glo
bal
Nor
th A
mer
ican
-bas
ed
Com
pan
y w
as h
avin
g d
if�
cult
y d
ecid
ing
wh
ich
T
hre
e re
sear
cher
s p
erfo
rmin
g m
anu
fact
uri
ng
com
pan
y w
ith
70
00
soft
war
e re
leas
e to
im
ple
men
tap
pro
xim
atel
y 2
4 h
of
inte
rvie
ws
wit
h 1
6 em
plo
yees
an
d 1
70
site
s al
l ov
er t
he
wor
ldA
ppro
xim
atel
y 5
00 c
oncu
rren
t u
sers
wor
ld-w
ide
mem
ber
s of
im
ple
men
tati
on
tea
m,
Fif
th o
r si
xth
in
th
eir
ind
ust
ry.
wer
e ex
pec
ted
incl
ud
ing
CIO
an
d c
onsu
ltan
tsM
ake-
to-s
tock
, as
sem
ble
-to
-ord
er a
nd
a s
mal
l P
roje
ct b
ud
get
app
roxi
mat
ely
$13
3.5
mill
ion
an
d
Rev
iew
ed i
nte
rnal
do
cum
enta
tion
an
d
pro
por
tion
of
engi
nee
r-to
-ord
er p
roce
sses
pla
nn
ed s
ched
ule
30
mon
ths
vid
eota
pes
of
ER
P p
roje
ct i
ntr
odu
ctio
n
Com
pan
y fo
rmed
th
rou
gh 1
997
mer
ger
of
Th
e E
RP
pro
ject
was
can
celle
d i
n 1
999
aft
er a
n
at l
ead
ersh
ip m
eeti
ng
Jan
uar
y 1
998
two
com
pan
ies
rou
ghly
eq
ual
in
siz
e an
d
exp
end
itu
re o
f $7
0 m
illio
nD
ata
colle
cted
in
US
in
Ju
ne
19
98p
revi
ou
sly
com
pet
ed i
n d
isti
nct
, in
du
stri
al
equ
ipm
ent
nic
hes
; lit
tle
inte
grat
ion
of
the
com
pan
ies
has
occ
urr
ed
Co
mp
any
CE
uro
pea
n s
ub
sid
iary
of
US
mu
ltin
atio
nal
M
ult
isit
e, p
an-E
uro
pea
n r
oll-
out
in p
roce
ss
Tw
o i
nte
rvie
ws
wit
h �
ve k
ey i
nfo
rman
ts
app
arel
co
mp
any
(eig
ht
site
s ov
er 4
yea
rs)
in s
um
mer
199
8M
ake
to s
tock
man
ufa
ctu
rin
gH
ead
qu
arte
rs w
ent
live
wit
h �
nan
ce m
odu
les
Tw
o r
esea
rch
ers
per
form
ing
Fac
ing
dec
linin
g m
arke
t d
eman
d f
or
core
D
ecem
ber
199
7 (
on t
ime)
an
d r
aw m
ater
ials
ap
pro
xim
atel
y 2
7 h
of
on-s
ite
inte
rvie
ws
pro
du
ctm
anag
emen
t in
Ju
ne
199
8 (6
mon
ths
del
ay)
wit
h 1
9 in
form
ants
in
Sep
tem
ber
19
98
Fir
st a
f�lia
te a
nd
sal
es o
f�ce
liv
e in
Ju
ly 1
998
Rev
iew
ed k
ey i
nte
rnal
do
cum
enta
tion
(6 m
onth
s d
elay
)D
ata
colle
cted
in
Bel
giu
m a
nd
Th
e M
ult
isit
e co
n�
gura
tion
: ei
ght
(log
isti
cal
and
�n
anci
al)
Net
her
lan
ds
com
pan
ies
wer
e se
t u
pD
istr
ibu
ted
arc
hit
ectu
re –
sep
arat
e se
rver
at
each
sit
e$5
.5 m
illio
n w
orth
of
mod
i�ca
tio
ns
and
in
terf
aces
(i
ncl
ud
es c
on
sult
ing)
Co
mp
any
D$
330
mil
lion
arm
of
glob
al e
ner
gy a
nd
T
hre
e d
ivis
ion
s ea
ch w
ith
sep
arat
e si
ngl
e-si
te i
nst
ance
T
wo
res
earc
her
s p
erfo
rmin
g en
gin
eeri
ng
�rm
loc
ated
in
Sca
nd
inav
iaof
sam
e E
RP
pac
kage
app
roxi
mat
ely
30
h o
f on
-sit
e in
terv
iew
s O
ne
thou
san
d �
ve h
un
dre
d e
mp
loye
esF
irst
div
isio
n w
ent
live
in J
anu
ary
1996
, o
n t
ime
and
w
ith
17
info
rman
ts f
rom
op
erat
ion
s,
Com
pan
y is
th
e re
sult
of
a 1
996
mer
ger
wit
hin
bu
dge
t�
nan
ce a
nd
IS
an
d o
ne
exte
rnal
IT
of
th
ree
com
pan
ies
pre
serv
ed i
n 1
998
as
Sec
ond
div
isio
n w
ent
live
May
19
98 6
mon
ths
beh
ind
co
nsu
ltan
t in
Sep
tem
ber
199
8th
ree
div
isio
ns
sch
edu
le a
nd
25
% o
ver
bu
dge
t d
ue
to b
ugg
y so
ftw
are
Rev
iew
ed k
ey i
nte
rnal
do
cum
enta
tion
Man
ufa
ctu
res
com
pon
ents
an
d s
yste
ms
that
an
d m
ore
cust
omiz
atio
ns
than
pla
nn
edD
ata
colle
cted
dat
a in
Sca
nd
inav
iase
rve
the
enti
re s
up
ply
ch
ain
of
the
elec
tric
al
Th
ird
div
isio
n w
as i
n t
he
pro
cess
of
imp
lem
enti
ng
pow
er i
nd
ust
ry
Learning from adopters’ experiences with ERP 253
Tab
le 2
con
tin
ued
Com
pan
y C
ompa
ny
des
crip
tion
ER
P i
mp
lem
enta
tion
des
crip
tion
–
Dat
a so
urc
esid
enti
�er
stat
us
at t
ime
of d
ata
colle
ctio
n
Com
pan
y E
Mu
ltin
atio
nal
con
glom
erat
e b
ased
in
th
e U
KIm
ple
men
tati
on b
egan
au
tum
n 1
997
Th
ree
rese
arch
ers
per
form
ing
Th
irty
ER
P p
roje
cts
pla
nn
ed o
ver
5 ye
ar p
erio
dap
pro
xim
atel
y 2
0 h
of
on-s
ite
inte
rvie
ws
wit
h 1
3 in
form
ants
in
Oct
ober
19
98
plu
s 3
h o
f in
terv
iew
s p
rio
r to
sit
e vi
sit
Com
pan
y N
Sm
all
pri
vate
US
com
pan
y m
anu
fact
uri
ng
Sin
gle-
site
im
ple
men
tati
onO
ne
hou
r te
lep
ho
ne
inte
rvie
w w
ith
h
ealt
h c
are
equ
ipm
ent
and
su
pp
lies
Init
ial
imp
lem
enta
tio
n i
n 1
99
3 an
d r
eim
ple
men
tati
on
com
pan
y pr
esid
ent
in M
arch
199
8in
19
97
Com
pan
y O
$1 b
illio
n e
lect
ron
ics
com
pan
y b
ased
in
M
ult
iple
im
ple
men
tati
ons
in d
iffe
ren
t in
tern
atio
nal
O
ne
hou
r te
lep
ho
ne
inte
rvie
w w
ith
th
e U
Ssi
tes
mem
ber
of
wor
ldw
ide
roll
-ou
t te
am i
n
Man
ufa
ctu
rin
g lo
cati
ons
in �
ve c
oun
trie
sF
irst
im
ple
men
tati
on o
f ea
rly
ER
P p
acka
ge i
n
Mar
ch 1
998
Six
th
ousa
nd
em
plo
yees
Ger
man
y th
en c
omp
any-
wid
e ro
ll-ou
tIn
th
e p
roce
ss o
f re
imp
lem
enta
tion
an
d w
orld
wid
e ro
ll-ou
t, s
tart
ing
wit
h m
anu
fact
uri
ng
Com
pan
y P
Jap
anes
e-ow
ned
au
tom
otiv
e su
pp
lier
Pla
nn
ed l
ive
dat
es i
n M
ay 1
998
for
Can
adia
n
On
e h
our
tele
ph
on
e in
terv
iew
wit
h
oper
atin
g in
Nor
th A
mer
ica
oper
atio
ns
and
Au
gust
199
8 f
or m
ain
of�
cem
anag
er o
f co
rpor
ate
serv
ices
an
d
On
e th
ousa
nd
em
plo
yees
in
Nor
th A
mer
ica
All
mod
ule
s to
go
live
at
once
corp
ora
te p
rod
uct
ion
pla
nn
ing
in M
arch
R
egio
nal
of�
ces
in �
ve s
ites
199
8E
xper
ien
cin
g ra
pid
gro
wth
Com
pan
y Q
$40
mill
ion
Can
adia
n e
lect
ron
ics
Bet
a te
st s
ite
for
earl
y E
RP
pac
kage
On
e h
our
tele
ph
on
e in
terv
iew
wit
h
sem
icon
du
cto
r m
anu
fact
ure
rIn
th
e p
roce
ss o
f re
imp
lem
enti
ng
ER
P w
ith
C
IO/I
T p
roje
ct m
anag
er i
n A
pri
l 19
98G
row
ing
at 2
0+%
per
yea
rn
ew s
oftw
are
Hig
h-t
ech
com
pan
y w
ith
str
on
g IT
ex
per
ien
ceC
omp
any
cult
ure
tol
erat
es c
han
ge
reas
onab
ly w
ell
Com
pan
y R
Sm
all
pri
vate
com
pan
y w
ith
pla
nts
in
th
e U
S,
Use
d E
RP
sin
ce 1
993
On
e h
our
tele
ph
on
e in
terv
iew
wit
h C
IO
Can
ada
and
th
e U
KC
urr
entl
y im
plem
enti
ng
ER
P u
pgr
ade
in A
pri
l 1
998
Com
bin
atio
n o
f as
sem
ble
to
ord
er a
nd
re
pet
itiv
e m
anu
fact
uri
ng
Com
pan
y S
$1 b
illio
n a
nd
gro
win
g U
S-b
ased
in
du
stri
al
En
terp
rise
rol
lou
t p
lan
ned
for
199
9O
ne
hou
r te
lep
ho
ne
inte
rvie
w w
ith
eq
uip
men
t m
anu
fact
ure
r w
ith
520
0
Des
crip
tion
of
‘un
it A
’, t
he
�rs
t d
ivis
ion
to
go l
ive
man
ufa
ctu
rin
g sy
stem
s m
anag
er o
f u
nit
em
plo
yees
wo
rld
wid
ew
ith
ER
P:
A i
n A
pri
l 1
998
Fif
ty y
ears
old
; a
who
lly o
wn
ed s
ub
sid
iary
of
Sin
gle-
site
im
ple
men
tati
on w
ith
ten
lic
ence
sa
maj
or i
nd
ust
rial
eq
uip
men
t co
mp
any
Wen
t liv
e in
Nov
emb
er 1
997
Pro
ject
bu
dge
t of
$1.
2 m
illio
nF
irst
rel
ease
im
ple
men
ted
was
‘ve
ry b
ugg
y’ a
nd
no
w
reim
ple
men
tin
g la
ter
rele
ase
Sti
ll h
ave
thre
e m
ajor
inte
rfac
es t
o h
om
egro
wn
sys
tem
sP
lan
to
rep
lace
wit
h E
RP
in
th
e fu
ture
254 Markus et al.
Tab
le 2
con
tin
ued
Com
pan
y C
omp
any
des
crip
tion
ER
P i
mp
lem
enta
tion
des
crip
tion
–
Dat
a so
urc
esid
enti
�er
stat
us
at t
ime
of d
ata
colle
ctio
n
Com
pan
y T
Sm
all
No
rth
Am
eric
an-b
ased
pro
du
cer
of
$1.5
mill
ion
pro
ject
Ele
ctro
nic
in
terv
iew
in
Ap
ril
199
8se
curi
ty s
yste
ms
Ser
ver
up
grad
ed n
um
erou
s ti
mes
du
e to
un
der
sizi
ng
Ap
pro
xim
atel
y 6
00 d
irec
t la
bou
r b
y E
RP
ven
dor
man
ufa
ctu
rin
g em
plo
yees
Du
e to
co
mp
any’
s p
erfo
rman
ce p
rob
lem
s, v
end
or
Fif
teen
to
twen
ty p
er c
ent
ann
ual
gro
wth
rat
e p
erfo
rmed
sp
ecia
l m
od
i�ca
tion
s in
ord
er t
o en
able
ov
er p
revi
ou
s 5
yea
rssp
ecia
l so
ftw
are
rele
ase
Com
pan
y U
Sm
all
Eu
rop
ean
ele
ctro
nic
s as
sem
ble
rF
irst
ER
P i
mp
lem
enta
tion
in
19
90;
log
isti
csO
ne
hou
r te
lep
ho
ne
inte
rvie
w i
n J
un
e T
wen
ty-�
ve t
o th
irty
per
cen
t an
nu
alR
eim
ple
men
ted
cu
rren
t re
leas
e19
98
gro
wth
rat
eF
ive
hu
nd
red
em
plo
yees
Com
pan
y V
$17
5 m
illio
n U
S-b
ased
man
ufa
ctu
rer
Wen
t liv
e on
ER
P e
arly
19
97
On
e an
d a
hal
f ho
ur
tele
ph
one
inte
rvie
w
Pro
du
cts
sold
in
100
+ c
oun
trie
s th
rou
ghE
lap
sed
tim
e fr
om i
nit
ial
pro
ject
con
cep
t to
go-
live
w
ith
IS d
irec
tor
in M
arch
19
98d
istr
ibu
tors
an
d d
eale
rsap
pro
xim
atel
y 2
.5 y
ears
Nin
e h
un
dre
d e
mp
loye
es,
mos
t of
wh
ich
are
O
ne
hu
nd
red
an
d �
fty
or m
ore
use
rs o
f th
e sy
stem
at h
ead
qu
arte
rs/m
anu
fact
uri
ng
faci
litie
sM
any
of t
he
pro
du
cts
are
engi
nee
red
to
ord
er
Com
pan
y X
Lar
ge m
anu
fact
uri
ng
�rm
Ro
ll-ou
t co
mp
lete
d D
ecem
ber
199
7O
ne
hou
r te
lep
ho
ne
inte
rvie
w i
n J
un
e D
etai
ls o
mit
ted
at
com
pan
y re
qu
est
199
8
Com
pan
y Y
Lar
ge U
S m
anu
fact
uri
ng
�rm
in
th
e T
wo
US
loc
atio
ns
and
th
ree
ph
ase
roll
-ou
tO
ne
hou
r te
lep
ho
ne
inte
rvie
w i
n J
un
eae
rosp
ace
ind
ust
ryF
irst
ph
ase
wen
t liv
e M
ay 1
997
199
8O
ther
tw
o p
lan
ned
to
be
com
ple
te b
y R
evie
wed
in
tern
al d
ocu
men
tati
onS
epte
mb
er 1
998
sampling (overselection of companies that had experi-enced or were experiencing dif� culties) and it may notbe a representative � nding. We do believe that somecompanies are successful on all three categories ofsuccess measures.
However, Ross and Vitale (2000) found that aperformance dip after initial implementation of an ERP system is very common. Many of our companiessimilarly experienced moderate to severe business dis-ruption when their ERP systems ‘went live’. They haddif� culty diagnosing problems (which had many possi-ble causes) and they had dif� culty recovering fromthem. They sometimes achieved ‘normal’ operationsonly by permanently increasing staf� ng levels and re-ducing expectations about labour ef� ciency. In general,ERP adopters seemed both physically and psychologi-cally unprepared for shakedown phase dif� culties.
Further, extreme dif� culties in the shakedown phaseappeared to have strong negative in� uences on compa-nies’ willingness to continue with the ERP experience.Several companies with shakedown phase problemsreported strong pressure to de-install their ERP system.Even when the ERP system was retained, there wasgreat unwillingness to upgrade to ‘enhanced’ versionsof the software. In essence, these companies imple-mented ‘legacy’ ERP systems.
Second, mixed ‘success’ results were observed evenwith a single phase. For example, a number of compa-nies achieved their budget and schedule targets, buthad to cut scope, often substantially (companies S, A and T). In the case of company T, these scopereductions led to failure later on: the company did notachieve the business results it had hoped for. However,
company S did achieve its desired business results,despite a massive cut in scope. While company Simplemented only 15% of the ERP functionality it hadoriginally planned to implement, the company claimedto have achieved substantial inventory reductions, asintended. This result shows that it is possible for‘failed’ projects to achieve eventual business success.
We found that companies differed substantially inhow they de� ned success in the project phase becausethey differed in their de� nitions of the project itself.Some companies de� ned the project as ‘implementingERP as quickly and cheaply as possible’. Others de� nedthe project as ‘adopting best practices enabled by ERP’ (which entails business process re-engineering).Still another de� ned the project as ‘achieving com-monality of systems and business practices in a de-centralized organization’ (which entails a process oforganizational development and consensus building).In general, the larger the organization’s de� nition of theproject the more willing the organization was to expandthe project’s budget and schedule. These companieswere less likely to judge the overall ERP experience asunsuccessful when the project budget and schedulewere not met.
We found that larger organizations tended to de� nethe ERP experience in much more expansive termsthan smaller ones. They often demanded businessresults from ‘IT’ projects. In many cases, these orga-nizations were planning for multiple (perhaps dozensof) ERP installations and realized the importance oflearning how to implement and upgrade ERP systemsbetter each time. They were more likely than smallerorganizations to start planning for the onward andupward phase during the project phase.
Third, as Larsen and Myers (1997) observed, somecompanies that achieved ‘success’ in the project phasecould be classi� ed as failures later on. Either they expe-rienced substantial dif� culties during the shakedownphase (companies E and N) or they reported a lackof business bene� ts during the onward and upwardphase (company Q). Similarly, one of the companiesstudied by Dolmetsch et al. (1998) successfully imple-mented SAP R/3 (a particular ERP system) within 4months but was later disappointed not to have achievedbusiness performance improvements because it had notre-engineered its processes.
We were surprised that several companies in theonward and upward phase could not say whether theyhad achieved business bene� ts from using ERP withany con� dence. They gave a variety of related reasonsfor their inability to assess their results.
(1) The ERP system had been adopted for tech-nical reasons (e.g. Year 2000, cost or lack ofcapacity in their current system) and not forbusiness reasons.
Learning from adopters’ experiences with ERP 255
Table 3 Companies studied by stage of ERP experiencecycle
Stage of experience cycle reached Company at time of data collection identi� er
Project phase Company BCompany OCompany P
Shakedown phase Company CCompany ECompany XCompany Y
Onward and upward phase Company ACompany DCompany NCompany QCompany RCompany SCompany TCompany UCompany V
256 Markus et al.
Tab
le 4
Pro
ble
ms
and
ou
tcom
es e
xper
ien
ced
by
ph
ase
Com
pan
ies
by
Pro
ject
ph
ase
pro
ble
ms
and
ou
tcom
esS
hak
edow
n p
has
e p
rob
lem
s an
d o
utc
om
esO
nw
ard
an
d u
pw
ard
ph
ase
pro
ble
ms
and
p
has
e o
f ou
tcom
esex
per
ien
ce c
ycle
at
tim
e of
stu
dy
Com
pan
ies
in p
roje
ct p
has
e
Com
pan
y B
In-p
roce
ss E
RP
pro
ject
hal
ted
an
d r
ech
arte
red
N
/AN
/Aw
hen
com
pan
y m
erge
dE
RP
sys
tem
fac
ed h
uge
org
aniz
atio
nal
in
tegr
atio
n
issu
esT
he
ER
P p
roje
ct w
as c
ance
lled
in
19
99 a
fter
an
ex
pen
dit
ure
of
$7
0 m
illio
n
Com
pan
y O
Had
to
mod
ify
ER
P s
oftw
are
for
esse
nti
al
N/A
N/A
fun
ctio
nal
ity
des
pit
e p
olic
y ag
ain
st i
tF
irst
im
ple
men
tati
on p
artn
er r
epla
ced
for
lac
k o
f re
leva
nt
exp
erie
nce
ER
P p
roje
ct c
om
bin
ed w
ith
com
pan
y-w
ide
stan
dar
diz
atio
n a
nd
re-
engi
nee
rin
g;
big
ch
ange
in
man
agem
ent
issu
es
Com
pan
y P
Had
to
mod
ify
ER
P s
oftw
are
for
esse
nti
al
N/A
N/A
fun
ctio
nal
ity
des
pit
e p
olic
y ag
ain
st i
tP
roje
ct t
eam
had
dif
�cu
lty
gett
ing
invo
lvem
ent
from
lo
cal
site
s
Com
pan
ies
in s
hake
do
wn
ph
ase
Com
pan
y C
Pro
ject
tea
m c
om
mu
nic
ated
wel
l w
ith
E
xper
ien
ced
sys
tem
per
form
ance
pro
blem
sN
/Am
anag
emen
t an
d s
ites
KP
Is d
eter
iora
ted
in
sh
ort
term
Man
agem
ent
cen
tral
ized
for
mer
ly d
ecen
tral
ized
C
ust
omer
s an
d s
up
plie
rs e
xper
ien
ced
neg
ativ
eIS
un
its
for
ER
P i
mp
lem
enta
tion
; lo
cal
site
s ef
fect
sre
sist
edM
anag
ers
not
hap
py
wit
h r
epor
tin
g ca
pab
iliti
esE
xper
ien
ced
co
st a
nd
sch
edu
le o
verr
un
s
Com
pan
y E
On
tim
e an
d w
ith
in b
ud
get
KP
Is d
eter
iora
ted
in
sh
ort
term
N/A
Exp
ecte
d s
cop
e ac
hie
ved
Pro
ject
tea
m d
id e
xcel
len
t jo
b o
f b
uild
ing
con
sen
sus
arou
nd
nee
d f
or c
omm
on s
yste
ms
in
this
dec
entr
aliz
ed c
omp
any
Com
pan
y X
Not
dis
cuss
edE
xper
ien
ced
dif
�cu
ltie
s w
ith
dat
a co
nve
rsio
nN
/AE
xper
ien
ced
sys
tem
per
form
ance
pro
blem
sK
PIs
det
erio
rate
d i
n s
hor
t te
rm
Com
pan
y Y
Bu
dge
t an
d s
ched
ule
ove
rru
ns
Exp
erie
nce
d s
yste
m p
erfo
rman
ce p
robl
ems
N/A
Exp
erie
nce
d s
oftw
are
bu
gsK
PIs
det
erio
rate
d i
n s
hor
t te
rmC
ust
omer
s an
d s
up
plie
rs e
xper
ien
ced
neg
ativ
e ef
fect
s
Learning from adopters’ experiences with ERP 257
Tab
le 4
con
tin
ued
Co
mp
anie
s b
y P
roje
ct p
has
e p
rob
lem
s an
d o
utc
omes
Sh
aked
own
ph
ase
pro
ble
ms
and
ou
tcom
esO
nw
ard
an
d u
pw
ard
ph
ase
pro
ble
ms
and
p
has
e of
ou
tcom
esex
per
ien
ce c
ycle
at
tim
e of
stu
dy
Co
mp
anie
s in
on
war
d a
nd
up
war
d p
has
e
Co
mp
any
AO
n s
ched
ule
an
d w
ith
in b
ud
get
KP
Is d
eter
iora
ted
in
sho
rt t
erm
Tu
rno
ver
of e
xper
ien
ced
use
rs a
nd
su
pp
ort
Sco
pe
cuts
per
son
nel
Use
r sk
ill w
ith
sys
tem
is
low
So
me
imp
rove
men
ts i
n k
ey b
usi
nes
s m
easu
res
Oth
er p
lan
ned
im
pro
vem
ents
not
ach
ieve
d
(ow
ing
to s
cope
cu
ts)
Sys
tem
may
be
part
ially
de-
inst
alle
d
(ow
ing
to u
nan
tici
pat
ed m
erge
r)
Co
mp
any
DO
ne
div
isio
n o
n t
ime
and
wit
hin
bu
dge
tE
xper
ien
ced
hea
vy d
ata
entr
y er
rors
by
use
rsU
ser
skill
wit
h s
yste
m r
emai
ns
low
an
d
A s
econ
d d
ivis
ion
exp
erie
nce
d s
ched
ule
del
ays
Had
to
incr
ease
sta
ff t
o co
pe
wit
h e
rror
ser
rors
rem
ain
hig
hd
ue
to s
oftw
are
mod
i�ca
tion
sK
PIs
det
erio
rate
d i
n s
hort
ter
mS
yste
m n
ot u
sed
in
man
ager
ial
dec
isio
n
mak
ing
Insu
f�ci
ent
pla
ns
for
on
goin
g sy
stem
su
pp
ort
and
bu
sin
ess
impr
ove
men
t
Co
mp
any
NA
ccep
tab
le p
roje
ct o
utc
om
es d
esp
ite
hea
vy
Dis
astr
ous
per
form
ance
pro
ble
ms
Nev
er a
chie
ved
nor
mal
op
erat
ion
scu
stom
izat
ion
sS
ever
e b
usi
nes
s d
isru
pti
on
Exp
erie
nce
d p
erm
anen
t lo
ss o
f b
usi
nes
sS
ever
e n
egat
ive
imp
act
on c
ust
omer
s an
d
sup
plie
rs
Co
mp
any
QA
ccep
tab
le d
espi
te e
nti
rely
in
-hou
se
Use
rs c
hal
len
ged
by
con
vers
ion
to
clie
nt-
serv
er
Bu
sin
ess
imp
rove
men
ts w
ere
no
t so
ugh
t as
im
ple
men
tati
on w
ith
no
pri
or e
xper
ien
ceen
viro
nm
ent
par
t of
ER
P i
mp
lem
enta
tion
Sev
ere
syst
em p
erfo
rman
ce p
rob
lem
s
Co
mp
any
RE
xper
ien
ced
po
or c
onsu
ltin
g ad
vice
Exp
erie
nce
d m
any
soft
war
e er
rors
du
e to
lac
k of
B
usi
nes
s im
pro
vem
ents
wer
e n
ot
sou
ght
as
Mad
e ex
cess
ive
soft
war
e m
odi�
cati
ons
inte
grat
ed s
yste
m t
esti
ng
par
t of
ER
P i
mp
lem
enta
tion
Exp
erie
nce
d d
if�
cult
y re
cove
rin
g fr
om e
rror
s
Co
mp
any
SW
ith
in b
ud
get
Not
dis
cuss
edP
lan
ned
bu
sin
ess
ben
e�ts
ach
ieve
dG
reat
ly b
ehin
d s
ched
ule
Gre
atly
red
uce
d s
cop
e
258 Markus et al.
Tab
le 4
con
tin
ued
Com
pan
ies
by
Pro
ject
ph
ase
pro
ble
ms
and
ou
tco
mes
Sha
ked
ow
n p
has
e p
rob
lem
s an
d o
utc
omes
On
war
d a
nd
up
war
d p
has
e p
rob
lem
s an
d
ph
ase
of
outc
omes
exp
erie
nce
cyc
le
at t
ime
of s
tud
y
Com
pan
ies
in o
nw
ard
an
d u
pw
ard
ph
ase
Com
pan
y T
Wit
hin
bu
dge
tN
ot d
iscu
ssed
Hea
d c
oun
t in
crea
sed
in
stea
d o
f d
ecre
ased
S
ched
ule
ove
rru
ns
as p
lan
ned
Sco
pe
cuts
Imp
rove
d r
epo
rtin
g an
d d
ata
acce
ssS
ome
pla
nn
ed b
usi
nes
s im
pro
vem
ents
not
ac
hie
ved
Com
pan
y U
Pro
ject
tea
m t
ook
exce
ssiv
ely
fun
ctio
nal
E
xten
sive
rec
on�
gura
tion
occ
urr
ed a
s a
resu
lt o
f P
oor
dat
a q
ual
ity
con
tin
ues
to
ham
per
ap
pro
ach
to
imp
lem
enta
tion
lear
nin
g ab
out
inte
grat
ed o
per
atio
ns
bu
sin
ess
Man
agem
ent
sup
por
t ga
ined
qu
ite
late
K
PIs
im
pro
ved
aft
er r
eco
n�
gura
tion
Com
pan
y h
as g
ain
ed i
mp
rove
d a
war
enes
s in
pro
cess
of b
ene�
ts o
f cr
oss-
fun
ctio
nal
in
tegr
atio
nM
anag
emen
t is
ext
rem
ely
sati
s�ed
wit
h
syst
emU
pgr
ade
to n
ew r
elea
se i
s p
lan
ned
Com
pan
y V
Ign
ored
ad
vice
of
exp
erie
nce
d i
mp
lem
enta
tion
E
xtre
me
bu
sin
ess
dis
rup
tion
ow
ing
to
Per
man
ent
loss
of
cust
omer
sco
nsu
ltan
tsco
n�
gura
tio
n e
rror
s, s
yste
m i
nte
grat
ion
pro
ble
ms
Bu
sin
ess
pro
cess
es s
trea
mlin
edD
id n
ot f
ollo
w c
onve
nti
onal
im
ple
men
tati
onan
d i
nco
mp
lete
/inac
cura
te d
ata
Com
pan
y is
now
rea
dy
to u
nd
erta
ke
met
hod
olog
yK
PIs
det
erio
rate
d i
n s
ho
rt t
erm
pro
cess
re-
engi
nee
rin
gE
xces
sive
pro
ject
man
ager
tu
rnov
er (
�ve
tim
es)
Cu
stom
ers
and
su
pp
liers
exp
erie
nce
d n
egat
ive
effe
cts
(2) No business goals had been set for the ERPproject.
(3) The company did not manage by metrics.(4) Existing systems did not allow the company to
measure where it was on key business metricsprior to the implementation of the ERP system.
(5) The company did not perform a post-imple-mentation audit of the ERP project in order toassess whether projected bene� ts were achieved.
In general, companies that do not deliberately setout to achieve measurable business results do notobtain them (or do not realize that they have obtainedthem). Further, the inability to document measuredbene� ts from an ERP implementation appears todiscourage organizations from undertaking futureupgrades and/or migrations.
In conclusion, success in the ERP experience is multi-dimensional and often hard to measure. Early success(or success on project measures) is not closely linkedwith later success (success on business measures) andearly failure (failure on project measures) is not tightlylinked with later failure (failure of business measures).Clearly then, success in an ERP experience is not pre-determined by a set of success factors in place at the start of the project and continuing unchanged through-out. Either conditions change over the course of theexperience or different types of actions are required atdifferent phases and the ways in which a companyresponds to conditions at each phase in� uence thesubsequent progress and ultimate success of the ERPexperience. This observation suggests one obviousnormative recommendation: companies should be con-cerned with success in all phases of the ERP experienceand should not concern themselves exclusively withwhat happens during the project phase. In addition, thisobservation suggests one obvious research issue: Inorder to understand the success of an ERP experience,one needs to look at what goes on (e.g. problems expe-rienced and attempts at problem resolution) at eachphase of the experience cycle. In the next section, wefocus more deeply on why the companies we studiedachieved suboptimal success.
Findings about adopters’ problems with ERP
We asked adopters what problems they experiencedwith ERP systems, how they had dealt successfullywith these problems (if they had) and what they hadlearned as a result of their experience. We also formedour own impressions of their experiences based onwhat we observed when we visited their companies.We came away with a deep respect for the challengesthey faced. If what they were trying to do was easy,more of them would have been successful on allmeasures. However, many of the problems they
experienced were ‘wicked’, that is hard to recognizeand diagnose due to multiple interacting causes andvarying symptoms and effects. In this section, wedescribe what we believe to be the most dif� cult prob-lems adopters experienced and why the problemsoccurred.
Project phase problems
The most challenging project phase problems reportedby our respondents involved software modi� cations,system integration, product and implementationconsultants and turnover of project personnel.
Software modi� cations Almost every analyst of the ERPexperience strongly advises companies to avoid modify-ing the software. Companies are advised to live withexisting ERP functionality and to change their pro-cedures to adapt to it. However, we found the following.
(1) Many adopters could not avoid some degree ofERP software modi� cation. In some cases, ERPpackages are selected on a centralized basis inorder to � t the majority of corporate needs.Often, there are a few sites that cannot operateeffectively with the software’s functionality, evenif people there are in principle willing to modifytheir business processes. For example, onecompany reported having an order of magni-tude more entities (e.g. sales representatives)than were allowed by the relevant � eld size in the software package. Other companiesexplained that the software simply did not � tbusiness rules around commissions and royal-ties and that these rules could not be changedwithout serious negative business implications.
(2) Many adopters had dif� culty in getting modi� -cations to work well. They complained aboutimplementation consultants who did not deliverwell-tested and working modi� cations in atimely manner.
(3) Most distressingly, several adopters reportedthat, after wrestling with modi� cations (andsometimes failing to make them work well), theyeventually learned that their modi� cations wereunnecessary after all. They had usually madeplans for software modi� cations early in theproject phase when they did not understand thesoftware thoroughly (in particular the integra-tions across modules). Later on when theyunderstood the software better, they discoveredways of implementing the needed capabilitieswithout modi� cations.
For a more general treatment of the issues involved intailoring ERP software to a company’s speci� c needs seeL. Brehm, A. Heinzl and M. L. Markus (forthcoming).
Learning from adopters’ experiences with ERP 259
Problems with system integration ERP systems are soldas ‘integrated packages’, implying that they containeverything one needs and that ERP software con� gura-tion (plus tailoring) is the major activity of the projectphase. However, there are a number of respects inwhich this is not so.
(1) First, an ERP system needs to be integrated withthe computing platform on which it will run. Wefound that companies had great dif� cultyintegrating their enterprise software with a pack-age of hardware, operating systems, databasemanagement systems software and telecom-munications systems suited to their particularorganization size, structure and geographic dis-persion. They reported having dif� culty � ndingexperts who could advise them on the preciseoperating requirements of their ERP con� gura-tion. They described having made unplannedupgrades of processors and memory to supporttheir systems. One company reported makingseveral changes of database management systembefore � nding one that ‘worked’.
(2) Second, for all that ERP systems are said to becomprehensive packages that cover every orga-nizational function, most of the companies westudied (large and small) reported needing toretain some legacy systems that performedspecialized functions not available in ERP pack-ages. (Alternatively, they acquired specializedsoftware from third parties.) These systemsneeded to be interfaced with ERP systems – a process, that is both challenging and expen-sive.
(3) A particular area in which many organizationsfound ERP systems de� cient was that of datareporting. ERP systems are essentially transac-tion processing systems that do not (withoutexpensive add ons) solve companies’ needs fordecision support. For descriptions of themeasures companies must often take to solvetheir ERP-related data reporting problems, seethe cases of Microsoft (Bashein et al., 1997) andMSC Software (Bashein and Markus, 2000).
Problems with product and implementation consultantsERP implementations are socially complex activities. As many as a dozen or more external companies –including the ERP vendor, vendors of ERP productextensions, vendors of supporting hardware, softwareand telecommunications capabilities, implementationconsultants and so forth – may be involved in different aspects of an organization’s ERP experience.Coordinating the efforts of all these � rms is, to put itmildly, a challenge. We found the following.
(1) Few IT products and services � rms were willingto take end-to-end responsibility for coordi-nating all parties. In addition, adopters wereoften rightly reluctant to cede authority forproject management to an outside party, evenwhen they were willing to pay the steep fees foroutside assistance.
(2) IT products and services � rms generally seemto resent taking subordinate roles to other such� rms. They do not to cooperate well. There ismuch � nger pointing when problems occur.
(3) Despite representations during the sales cycle,there was widespread lack of knowledge about the details of ERP products, particularly whereintegrations, tools and interfaces with ‘partner’products were concerned.
(4) Because IT products and services � rms aregrowing rapidly, they � nd it dif� cult to providecontinuity in personnel assigned to adopterprojects and adopters strongly value continuityin personnel.
(5) Several adopters reported having had con� icts(sometimes severe) with IT products andservices vendors over contractual provisions(e.g. pricing and billing arrangements) andproject direction (e.g. project management).
Turnover of project personnel An all too common com-plaint was the frequency with which adopters lose key personnel experienced with ERP or supportingtechnologies. As already noted, external serviceproviders themselves are unable to maintain continuityof customer support personnel. In addition, adoptersfrequently reported the following.
(1) Losing key IT specialists and user representa-tives working on the project while the projectwas going on, often despite handsome retentionbonuses.
(2) Losing experienced people after the project wascomplete. Many IT specialists thrive on projectwork and view assignment to a ‘competencecentre’ (support unit) as unpleasant mainte-nance work.
In short, the project phase of the ERP life-cycle posedsevere challenges for the adopters we studied and notall companies resolved these problems well. In somecases, unresolved issues ‘left over’ from the projectphase became the source of problematic outcomes laterin the shakedown phase.
Unfortunately, it was also the case that companiesexperienced problems that had originated in the projectphase, but which were not perceived as problems orrecti� ed at that time, during the shakedown phase.Although these problems are more rightly classi� ed by
260 Markus et al.
their origins as project phase problems, we list thembelow as shakedown phase problems because that iswhere their symptoms show up.
Shakedown phase problems
As mentioned earlier during the discussion on‘success’, many of our companies experienced nega-tive outcomes during the shakedown phase. Amongthe outcomes experienced were the following.
(1) Performance problems with the ERP system(and underlying IT infrastructure).
(2) A slow down in business processes.(3) Errors made by users entering data into the
system.(4) Increased staf� ng required to cope with slow
downs and errors.(5) A drop in the company’s key performance indi-
cators.(6) Negative impacts on customers and suppliers
from an inability to answer their queries andfrom delayed shipments and payments.
(7) A need for manual procedures for addressinglack of functionality in ERP software.
(8) Data quality problems.(9) Inadequate management reporting.
This list is an uncomfortable mélange of symptoms ofleftover problems (e.g. performance problems with thesystem and slow down in processes), attempts toresolve problems (e.g. manual processes, workaroundsand increased staf� ng) that create new problems intheir turn and true outcomes – consequences of prob-lems (e.g. negative impacts on customers). Theseelements are dif� cult to disentangle analytically.However, after detailed examination, we concludedthat many shakedown phase dif� culties were causedby problems that occurred during the project phasebut were not recognized as problems or successfullyresolved at the time they occurred. The most impor-tant of these problems were approaching ERP imple-mentations from an excessively functional perspective,inappropriately cutting project scope, cutting end-usertraining, inadequate testing, not � rst improvingbusiness processes and underestimating data qualityproblems and reporting needs.
Approaching ERP implementations from an excessivelyfunctional perspective Cross-functional integration isstill a new concept to many organizations. It is far more natural for them to approach implementing ERPon a module-by-module basis and to assume that ERP modules correspond to traditional functionaldepartments in the organization (e.g. accounting,manufacturing and sales). Con� guration errors oftenfollow when adopters set up project teams without
appropriate cross-functional representation. For anexample, see Koh et al. (2000).
Inappropriately cutting project scope Knowledgeableproject managers know that exceeding the projectschedule is the major threat to project success (more soeven than budget overruns). Therefore, cutting scope isa common tactic when the project shows signs of miss-ing key milestones. Project managers are often temptedto cut scope according to what looks hardest to do;those who stay focused on ‘what is the minimum func-tionality we can implement in order to obtain thedesired business bene� ts?’ are more successful. As men-tioned earlier, several of the companies we studied cutscope when the schedule and budget ran short. Thesedeletions often made it necessary for users to adoptinef� cient manual processes in the shakedown phase.
Cutting end-user training Schedule pressures affecttraining as well as scope because end-user training istypically one of the last activities to occur in the pro-ject. Adopters frequently reported having underesti-mated the needs for end-user training. In particular,they told us that users needed additional training andeducation in non-ERP areas.
(1) Making the transition from ‘green screen’(mainframe software) to ‘client-server’ (PC-based software). Surprisingly, this was a majorhurdle in several adopter organizations.
(2) Understanding ERP and MRP (material require-ments planning) concepts. Some adoptersbelieved it necessary to conduct extensive APICS(a professional institution) education to accom-pany ERP training.
(3) Understanding cross-functional business pro-cesses. In many organizations, people understandwhat they do, but not how their work affectsothers. In the ERP setting, such a limited worldview leads to errors and misunderstandings.
(4) Recovering from data entry mistakes. BecauseERP systems are integrated, data entry errorshave many more rami� cations than do errors intraditional systems and they are much harderto correct. Adopters reported suffering from lackof training activities that addressed recoveryfrom data entry problems.
In some companies, training was not budgeted aspart of the ERP project itself, but was left to thebudgets and discretion of operating managers. Thismanagement policy increased the likelihood of inade-quate end-user training.
Inadequate testing, particularly of interfaces, modi� cations,integrations and exceptions Like scope and training,
Learning from adopters’ experiences with ERP 261
testing is often cut when the project schedule gets tight.Further, because many adopters lack extensive experi-ence with integrated software and with cross-functionalteaming, they are likely to overlook the need for con-ducting system (as opposed to module) tests. Areaswhere testing is most likely to be de� cient include ERPcross-module integrations, interfaces with legacy sys-tems, modi� cations, particularly those performed byexternal � rms (adopters often assume that externalprovider work is properly tested) and unusual businessscenarios and scenarios involving the input of erroneousdata. Several adopters told us that (they realized afterthe fact) they had not adequately tested their ERP soft-ware.
Not � rst improving business processes where this needs doingAdopters naturally want faster implementations andone of the best ways of shortening implementationschedules is to ‘implement the software � rst and re-engineer the business processes later’. This is greatadvice when adopters have reasonably sound businessprocesses to start with. However, some adopters do not.Some companies have found that failure to change theirbusiness processes leads to the following.
(1) Inappropriate software modi� cations. Onecompany we studied tried to implement ERPwithout changing either the software or its busi-ness practices. In the end, the company changedboth unnecessarily. The software modi� cationscould have been avoided through upfrontbusiness process improvements.
(2) Severe disappointment with ERP whenmanagers realized that getting business bene� tsfrom ERP required change in business practices(Dolmetsch et al., 1998).
Underestimating data quality problems and reporting needsOur review of a few companies’ detailed project plans revealed severe underestimation (even theirimplementation consultants missed this problem!) ofthe project tasks associated with data. In the early daysof a project, it is of course hard to know how manyand which legacy systems will have to be retained.However, even when the ERP system replaces alllegacy systems, data problems can be severe.
(1) Due to the nature of their businesses, adoptersmay need to retain legacy data for many years(e.g. for regulatory compliance or because theirproducts remain in service for many years).
(2) Adopters often underestimate the poor qualityof the existing business records that will be inputto ERP. Knowledgeable end-users often substi-tute for high-quality data in traditional systems:
they know what the numbers really are.However, because ERP systems are integrated,the data must be cleaner. Bad data may auto-matically trigger processes in distant areas wherepeople lack the knowledge to override thesystem.
(3) Most large adopter organizations have extensiveand complex data reporting needs. While theseneeds are best addressed with technologies otherthan ERP, adopters often believe that ERP willsatisfy them. Therefore, ERP project plans oftenneglect reporting issues and some adoptersbecome very disappointed with ERP systemsbecause their reporting needs were not well met.
In fact, our biggest surprise about the shakedown phasewas that, in the adopter’s eyes, high-quality data andgood reporting are absolutely essential for ERP success.End-users and line managers are unwilling to trust anduse systems if they do not trust the data and reports.Lack of user acceptance of data and reporting can leadto de-installation of the system or unwillingness toinvest in further upgrades. Note that achieving accep-tance of a common source of data is often a highlypolitical process, particularly in large, complex orga-nizations. However, if these politics are not wellmanaged during the project phase the success of theentire experience is at risk.
In short, the shakedown phase reveals the un-resolved or unrecognized problems of the projectphase. Many negative shakedown experiences could be avoided by giving adequate attention to cross-functional con� guration and testing of ERP software,end-user training, data conversion and management of legacy data, reporting needs and scenarios forrecovering from data input errors during the projectphase.
At the same time, steps taken during the shakedownphase for remedying these problems or their symptomsmay fail to solve the problems and may actually makematters worse. For example, we found that, becauseend-user training was inadequate and users did notunderstand how to back out erroneous transactions,companies often began to rely heavily on ‘key users’(project personnel) and IT staff to perform routinework that should have been done by users. As a result, the key users did not have time to conductbetter end-user training and IT staff did not have timeto work out platform problems and upgrades. Thesecompanies later found themselves extremely vulnerablewhen key users and IT staff began leaving for better paying jobs elsewhere. Similar observationsabout the persistence and negative consequences of‘workarounds’ have been made by Tyre and Orlikowski(1994).
262 Markus et al.
Onward and upward phase problems
Different problems characterized the onward andupward phase. As with the shakedown phase problems,problems appearing during the onward and upwardphase often had much earlier roots. The most impor-tant problems we observed in the onward and upwardphase are as given in the following subsections.
Unknown business results Many adopters who had beenusing ERP long enough to have business results didnot know whether they had realized improvements. Inmost cases, these companies had viewed ERP strictlyas a technology replacement decision and had notprepared business cases justifying ERP in terms ofbusiness bene� ts.
Disappointing business results Some adopters in theonward and upward phase reported that their businessresults had not been achieved. In some cases, theabsence of business results could be traced to inappro-priate scope-cutting decisions during the project phase. In other cases, the organization did not have aculture of managing the results, did not collect and usemetrics, did not demand business improvements and so forth. The lesson is clear: ERP bene� ts are notautomatic. They require human and organizationallearning, both of which take time and require focusedmanagement attention.
Fragile human capital Many adopters were not in astrong position to go forward with ERP because of thefragile state of their ERP human capital. Many organ-izations had lost and had dif� culty replacing ERPknowledgeable IT specialists and end-users. In someorganizations, the only end-users who were ERP knowl-edgeable were those who participated on the projectteam. In addition, we saw IS specialists routinely doingwork that belonged in end-user job responsibilities.This is a precarious situation for adopters. Not onlymay they fail to realize full business bene� ts from ERP,but they may also be unable to recover gracefully fromfuture problems. Further, they may not be able to makefuture technology upgrades and business improvementswithout outside help.
Migration problems We spoke to several adopters whowere on their second round of ERP implementation.Most reported having learned how poorly softwaremodi� cations convert during implementation of laterreleases. In some cases, this was seen as a positive learn-ing experience, because the organizations vowed neverto modify the ERP software again but to make essen-tial changes to their business processes. However, we suspect that most companies that have dif� culties in upgrading will simply stop enhancing their ERP
systems. These organizations will in effect haveimplemented legacy ERP systems, obviating one of themajor bene� ts of using packaged software – the abilityto outsource the ongoing maintenance and enhance-ment of software to a vendor (Brehm and Markus,2000).
In short, the onward and upward phase reveals theunresolved or unrecognized problems of earlier phases.In some cases, onward and upward phase problemscould have been avoided by taking action during theproject phase:
(1) Doing a much better job of end-user trainingduring the project phase.
(2) Starting the project phase with plans for long-term maintenance and migration.
(3) Documenting the reasons for con� gurationdecisions, not just the parameters, so that peoplenot involved in the project phase can get up tospeed quickly.
(4) Not disbanding the project team when theproject goes live, but instead staf� ng a compe-tence centre for managing future evolution andlearning.
In other cases, however, preventing and resolvingonward and upward phase problems must occur wellbefore the project phase even begins. Markus andTanis (2000) discussed the importance of what we callthe chartering phase, which is often unacknowledgedin less successful ERP adoptions, in which key busi-ness decisions related to the ERP system are made. In many cases, only senior executives (not projectmanagers and team members) can address pre-existingorganizational challenges that threaten ERP success.Among such challenges are the following, which wereobserved in several of our study companies.
(1) Lack of results orientation in the business is akey factor in failure to achieve business results.This is not something that an ERP project teamcan � x.
(2) A culture resistant to change is another bigimpediment to ERP success. Project teams candesign and execute change managementprogrammes, but senior executives must workto make these efforts a success.
(3) When top managers do not buy in to the goalsand plans of the ERP project team, the chancesfor success are weak. Good project managerscan contribute to buy-in by good and frequentcommunication, but again success requires aconcerted effort at the top, before and duringthe project.
Waiting to resolve these problems until the symp-toms � rst appear – often as late as the onward and
Learning from adopters’ experiences with ERP 263
upward phase – can be a recipe for failure. Remedialactions taken late in the experience often fail to solvethe problems. The more likely outcome when problemresolution is delayed is termination of the system.
Conclusions and suggestions for futureresearch
The implementation of ERP systems in organizationsis an enormously complex undertaking. ERP systemscan affect nearly every aspect of organizational perfor-mance and functioning and measures of ERP systemssuccess must re� ect this fact. Our � ndings show thatdifferent measures of success are appropriate atdifferent points in the ERP experience cycle and thatthe outcomes measured at one point in time are onlyloosely related to outcomes measured later. This occursbecause the experience cycle is a process (or really aset of processes) and not a mechanical connectionbetween starting conditions and � nal results. Over thecourse of this process several things can happen toin� uence the � nal outcomes observed: starting condi-tions can change, problems can arise (which may ormay not be recognized) and steps can be taken toaddress them (which may or may not be successful,possibly creating new problems in their wake).
In short, the connections between starting condi-tions, experienced problems and outcomes in the ERPexperience are not deterministic. While this can be con-strued as bad news for academic theory, it is good newsfor both ERP adopters and for IS researchers. For ERPadopters it means that it is possible to succeed with ERPdespite bad luck, some mistakes and even early failures.For researchers it means that there is much more workto be done in order to understand problem recognitionand resolution behaviours and how they interact toresult in successful and unsuccessful outcomes.
One particular area that deserves much futureresearch is what we have called the chartering phase –this was often unacknowledged and unful� lled in the organizations we studied. In this phase, whichshould occur before a ‘project’ is ‘chartered’ (hence thename), senior executives in consultation with othersmake important business decisions about the objectivesof the project, the decomposition of the project intomanageable chunks, the level of budget to be allocatedto the project and shakedown phases of each chunk, anappropriate project leader and/or implementation part-ner and so forth. Further research is needed on howcompanies actually make or avoid making these deci-sions, what factors they consider and those that they donot, whom they consult and follow and the speci� cimplications of these decisions for the problems andoutcomes experienced later in the experience cycle.
Acknowledgements
We gratefully acknowledge the support of GordonMosinho for this research. Andrew Martin and ChrisSauer provided helpful comments.
References
Ballantine, J., Bonner, M., Levy, M., Martin, A., Munro, I.and Powell, P.L. (1996) The 3-D model of informationsystems success: the search for the dependent variablecontinues. Information Resources Management Journal, 9(4), 5–14.
Bashein, B.J. and Markus, M.L. (2000) Data Warehouses:More Than Just Mining (Financial Executives ResearchFoundation, Morristown, NJ).
Bashein, B.J., Markus, M.L. and Finley, J.B. (1997) SafetyNets: Secrets of Effective Information Technology Controls(Financial Executives Research Foundation Inc.,Morristown, NJ).
Brehm, L. and Markus, M.L. (2000) The divided softwarelife cycle of ERP packages. In Proceedings of the FirstGlobal Information Technology Management (GITM)World Conference.
Brehm, L., Heinzl, A. and Markus, M.L. (forthcoming)Tailoring ERP systems: a spectrum of choices and theirimplications. Proceeding of the 34th Annual HawaiiInternational Conference on System Sciences, 3–6 January,Maui, Hawaii.
Bulkeley, W.M. (1996) A cautionary network tale: Fox-Meyer’s high-tech gamble. Wall Street Journal InteractiveEdition.
Davenport, T.H. (1998) Putting the enterprise into theenterprise system. Harvard Business Review, 76(4),121–31.
deLone, W.H. and McLean, E.R. (1992) Informationsystems success: the quest for the dependent variable.Information Systems Research, 3(1), 60–95.
Dolmetsch, R., Huber, T., Fleisch, E. and Osterle, H. (1998)Accelerated SAP: 4 Case Studies (Institute for InformationManagement, University of St Gallen School forAdministration, Economics, Law, and Social Sciences).
Koh, C., Soh, C. and Markus, M. L. (2000) Process theoryapproach to ERP implementation and impacts: the caseof Revel Asia. Journal of Information Technology Casesand Applications, 2(1), 4–23.
Larsen, M.A. and Myers, M.D. (1997) BPR success orfailure? A business process reengineering model in the� nancial services industry. In Proceedings of theInternational Conference on Information Systems, pp.367–82.
Lyytinen, K. and Hirschheim, R. (1987) Information systemsfailures – a survey and classi� cation of the empiricalliterature. In Oxford Surveys in Information Technology,Vol. 4, Zorkoczy, P. I. (ed.) (Oxford University Press,Oxford), pp. 257–309.
Markus, M.L. and Tanis, C. (2000) The enterprise systemsexperience – from adoption to success. In Framing theDomains of IT Research: Glimpsing the Future Through the
264 Markus et al.
Past, Zmud, R.W. (ed.) (Pinna� ex EducationalResources, Cincinnati, OH), 173–207.
Peters, T.J. and Waterman, R.H. (1982) In Search ofExcellence: Lessons From America’s Best-run Companies(Harper & Row, New York).
Ross, J.W. and Vitale, M. (2000) The ERP revolution:surviving versus thriving. Information Systems Frontiers,in press.
Sauer, C. (1993) Why Information Systems Fail: A Case StudyApproach (McGraw-Hill, London).
Tyre, M.J. and Orlikowski, W.J. (1994) Windows of oppor-tunity: temporal patterns of technological adaptation inorganizations. Organization Science, 5(1), 98–118.
Yin, R.K. (1994) Case Study Research: Design and Methods,(Sage Publication, Thousand Oaks, CA).
M. Lynne Markus is Professor (Chair) of ElectronicBusiness at City University of Hong Kong, on leave fromthe Peter F. Drucker Graduate School of Management,Claremont Graduate University. Professor Markus hasover 20 years of experience researching information tech-nology in organizations. Her research on enterprise sys-tems has been funded by the National ScienceFoundation, the Financial Executives ResearchFoundation, SIM International, and Baan Research.
Sheryl Axline is a Ph.D. candidate at the School ofBehavioral and Organizational Sciences, ClaremontGraduate University. She has studied team and orga-nizational learning issues around ERP projects since
1997. She has practical experience in human resourcesand career development. Sheryl’s dissertation researchdeals with team learning organizational memory, andinformation technology.
David Petrie is a Ph.D. student at the School ofInformation Science, Claremont Graduate University.Petrie has studied the business value and IT architec-tural issues around ERP implementations since 1997.He has 20 years of practical IS experience with anemphasis on data warehousing and database mar-keting. David’s dissertation research concerns howcompanies deal with technological discontinuities suchas that created by the Internet, web-hosted software,and business-to-business e-commerce. He teaches atthe University of Redlands.
Cornelius Tanis is a consultant with Coach & Commit-ments, based in Utrecht, The Netherlands. He formerlyworked with Key Performance International and wasResearch Program Director with Baan Research.
Address for correspondence:M. Lynne Markus,Department of Information Systems,City University of Hong Kong,83 Tat Chee Avenue,Kowloon, Hong Kong, PR China,e-mail: [email protected]
