American Marketing Association

Return on Quality (ROQ): Making Service Quality Financially Accountable

Author(s): Roland T. Rust, Anthony J. Zahorik and Timothy L. Keiningham
Source: Journal of Marketing, Vol. 59, No. 2 (Apr., 1995), pp. 58-70
Published by: American Marketing Association
Stable URL: http://www.jstor.org/stable/1252073
Accessed: 25-10-2015 11:47 UTC

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at http://www.jstor.org/page/
info/about/policies/terms.jsp

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content
in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship.
For more information about JSTOR, please contact [email protected].

American Marketing Association is collaborating with JSTOR to digitize, preserve and extend access to Journal of Marketing.

http://www.jstor.org

This content downloaded from 128.6.218.72 on Sun, 25 Oct 2015 11:47:44 UTC
All use subject to JSTOR Terms and Conditions
 RolandT. Rust, AnthonyJ. Zahorik,&TimothyL. Keiningham

Return on Quality (ROQ): Making

Service Quality Financially
Accountable
Many companies have been disappointed by a lack of results from their quality efforts. The financial benefits of qual-
ity, which had been assumed as a matter of faith in the "religion of quality,"are now being seriously questioned by
cost-cutting executives, who cite the highly publicized financial failures of some companies prominent in the quali-
ty movement. In this increasingly results-oriented environment, managers must now justify their quality improve-
ment efforts financially. The authors present the "returnon quality"approach, which is based on the assumptions
that (1) quality is an investment, (2) quality efforts must be financially accountable, (3) it is possible to spend too
much on quality, and (4) not all quality expenditures are equally valid. The authors then provide a managerial frame-
work that can be used to guide quality improvement efforts. This framework has several attractive features, includ-
ing ensured managerial relevance and financial accountability.

T he quality revolution has taken over the thinking of changes to imitate successful quality-driven companies
much of American industry (Dean and Evans 1994). withoutgraspingthe need to change the fundamentalcultur-
American manufacturers,hard pressed by the stan- al underpinningsof the firm, with disastrousresults (Grant,
dards of precision and durability of the products of their Shani, and Krishnan1994). And firms that have been laud-
Japanese counterparts,have adopted "quality"as their new ed for their quality orientationhave run into financial diffi-
idol with a near-religiousfervor(Greising 1994). The "qual- culties, in partbecause they spent too lavishly on customer
ity revolution"has spread through manufacturingand ser- service.
vice industries as customers have begun to demand more For example, the Wallace Company won the Malcolm
and better levels of performance,and managershave found BaldrigeNationalQualityAwardin 1990. Howeverthe high
that growing, highly competitive marketsrequire that cus- levels of spending on quality that enabled them to win the
tomers must be satisfied by their purchasesor they will go Baldrige also produced unsustainable losses,1 and within
elsewhere (Rice 1990). Signs of this fervorare evident in the two years they were bankrupt(Hill 1993). Similarly,Florida
managementsection of any bookstore. The late 1980s and Power & Light spent millions to compete for Japan'spresti-
early 1990s have seen an explosive growthin the numberof gious Deming Prize (Wiesendanger1993). Inattentionto ris-
titles espousing the way of quality-from anecdotes about ing costs caused a backlash by rate payers, resulting in its
how quality has saved companies to managementguides on quality programbeing dismantled(Training1991).
how to instill quality principles in one's own organization From the experiences of these companies, and common
(e.g., Deming 1986; Heskett, Sasser, and Hart 1990; Juran sense, it is clear that there are diminishingreturnsto expen-
and Gryna 1980). ditureson quality.Improvingquality helps up to a point, but
However, the quality revolution is not without its casu- past that point furtherexpenditureson quality are unprof-
alties. For example, the fervor for quality circles as a itable. Of course, many quality improvementsresult in a re-
panacea for managementproblems has died with the real- duction in costs that more than makes up for the quality ex-
ization that they are not suited to all cultures (Arnold and penditures (Bohan and Homey 1991; Carr 1992; Crosby
Plas 1993). Many firms have attemptedto adopt cosmetic 1979; Deming 1986). However, such improvements are
more prevalentin manufacturingand the more standardized
services (e.g., fast food restaurants)than they are in the
RolandT. Rustis Professorof Marketing and Directorof the Centerfor
Services Marketing, and AnthonyJ. Zahorikis an AssistantProfessor, highly customized, big-ticket services that constitute the
OwenGraduateSchoolof Management, Vanderbilt TimothyL.
University. growthindustriesof the informationage (e.g., electronic in-
Keiningham is CEO,Copernican Systems,Inc.TheauthorsthankVander- formation services) (Fornell, Huff, and Anderson 1994).
bilt'sCenterforServicesMarketingandthe Dean'sFundforResearchof This is because customization inhibits economies of scale
the OwenGraduateSchoolof Management forproviding
partialsupport and thus makes individualimprovementsless cost-effective
for this project.They also thankthe PromusCompaniesand Union
PlantersBankfortheircollaboration in the operationalization
(Anderson,Fornell, and Rust 1994).
of the ap-
proach.Anearlierversionof thisworkappearedas Marketing ScienceIn-
stituteTechnicalWorking Paper94-106. 1Anotherfactor leading itoWallace's collapse was a decline in
demandfor its products.

Journal of Marketing
58 / Journalof Marketing,April1995 Vol. 59 (April 1995), 58-70

This content downloaded from 128.6.218.72 on Sun, 25 Oct 2015 11:47:44 UTC
All use subject to JSTOR Terms and Conditions
 Quality improvementin services thus increasingly im- Drew (1991a, b) demonstratebehavioralimplicationsof the
plies spendingon qualityto improverevenuesratherthanre- customer satisfaction of telephone customers. Fornell
duce costs. How to make profitabledecisions about quality (1992) documents the aggregate financial implications of
expendituresis the key managerialproblem. This involves customer satisfaction across many industries in a huge
justifying all quality improvementefforts financially,know- Swedish study, and these results were later extended (An-
ing where to spend and not to spend on quality improve- derson, Fornell, and Lehmann 1994). Rust, Subramanian
ment, and knowing when to reduce spending. and Wells (1992) document the financial impact of com-
What is needed is a method to help managers decide plaint recovery systems. Nelson and colleagues (1992) find
where they are likely to get the greatest response for their a statistical link between patient satisfaction and hospital
limited resources.In general,expenditureson quality do not profitability.Anderson and Sullivan (1993) and Boulding
have obvious profit implications (Aaker and Jacobson and colleagues (1993) explore the impact of service quality
1994). They do not necessarily reduce costs and often in- on repurchaseintentions, Kordupleski,Rust, and Zahorik
crease them, at least in the short term (Griliches 1971). (1993) show the links between productquality,service qual-
Many quality efforts deal not with tangible products, but ity, and marketshare, and Rust and Zahorik(1993) explore
with intangibleaspects of service, such as the qualityof per- the diminishing returns and market share implications of
sonal interactions(Grant,Shani, and Krishnan1994). And, quality expenditures.Hauserand colleagues (1994) analyti-
unlike short-termsales promotions, the results are not im- cally show the financial implicationsof using customer sat-
mediately measurable in terms of sales. Nevertheless, the isfaction in employee incentive systems. (For a review of the
chain of effects from increased quality can be traced to the early literaturein this area, see Zahorikand Rust 1992.)
firm's profits. These studies are unanimous in finding that customer
The benefits of quality improvements come in two satisfactionand service qualityhave a measurableimpacton
forms. One effect is the improved ability of the firm to at- customer retention, market share, and profitability.These
tract new customers, due to word of mouth, as well as the findings have been recentlyunderscoredby anotherinterest-
firm's ability to advertisethe qualityof its offerings.This ef- ing finding:Juran,in one of his last lectures, was asked why
fect is in many ways analogousto productrepositioningand so many Baldrige award winners were financially unsuc-
is part of "offensive marketing"-those actions that seek to cessful. He said thathe was sure thatan investorinvesting in
attractnew customers.2 the Baldrige winners would beat the market.Business Week
The second result is that when current customers are magazine followed up, and discovered that an investor
more satisfied with the productsthey buy, they become re- putting money on the Baldrige winners the day they were
peat customers.Small increasesin retentionratescan have a announcedin fact would have received an 89% return,ver-
dramaticeffect on the profits of a company (Dawkins and sus 33% for the Standard& Poor 500 (Business Week1993).
Reichheld 1990; Fomell and Wernerfelt1987, 1988; Payne Thus, a link between quality and financialreturnexists, and
and Rickard 1993; Reichheld and Sasser 1990) for several the challenge is to provide operationalmethods for measur-
reasons:Existing customerstend to purchasemore thannew ing the link.
customers(Rose 1990), the efficiencies in dealing with them
is greater,and, comparedwith the cost of winning new cus-
tomers, selling costs are much lower-said to be on average The ROQApproach
only 20% as much, according to a much-quotedstudy for We describe an approachto making quality expendituresfi-
the U.S. Departmentof ConsumerAffairs (e.g., see Peters nancially accountable. The return on quality (ROQ) ap-
1988). Retainingcurrentcustomersthroughhigher levels of proachis characterizedby the following assumptions:
satisfaction is called "defensive marketing"(Fornell and
Wernerfelt1987, 1988). 1. Qualityis an investment,
2. Qualityeffortsmustbe financiallyaccountable,
A New Quality Movement: Financial 3. It is possibleto spendtoo muchon quality,and
Accountability 4. Not all qualityexpendituresareequallyvalid.
In an era of cost cutting, quality expendituresmust be made
The assumption that quality is an investment is a con-
financially accountable.This managerialneed has resulted
in a new quality movement in the marketingliterature,in scious attempt to place quality improvementexpenditures
which customersatisfactionand service quality are not only on an equal basis with other investment decisions. The al-
measuredbut also statisticallyrelated to customerretention ternativeis for quality expendituresto be made because of
and marketshare.Although analysis of the PIMS (ProfitIm- consistency with a quality culture or because eventual re-
turnsare takenon faith. Qualityexpendituresgenerallyhave
pact of MarketStrategy)databasesuggested some years ago
that a link between quality and profitability might exist not been treated as an investment by most companies, be-
(Buzzell and Gale 1987), a veritableexplosion of interestin cause there has been no solid basis for assessing financial
this area has occurredsince 1991. For example, Bolton and impact (Spitzer 1993). We provide a frameworkthat can be
used to evaluatethe financialimpactof qualityimprovement
2Conjointanalysismethodscanbe usedto determinethe "pull" efforts; thus enabling quality to be considered an invest-
thatupgradedqualitymighthave for customersof otherbrands ment.The second assumption,thatqualityefforts must be fi-
(DeSarboet al. 1994;GreenandSrinivasan
1978,1990;Windet al. nancially accountable,can thus be seen as synergistic with
1989). the assumptionthat quality is an investment.

Returnon Quality/59

This content downloaded from 128.6.218.72 on Sun, 25 Oct 2015 11:47:44 UTC
All use subject to JSTOR Terms and Conditions
 FIGURE 1 to greaterprofitability.The effect of word-of-mouthis very
A Model of Service Quality Improvement difficult to measurein a practicalbusiness situation;thus we
and Profitability use dotted lines in Figure 1 to indicate links that we do not
formally model. We can see immediately that our projec-
tions of service quality effects will be conservative, in that
they will ignore the positive benefits that might arise from
word-of-mouth.
The conceptual logic of Figure 1 can be convertedinto
general equationsthat model this chain of effects. If X is an
indicatorvariable(1 if the improvementis made, and 0 if it
M
tWor is not) and AQ represents"actual"(objective) service quali-
IWord-of-Mouth I ty,4then
(1) AQ = fl(X) + el
Attractionof
New Customers where El is a randomerrorterm reflecting all other nonsys-
:?.. . . . . . . . .?.?.?.?.?.?.??.I tematic influences on service quality. Let S be a vector of
measurementsof consumerattitudes,emotions, and percep-
tions (not all of these will typically be measuredin any one
application), including variables such as perceived service
quality,disconfirmation,and customer satisfaction.Then
If quality is an investmentand improvementefforts are (2) S = f2(AQ,E) + E2

financially accountable, then it is inevitable that some ef- where E is a vector containingcustomerexpectationsand all
forts will be evaluated as being ineffective, either because other systematic influences on S other than actual quality,
too much is to be spent (diminishing returns) or the im- and e2 includes all non-systematicinfluences. Also if CR is
provementmoney is spent for the wrong things (inefficient the cost reductionsrealized, then
use of funds). The assumptionsthat it is possible to spend
(3) CR = f3(AQ) + 63
too much on qualityand thatnot all qualityexpendituresare
equally valid thus are seen to be completely consistent with where 63 is a randomerrorterm.Note thatthe way this func-
the first two assumptions. tion is defined, if we assumed that the expectation of the
In the next section, we provide an overview of the theo- errorterm was 0, then E[f3(currentAQ)] = 0. That is, if we
retical frameworkof ROQ and the submodelsthatconstitute effect no improvement, then we do not expect any cost
the ROQ approach.We also develop the equationsnecessary reduction.
to project market share, net present value of quality im- Customerretention,R, then results from customer atti-
provementeffort, and returnon investmentof a quality im- tudes and perceptionsby
provementeffort (ROQ). In the thirdsection, we discuss is- (4) R = f4(S) + 64
sues in implementing the ROQ approach,including mea-
surement alternatives and the ROQ quality improvement where 64 is a randomerrorterm reflecting other factors af-
process. In the fourthsection, we presentan example appli- fecting retention.If MS is a vector reflecting business per-
cation, and, in the final section, summarizethe approachand formancevariablessuch as revenues and marketshare, MV
provide directionsfor furtherresearch. is a vector of all other systematiccompany- and market-spe-
cific variablesaffecting marketshare, and PROFITis some
measureof profits, then
EvaluatingFinancial Impact (5) MS = f5(R,MV) + ?5
Overview
(6) PROFIT= f6(MS,CR)+ 86
We model the relationshipbetween service qualityimprove-
ment efforts and profitabilityas a chain of effects (see Fig- where the errortermsreflect the realizationthat revenueper
ure 1).3 The improvementeffort, if successful, results in an customeris a randomvariable,which makes equation5 sto-
chastic and adds variabilityto equation 6.
improvementin service quality.Improvedservice qualityre-
sults in increased perceived quality and customer satisfac- In subsequentsections, we show how this simple model
tion and perhapsreducedcosts. Increasedcustomersatisfac- can be operationalizedwithin a particularfirm, taking into
tion in turnleads to higher levels of customerretention,and accountthe specific measurementpracticesof thatcompany.
also positive word-of-mouth.Revenues and marketsharego First, however, we show how equations 5 and 6 can be ex-
panded and operationalized to determine the profitability
up, drivenby higher customerretentionlevels and new cus-
tomers attracted by the positive word-of-mouth. The in-
creased revenues, combined with the decreased costs, lead 4Wearenotarguingthatobjectivequalityis in anywaymoreim-
portantthanperceivedquality.Thedistinctionbetweenactualand
perceivedservicequalityis necessaryto permitthepossibilitythat
3Another conceptualframework thatshowsthe sourcesof prof- cost reductionsmay resultfrom qualityimprovementsthat are
itabilityarisingfromqualityis thatof Garvin(1984). to (orirrelevant
transparent to) thecustomer.

60 / Journalof Marketing,
April1995

This content downloaded from 128.6.218.72 on Sun, 25 Oct 2015 11:47:44 UTC
All use subject to JSTOR Terms and Conditions
implications of change in customer retention,and ultimate- Ourmarketsharefor periodt is then estimatedto be the sum
ly to projectthe ROQ. of equations7, 8, and 9, divided by Nt.
Taken collectively, equations 1 through 9 give us the
Drivers of Market Share means to evaluatethe marketshareimpactof improvingper-
Customerretentionhas a majorimpact on marketshare,but ceived service quality or customer satisfaction.Because ef-
it is not the only factor.Marketgrowthrate, market"churn" fecting these shifts typically involves spending for quality
(customersenteringor leaving the market,even given a sta- improvement,we must determinewhether the returnon in-
ble marketsize), competitors'retentionrates, and the effec- vestment (ROQ) is adequate.
tiveness of offensive marketing efforts (e.g., advertising, If we assume a constantgrowthrateG, then marketsizes
can be projected out several periods. Then assuming that
price, convenience) all play a role in determining market
share.Given estimatesof these otherfactors,which are gen- churn (C), attractiveness (A), competitors' retention rate
erally availablefrom either internalcompany data, compet- (R'), and new retentionrate (R[X], where retentionrate is
itive analyses, or customer surveys, we can estimate the ef- expressed as a function of expenditurelevel X) are constant
over time, we can project marketshares out several periods
fect of our customerretentionrate on marketshare.We treat
as well. Assuming that average price and contributionmar-
the marketas a duopoly, in which all other competitorsare
gin (Y) remainconstant6(we will not raise price to cover the
lumped into one. We thus conduct an "us versus them" costs of quality improvement) and given a financial dis-
analysis, for two reasons:First,it reducesthe amountof data
counting factor (I) that reflects the cost of capital, we can
to be collected, and second, we do not need to determinethe calculate the net presentvalue of the profit flow over a time
effect of retention for the competitors individually. Given horizon of P periods:
this approach,we define the following:
p
Nt = marketsize at periodt, (10) NPV = Z(1 + I)-k[YMt+k(l+G)kNt- Xt+k].
k=l
Mt= marketshareat periodt,
C = chur, operationalized as the percentof customersleav- To calculate the ROQ, let us now assume that there is an up-
ing themarket, front expenditureof F' to initiate a quality improvementef-
R = ourretentionrate, fort for a particularbusiness process (or dimension of a
R' = competitors'retentionrate(definedas l-Pr(switchto us process) and after that an annual maintenanceexpenditure
or leavethemarket)),and of F (net of any cost reductions, which can be viewed as
A = percentage of newcustomerswhochooseus. negative expenditures).Let F0 indicate the currentlevel of
Then our numberof customersin periodt-1 is Mt_iNt_1,and annualexpenditure.Then the net presentvalue of additional
the competitorshad (1-Mt_l)Nt_1customers.The numberof spending is
p
new customers in period t is the sum of chum (CNt_l) and
marketgrowth5(Nt-Nt_1). (11) NPVAS= F' + Z (F -
k=l
Fo)(l+ I)-k
Operationally, if we assume that churn is the same = F' + (F - Fo)[(l - (1 + I)-P) / I)].
across all competitors,C can be estimatedfrom the percent- Let NPV and NPVo be the net present values of the profit
age of one's own customerswho leave the market,obtained streams for the quality improvementeffort and status quo,
from exit surveys. R' can be estimatedby taking one minus
respectively.Then the ROQ is
the ratioof the numberof new customerswho have switched
from competitors, divided by the number of competitors' (12) ROQ = (NPV - NPVo) / NPVAS.
customersin the previousperiod.This can be obtainedfrom This measure permits managementto consider quality im-
a survey of new customers.A can be estimatedas the num- provementas an investment,in terms of the financial return
ber of new customers divided by the number of total new generatedfrom the quality improvementexpenditure.
customers, for the period preceding the analysis or perhaps
obtainedfrom several periods of data.
Number of customers in time t is the sum of the cus- ImplementationIssues
tomers retained,plus the numberof customers who switch
to us, plus the numberof new customers.We can obtain the Ensuring Managerial Relevance
sizes of these subgroupsfrom the inputslisted previously,as It is importantto recognize that the operationalcontext of
follows: customer satisfaction/servicequality measurementis quali-
ty improvement.Thus, the questionnairestructuremust con-
(7) customersretained= RMt_iNt_ nect to managerial processes over which individual man-
agers can claim ownership. From a managerial point of
(8) customersswitching to us = (1 - R' - C)(1 - Mt_)Nt_1 view, one of the most important goals of any
(9) new customers = A[CNt_1+ Nt - Ntl]
6Thisassumes(for simplicityof exposition)that we will not
= A[Nt - (1 -
C)Nt-1].
raisepriceto coverthe costsof qualityimprovement andalso that
thereis no net reductionin variablecosts. If eitherof these as-
sumptionsis violated,it is easyto incorporate
theseintothemodel
5Notethat"growth"
canbe positive,zero,or negative. by adjustingthecontribution marginaccordingly.

Returnon Quality/ 61

This content downloaded from 128.6.218.72 on Sun, 25 Oct 2015 11:47:44 UTC
All use subject to JSTOR Terms and Conditions
satisfaction/qualitymeasurementprogramis helping man- tives that have been proposed in the literatureand could be
agement pinpoint the managerialprocesses and subprocess- used within the ROQ framework.In many cases, the choice
es in which quality improvementefforts will have the great- of measure depends on not only theoreticalconsiderations,
est impact on customerretention. but also corporatehistory. If a particularretention, quality,
One of the most common problemswith customersatis- or satisfaction measure has been used for years, then the
faction surveys is that the results are too often not manage- costs in terms of discontinuitymight argue against change.
rially relevant (Kordupleski,Rust, and Zahorik 1993). To For this reason, we now consider several of the majormea-
counteractthis, the ROQ approachstructuresthe customer surementalternatives.
satisfaction survey around business processes. The idea is
thatbusiness processes (e.g., sales, billing, product)are how Repurchase intention. Typically, we will have a repur-
chase intention measure rather than retention itself (e.g.,
the business is organized.For example, there is likely to be
a particularmanagerwho is in charge of the billing process. Rust, Subramanian,and Wells 1992; Rust and Zahorik
If particularaspects of billing need to be improved, then 1993). The categories on the repurchaseintention question
there is no ambiguity about who should supervise the may vary from companyto company,but it may be useful to
have more than a "yes-no" question for repurchaseinten-
changes or take responsibilityfor the results. tion. This is because in a very good company,the numberof
The idea is to structurethe questionnaire around the
"no's" is small, leading to a loss of explanatorypower and
processes of the business and also to determinethe questions in turn resulting in a very large sample size requirement.
to ask within the process on the basis of exploratoryanaly-
sis with customers. These questions should be in the cus- After testing several alternativeforms in parallel tests, we
tomers' language, using the words the customersuse to talk conclude that a repurchaseintentionscale of the form "0%,
about the relevant topics in focus groups. This approach 20%, 40%, 60%, 80%, 100%"works well with respondents
and producesan acceptableamountof variation.
guaranteesboth that the questionnairecovers topics which
are relevant to customers in words that customers use and Of course, we must recognize that the repurchaseinten-
thatthe resultsof the survey will be relevantto specific busi- tion category may not reflect the true probabilityof repur-
ness processes. We thus facilitate translatingexternal cus- chase. It is importantto recognize that the intention to re-
tomer information into internal process improvement purchaseis not the same thing as actual repurchase.These
information. categoriesmust be calibratedby following up on samples of
Figure 2 shows how the ROQ approachlinks customer respondentsto determinetheir actual repurchaseprobabili-
retentionto business processes to enable targetedqualityim- ties (Kaarre 1994). However, once this calibrationis done,
provement efforts. Measures of customer retention (or its the repurchaseintention question may be used directly for
surrogate,repurchaseintention) are linked to overall satis- some time, without furthercalibration.The probabilitiesin
faction measures. These overall satisfaction measures, in the scale are simply replacedby the calibratedprobabilities
turn, are linked to the business processes, and ultimately to (for example, only 95% of the "100%"category may repur-
dimensions or subprocesses within each process. The moti- chase, on average,which means that we simply treatthe top
vation for this structureis managerial:Quality improvement category as .95 for analysis purposes).
efforts must be targetedto the process and subprocesslevel Service quality. The direct measurementof perceived
to be actionable. service quality has been proposed by many authors in the
Measurement Alternatives academicliterature(e.g., Bolton and Drew 199la, b; Cronin
and Taylor 1992, 1994). This approachis also adopted by
The ROQ approach does not require any particularmea-
many practicingmarketingresearchers,who typically mea-
surementapproach.There are several measurementalterna- sure perceived service quality on somethinglike a "Poor"to
"Excellent"scale. Such researchersuse a varietyof numbers
FIGURE 2 of scale points, including 3, 5, 7, and 10.
Drivers of Customer Satisfaction
Given this measurementscheme, we would calibratethe
impactof overall service quality(OQ) on retention(R) using
the following equation:
(13) Ri = bo+ blOQi+ Ei
where the i subscriptrefers to individual i, si is a random
normaldisturbanceterm, and bo and b1 are regressioncoef-
ficients. Noting that Ri is a variablethat ranges from 0 to 1,
one might also use a logistic regression (Rust and Zahorik
1993), but that is probablynot necessary,because potential
quality shifts are usually small, which ensures that the re-
tention shift will also usually be small, and the logistic rela-
tionship between the variables will thus be approximated
0
0
0
0
well by a linear relationship.
Similarly,the overall service quality is linked to the per-
0 0

ceived quality levels of the processes:

62 / Journalof Marketing,April1995

This content downloaded from 128.6.218.72 on Sun, 25 Oct 2015 11:47:44 UTC
All use subject to JSTOR Terms and Conditions
(14) OQi = PQiPq + jYi ilar to disconfirmationin the customersatisfactionliterature,
where PQi is a vector of perceived quality levels for the except for the standardof disconfirmation.
Recently, several studies have pointed out the pitfalls
processes, as perceived by individual i, Pq is the corre- with using difference scores (Brown, Churchill, and Peter
sponding coefficient vector, and yi is a random errorterm.
The process quality levels can then be linked to the sub- 1993; Carman 1990; Cronin and Taylor 1992; Peter,
Churchill,and Brown 1993; Teas 1993). In particular,the re-
process quality levels using equations identical in form to
equation 14. The numberof levels of subprocessesis limit- liability of a differencescore is well known to be not as high
as the reliability of the constituent measures (Lord 1958).
ed only by the length of the questionnaireand the fatigue of
This has led many authors to recommend the direct mea-
the respondents.
surement of disconfirmation (Babakus and Boller 1992;
Estimationof equation 14 can be problematicbecause of
Carman 1990; DeSarbo et al. 1994; Devlin, Dong, and
potentially severe multicollinearity between the predictor Brown 1993; Oliver 1980). The direct measurementof dis-
variables(Petersonand Wilson 1992). Thus, we recommend
confirmationresults in a scale of the type "Muchbetterthan
using an estimationmethod that is robustto multicollineari-
expected"to "Muchworse than expected."
ty, such as ridge regression(Hoerl and Kennard1970) or the
equity estimator (Krishnamurthiand Rangaswamy 1987). Regardlessof whetherthe disconfirmationmeasure is a
The equity estimator,in particular,appearsto performbetter service quality gap or a direct disconfirmationmeasure,the
than ridge regression in the presence of multicollinearity structureof the analysis remains essentially the same. The
(Rangaswamyand Krishnamurthi1991). Recently,the equi- only difference is that the disconfirmationmeasurereplaces
the service quality measurein equations 1 and 2.
ty estimatorhas been questionedon both theoreticaland em-
pirical grounds (Hill and Cartwright1994), but the original Customerdelight. The importanceof customer delight
authors convincingly defended themselves by demonstrat- (or positive surprise), as opposed to mere satisfaction, has
ing that the equity estimatoris still generally more accurate been previously noted by researchersand managerialtheo-
when high levels of multicollinearityare present (which is rists (Chandler1989; Deming 1986; Oliver 1989; Westbrook
usually the case with service quality/customersatisfaction and Oliver 1991; Whittaker 1991). Our approachis broad
data;Krishnamurthiand Rangaswamy 1994). enough to accommodatedelight as a driverof customerre-
Customersatisfaction. Many researchershave explored tention-one way is to specify three categories: delight,
the natureand effects of customersatisfaction(e.g., Cadotte, mere satisfaction,and dissatisfaction(DeSarbo et al. 1994).
Woodruff,and Jenkins 1987; Churchilland Surprenant1982; This can be done by either using a three-pointmeasurement
Erevelles and Leavitt 1992; Oliva, Oliver, and MacMillan scale or dividing a larger scale into three groups. (For ex-
1992; Oliver 1980; Oliver and DeSarbo 1988; Oliver and ample, a five-point scale might convert 5 to "Delight,"4 to
Swan 1989; Tse and Wilton 1988). Customer satisfaction "Satisfied,"and 1-3 to "Dissatisfied.")
measures have also been linked to business performance Retention (or repurchaseintention)might then be relat-
(Anderson,Forell, and Lehmann1994;Andersonand Sulli- ed to satisfactionand delight by
van 1993; Forell 1992; Rust andZahorik1993). Also, many (15) Ri = bo + blOSi + b2ODi + ei
companies routinely measure customer satisfaction rather
than service quality (Devlin, Dong, and Brown 1993). where OSi is a dummy variable reflecting overall satisfac-
If customer satisfaction is measured instead of service tion (1 if satisfied or delighted, 0 otherwise), ODi is a
quality, the resulting equations are identical to those listed dummy variablereflecting overall delight (1 if delighted, 0
previously,with the exception that satisfactionmeasuresare otherwise), Ri and ei are defined as previously, and bo, bl,
substitutedfor the service quality measures. and b2 are regression coefficients. We might then relate
overall satisfaction to satisfaction with the processes and
Disconfirmation.The "expectancydisconfirmation"par- overall delight to delight with the processes, using the
adigm in the customer satisfactionliteratureposits that cus-
tomer satisfactionresults in large partfrom the disconfirma- following:
tion of priorexpectation(Day 1984; Oliver 1980; Olshavsky (16) Osi = PSIps + yj
and Miller 1972; Olson and Dover 1976; ). That is, if the
performanceof a service providermeets or exceeds expec- (17) ODi = PDiPd + bi
tations, then the customer is more likely to be satisfied. If
the performance fails to meet expectations, then the cus- where PSi is a vector of satisfaction dummy variables re-
tomer is more likely to be dissatisfied. flecting whetherindividuali was satisfiedwith each process,
A parallel research stream in service quality involves PDi is a vectorof delightdummyvariablesreflectingwhether
individuali was delighted with each process, Ps and 3d are
similar ideas. In that literature, dominated by the
vectors of regressioncoefficients, and yi and 5i are random
SERVQUAL approach(Parasuraman,Zeithaml, and Berry error terms, assumed normal for purposes of estimation.
1985, 1988; Zeithaml, Berry, and Parasuraman1993), cus-
tomers are asked for levels of agreementwith statementsin- Again, one could substitutea logistic regressionat this stage,
but it is probablynot necessaryfrom a practicalstandpoint.7
volving perceivedand ideal service performance.These lev-
els of agreement are termed "perceptions"and "expecta-
7Note also that the coefficients obtained by this procedurewill
tions," respectively. Subtractingthe "perception"minus the be equivalent,uponrescaling,to coefficientsthatwouldbe ob-
"expectation,"yields a gap, which is conceptuallyquite sim- tainedby two-groupdiscriminantanalysis.

Returnon Quality/ 63

This content downloaded from 128.6.218.72 on Sun, 25 Oct 2015 11:47:44 UTC
All use subject to JSTOR Terms and Conditions
 Cumulativefocus versus transactionfocus. The compa- FIGURE 3
ny must decide whether its customer satisfaction/service The ROQ Quality Improvement Process
quality questions will measurethe general perceptionof the
company's level of quality8 (cumulative focus) or the per-
ception of the company's level of quality on the most recent Stage 1: PreliminaryInformationGathering
transaction(transactionfocus). An example of a cumulative Customersurveys
question would be "Acme's billing quality is: Excellent, Marketinformation
Good, Fair, or Poor."An example of a transactionquestion Internalinformation
would be "On your most recent billing transaction with Heavy use of managementjudgment
Acme, Acme's billing quality was: Excellent, Good, Fair,or
Poor."
There are good reasons to select either option. Some re- Stage 2: Identificationof PossibleOpportunities
searchers have found that cumulative questions correlate
better with customer retention and other behavioral out- I
comes (Fornell 1992; Reichheld and Sasser 1990). This sug- Stage 3: LimitedTestingof Improvementsto
gests that if the company's primarymeasurementobjective DetermineEffectiveness
is to predictbehavior,then cumulativequestionsmay be pre-
ferred. On the other hand, suppose the company's primary
measurementobjective is to monitorthe progressof a qual-
Stage4: FinancialProjectionsBasedon HardData
ity improvementeffort. In such a case, the cumulativemea-
sure may be a poor choice, because it reflects both current
transactions and transactions that have taken place over
t
time. A transactionmeasure, because it measures only the Stage 5: Full Rolloutof QualityImprovementEfforts
I

most recent transaction,reflects qualityimprovementsfaster I

and provides a more accuratepicture of the currentperfor-
mance of the company in supplying quality to the customer
(Bolton and Drew 1991a). The ROQ approachcan be used alized at this point that these financial projections rely at
with either option. least partially(and perhapsto a greatdegree) on "soft"man-
The ROQ Quality Improvement Process agerialjudgment ratherthan hard data. Thus, it is often ad-
visable to verify the key assumptions before committing
Few managers will take one run through this approach, large amountsof money.
guessing at inputs where needed, and then commit substan-
tial sums of money to a particularquality improvementef- Stage 3: Limitedtesting of improvements.Then a limit-
fort. Rather, information must be gathered to support the ed-scale test of the proposed quality improvementeffort is
conducted.For example, if a hotel chain wants to test a par-
original managerialestimates, with financialresourcescom-
mittedonly aftermanagementis convinced thatthe inputsto ticularquality improvementeffort, then a randomsample of
the system are sufficiently solid. Figure 3 shows how this the chain's hotels can be used in a test. Of particularinterest
is whether the costs are as anticipated and the extent to
process can be visualized as consisting of five distinct
which the expected shift in percent satisfied or percent de-
stages: (1) preliminaryinformationgathering,(2) identifica-
tion of possible opportunities, (3) limited testing of im- light actually occurs. The numbersfrom the test provide es-
timates of the effectiveness of the quality improvementef-
provements, (4) financial projections based on hard data,
and (5) full rollout of quality improvementefforts. fort, based on harddata.

Stage 1: Preliminary information gathering. The first Stage 4: Financialprojectionsbased on hard data. Then
it is back to the financial impact equations, with the revised
stage involves collecting customer survey data, information
aboutthe market,and whateverinternalinformationis avail- cost and effectiveness estimates now used as input. This re-
able and plugging information holes, wherever necessary, sults in more accurateestimates of the ultimatefinancial re-
with managerialestimates (Griffinand Hauser 1993). In this sults, such as ROQ, NPV, optimal expenditure level, and
marketshare trajectory,providing a final check of whether
stage, there is typically heavy use of managerialjudgment,
because therewill inevitablybe informationgaps in even the the proposedimprovementeffort is financiallyjustified.
most sophisticatedcompanies (Little 1970). Stage 5: Full rollout of quality improvementefforts.
Stage 2: Identificationof possible opportunities.In the Managementcan now roll out the quality improvementef-
next stage, the manager identifies the best opportunityfor fort, confident that a strong financial returnwill result from
profitablequality improvement.Using the inputs from stage the necessary expenditure.Equations 1 through 8 can then
1, the financial implicationsof the opportunity(ROQ, NPV, be used to approximatethe actual ROQ, on the basis of the
and marketshare trajectory)are all estimated.It must be re- satisfactionimprovementsactually obtained.
Continualimprovementthen demandsthat management
8Quality is used here as an example. Obviously, the company go back to stage 1 and reevaluatethe business processes for
might instead be measuring customer satisfaction or furtherpotential opportunities,on the basis of updatedcus-
disconfirmation. tomer survey data.

64 / Journalof Marketing,April1995

This content downloaded from 128.6.218.72 on Sun, 25 Oct 2015 11:47:44 UTC
All use subject to JSTOR Terms and Conditions
 Summary of Managerial Inputs 11. The contributionmarginfrom an average customer must
be estimated. This is obtained either per transaction (in
Estimatingthe ROQ requiresactive managementparticipa- transactionbusinesses) or per period (in businesses entail-
tion. Specifically, there are many things that must be mea- ing continuousrelationships).This figure is obtainedfrom
sured, estimated, or otherwise quantifiedto apply the ROQ internalaccountingdata.
approach. Although we have introduced these inputs 12. The cost of capital must be determined.Companies typi-
throughoutthe article wherever appropriate,we now take cally use this numberfor financialplanning.
stock and catalog exactly what information management 13. The time horizon must be specified. This is the length of
must provide to operationalizethe ROQ approach. time in which an investment project must prove itself.
Again, this is a common consideration in financial
1. The key managementprocesses must be identified.These analysis.
processes are identified by management.Many companies 14. A specific quality improvementalternativemust be identi-
have identified their key processes as part of TQM or
fied. Typically,the quality improvementpossibility is cho-
reengineeringprograms;however we should point out that sen to improve an importantprocess, as identified by the
the processes identified in that way may not be exactly statisticalanalysis.
what are needed here. The ROQ approachis customerfo-
cused rather than externally focused, which means that 15. The additionalexpendituresrelatedto this improvementef-
only those processes that directly affect the customer fort mustbe estimated.These estimates may include an up-
should be included. front expenditure to launch the effort, plus per-period
maintenancecosts.
2. Key dimensions of each process must be obtained. Ex-
16. Cost savings must be estimated. For some quality im-
ploratoryresearchis used to determinefrom customers,in
the customers' own words, the most importantaspects of provementefforts, a significantpartof the gain is from cost
each process. savings. Although these are not easy to calculate, there has
been considerableprogress in recent years in quantifying
3. Customerretention(or repurchaseintention)must be mea- cost savings from quality improvementefforts (Campanel-
sured. Ideally, customer retention behavior is obtained la 1990; Carr1992; Gryna 1988; Nandakumaret al. 1993).
from a database,and then is matched with customer satis-
17. The satisfaction shift must be estimated. Until markettest
faction surveys (Kaarre1994). Because most organizations
data are available, management must supply initial esti-
are not so fortunateto have such a database,it is often nec-
mates of the effect of the improvementeffort. The effect is
essary to obtain a repurchaseintention measure as part of
the customer satisfactionsurvey. expressed in a way that is compatible with the company's
measurementscheme (e.g., shift in mean satisfaction or
4. Customer satisfaction (or a suitable substitute) must be mean perceivedquality,shift in percentsatisfied or percent
measured. Almost all major companies already measure delighted).
customer satisfaction, service quality, or disconfirmation
18. Market test data may be obtained. (optional) Obtaining
on a routine basis. However, to apply the ROQ approach,
markettest data can help by supplying objectively derived
the survey questions must be grouped into business estimates of the satisfactionshifts.
processes, overall process satisfaction must be measured,
and overall satisfaction must be measured as well. Thus,
there typically must be minor additions to the typical sur- Other Issues
vey to be able to apply the ROQ approach. Successful implementationof the ROQ approachrequires
5. Marketsize must be measured.This involves defining the
considerationof several practical issues. Among the most
marketin which the firm is competing and then determin-
ing the numberof customersin the market. importantare modificationof existing satisfactionquestion-
6. Currentmarketshare must be estimated.Almost all com- naires, conversion of already existing satisfaction scales,
and marketsegmentation.
panies alreadyobtain good estimates of marketshare.
7. Chur must be estimated. Churn, the percentage of cus- Many companies that might want to adopt the ROQ ap-
tomers leaving the marketaltogetherin a particularperiod, proachmay alreadyhave existing customer satisfactionsur-
is often difficult to obtain directly.Approximatechur can veys (Honomichl 1993), and implementing too great a
be obtained from secondary sources, such as income tax change in the existing survey can cause problems,both with
recordsand census reports. monitoring satisfaction scores over time and in educating
8. The company's current retention rate must be estimated. managersabout what the scores mean. In general, it is ad-
On the basis of internalcompany records (usually) or pri- visable to modify the surveys as little as possible. Fortu-
mary research (sometimes), the probabilityof a customer nately, it is usually possible to employ the ROQ approach
choosing this company on the next transaction(if choice is with only minormodificationsto the original survey.In par-
based on transactions,such as is truein hotel stays, for ex-
ticular, the questions most often missing from the typical
ample), or stays with the company until the next period (if
customer satisfaction survey are overall satisfaction ques-
long-term relationships are involved, as in banking ac-
counts, for example) must be obtained. tions (American Productivity& Quality Center 1994). It is
9. The attraction percentage must be obtained.9 This is the essential, in using the ROQ approach,that overall satisfac-
percentageof new (to the market)customerschoosing this tion questionsbe presentfor the service overall and for each
company (not the same thing as marketshare). of the processes overall (see Kordupleski,Rust, and Zahorik
10. The marketgrowth rate must be estimated. Most compa- 1993). It is also necessaryto make sure thateach satisfaction
nies alreadyprojectthis. question relates to a particularmanagerialprocess (e.g., in-
stallation, billing). This ensures "ownership" of the results
9Wehave also developed approximationequationsfor ROQthat by specific managers or teams and makes the survey results
do not requirethis input. more actionable.

Returnon Quality/ 65

This content downloaded from 128.6.218.72 on Sun, 25 Oct 2015 11:47:44 UTC
All use subject to JSTOR Terms and Conditions
 FIGURE 4 The company's managers had already structured its
Process Impact on Overall Satisfaction questionnaire
by businessprocesses,so no majormodifica-
tions were required.To examine nonlinear customer satis-
faction effects, we converted their five-point satisfaction
Bathroom scale to a three-pointscale. For convenience, we refer to re-
Room spondentsrecording5 as "delighted,"3 and 4 as "satisfied"
_and
_-^^^^^^^^^^^^^^f~~ 1 and 2 as "dissatisfied."10
Analysis of repurchaseintentionas a function of overall
Restaurant satisfactionand delight showed that the disappointedgroup
Facilities had only a 45% probabilityof returning,whereas the satis-
fied group had a 95% probabilityand the delighted group
Grounds
had a 97% probability.This indicates that the hotel chain's
RoomService biggest benefits are derived from converting customers from
'
. .0.15
.~~ dissatisfied to satisfied, which is likely accomplished by
0 .05 .1 0 .1 .2
. .25
.25
solving or avoiding problems.However,this must be traded
Equity Estimator Coefficient off against the fact that only 9% of the chain's customers are
dissatisfied,whereas75% are merely satisfied.Thus, thereis
a greater benefit from shifting dissatisfied customers, but
Segmentation is another issue about which the re- there are fewer of them to shift. In this case, it was decided
searcher must be careful. Heterogeneity in the population that the huge jump in repurchaseprobabilitymade trying to
may mean that some customersrespondwell to some quali- reduce the proportion of dissatisfied customers the top
ty improvementefforts, whereas others respond less well priority.
(DeSarbo 1993). Thus, it is advisableto conductpreliminary The managersthen had to addressthe issue of where to
analysis to see whether segments respond differently and improvesatisfaction.They recognized the importanceof fo-
then to conduct ROQ analysis on a segment basis, essential- cusing on a small numberof potential improvementsrather
ly considering each segment individually,perhaps address- than generatinga "laundrylist" of problems and opportuni-
ing first the segment currentlymost importantto the busi- ties. To make sure their effort was focused, they decided to
ness's strategicplan. select one key area for improvementby using the ROQ ap-
proach to determine the relative impact of satisfaction on
An Illustrative Application each of the processes on overall satisfaction.Figure4 shows
the result of the equity estimator regression coefficients
Thoroughtesting of the ROQ approachwill requireseveral usg equaton From this we can see that satisfaction
years to complete. However, we have operationalizedthe wth the bathroomhas the largest mpact on overall satisfac-
ROQ approachin a PC-based decision
asystem and
supportt tion. Similarly,they investigatedthe dimensions within the
implementedthe system at two companiesthat agreedto act bathroom(Figure5) and determinedthat cleanliness was the
as "beta test" sites. We reporthere some preliminaryresults most powerful variable
from one of those applications.The resultswe presentin this t i pi i w e o o o e
section are disguised to protect the proprietaryinterests of p t wod e acco ished, and
provement would be accomplished, and what it would
the company involved, but we have tried to preserve their cost.
general
pattem. cost.ll There
Thereare
arevarous
variousways to mprove bathroomcleanl-
ways to improvebathroom cleanli-
~genera~~l~ pattem~rn.^ ness. These include instructingthe staff to
cleaning spend
A National Hotel Chain more time cleaning each bathroom, assigning additional
personnel to clean bathrooms,or increasing the trainingon
Our example beta test site is a nationalhotel chain that has pow to clean bathroom.Ultimately,the managersdecided
been tracking customer satisfaction data for several years. at the appropriatemethod was to increase the amount of
The firm sends mail surveys to a random sample of cus- time the cleaning staff would spend on each bathroom.This
tomers within several days after they check out. The ques- cts moe, ecause the numberof rooms cleaned per day
tionnaireasks for ratingson seven differentprocesses of the by each cleaning person is reduced.
hotel service (e.g., the room, the grounds,the staff, the bath- Lmtedteng du es reeed e
Limited testing at a handful of hotels revealed the re rela-
room), with additionalquestions on as many as 15 dimen- tionship between time spent cleaning each bathroom and
sions of each of those processes. The survey also asks an stistion time spent c e ea t
satisfaction.The time spent is convertedeasily to cost, using
overall satisfaction question, as well as a question on the
likelihood of repurchase.The firm has,.?., , A ^ r. ,
?"average wage rates, resulting in the financial
the relationship
om
a 30% response rate, shown in Figure 6. Managers estimated that the company
and, through telephone follow-ups of nonrespondents,has
-us has was currently million annually ont this process
Ad,througdha
determined tehoe
that follw
the answers of
respondentsappearvery r-
of nonwrespondents,apa rep- di t o$1 t
A spending r t
dimension.Applicationof the ROQ approachthen calculat-
resentativeof the attitudesof nonrespondents.Ourdata con-
of77882tindividual
stenta
sisted attitud res our datacon-
pononrespaondentsg ed the implied NPV for several expenditurelevels across a
responses spanningone year.
realistic range of possible expenditures ($600,000 to $3.6
million annually).An expenditurelevel of $2.4 million an-
10Werecognizethattheremaybe some"5"respondents whoare
nottrulydelighted.Nevertheless
the"5"responseis thebestproxy 1 Thissectiondisguisessomeelementsof the to pro-
application
fordelightthatwe haveavailableon the survey. tecttheproprietary
interestsof theparticipating
firm.

66 / Journalof Marketing,April1995

This content downloaded from 128.6.218.72 on Sun, 25 Oct 2015 11:47:44 UTC
All use subject to JSTOR Terms and Conditions
nually is the best of the candidate expenditurelevels, with of the company and are inevitably considered discretionary
respect to maximizing the net present values of profit flows costs, which can then be cut when times are tough.
over the planning horizon. In this case, the cleaning staff To combat this trend, we have devised an approachthat
was instructedto spend almost two and a half times as long can quantifythe marketshareimplications,net presentvalue
cleaning each bathroom,with appropriatedetailed instruc- of the resultingprofit stream,and ROQ of a proposedqual-
tions about exactly how to use that time. Note that this ity expenditure.This approachcan also reveal where spend-
analysis could just as easily have revealedthata reductionin ing for quality improvementmay be appropriateand esti-
spending was in order. mate the optimal expenditurelevel. This approachuses cus-
The ROQ approachalso facilitates the estimationof the tomer satisfactionsurvey data, internalcompany data, com-
market share impact of the shift in satisfaction. Figure 7 petitive marketdata, test marketdata, and managerialjudg-
shows the marketshare trajectoriesprojectedto result from nients to form its estimates.
variouslevels of expenditure.We can see thatpast $2.4 mil- We have appliedthe ROQ approachin two companies as
lion annually,there is very little impacton marketshare.Be- beta tests. In the application described here, the company
cause of the dynamics of this particularmarket,long-term had a long historyof customersatisfactionmeasurementand
marketsharesare expected to decline for all spendinglevels quality improvement efforts. The implementation issues
eventually, but they will decline faster if a suboptimal thererelatedto adaptingthe system to theircultureand mak-
amountis spent on quality. ing do with some nonoptimalaspects of their surveyprocess
Finally we can calculate the projectedROQ.Assuming a to maintaincontinuity in their customer satisfaction moni-
time horizon of three years, and a discount rate of 15%, the toring. We recommend that those implementing an ROQ
net presentvalue of additionalprofitswere (referringto Fig- system within an alreadysophisticatedcompanypay special
ure 6) $177,121,000, an increase of $1.641 million. The ad- attentionto making sure that the ROQ quality improvement
ditional expenditure of $1.4 million annually, discounted
over threeyears, is $3.676 million. Thus, the projectedROQ FIGURE 6
for this quality initiativeis Impact of Quality Expenditures on Percent
(18) ROQ = $1.641/$3.676 = 44.6%. Disappointed
A 44.6% returnon investmentis a very healthy return.
Percent of Total
This makes it possible for the investment in quality to be Customers
considered alongside other investments the firm might
make, rather than merely being considered a cost, which
could be cut when times are tough.

Discussion and Conclusions

The quality maniaof the 1980s has yielded to more sober fi-
nancial analysis in the 1990s. The result is that many quali- 5%-
ty initiatives are currentlybeing forced to justify their exis-
tence on financialgrounds.If a companyis not able to spec- 0% . . . . .
ify the returnon investmentof its quality efforts (the ROQ), 0g
those quality efforts cannot compete with other investments 8 8 8
Service Improvement Effort ($000)

FIGURE 5
Impact of Process Dimension Satisfaction on
Process Satisfaction FIGURE 7
Projected Market Share Trajectories Resulting
For Bathrooms From Different Spending Levels

Cleanliness
Market Share

Tub 2.35% -
I
2.40%
Lighting/Mirror
2.45%
Vanity - $0
2.50%
Water Pressure --$IM
2.55% -A$2.4M
Towels X $4M
2.60%
Supplies 2.65%
I I

0 .05 .1 .15 .2 .25 .3 .35 .4 2.30%

0 1 2 3 4 5 6 7 8 9 10
Equity Estimator Coefficient Period

Returnon Quality/ 67

This content downloaded from 128.6.218.72 on Sun, 25 Oct 2015 11:47:44 UTC
All use subject to JSTOR Terms and Conditions
approach meshes closely with the existing quality improve- and (1991b), "A Multistage Model of Cus-
ment approach. Specifically, we recommend that each of the tomers' Assessment of Service Quality and Value,"Journal of
ConsumerResearch, 17 (March),375-84.
stages shown in Figure 3 be matched closely to, and bonded
Boulding, William, Ajay Kalra, Richard Staelin, and Valarie A.
to existing quality improvement processes within the com- Zeithaml (1993), "A Dynamic Process Model of Service Qual-
pany. In other words, the system must be positioned as a tool ity,"Journal of MarketingResearch, 30 (February),7-27.
for helping the company to do its existing work better, rather Brown, Tom J., GilbertA. Churchill,Jr., and J. Paul Peter (1993),
than as a new approach that will force management to work "Improvingthe Measurementof Service Quality,"Journal of
differently. Retailing, 69 (Spring), 127-39.
Business Week(1993), "Betting to Win On the Baldie Winners"
We were able to tell managers where additional invest-
(October 18), 8.
ment might be worthwhile and where it would not be rec- Buzzell, Richard and Brad Gale (1987), The PIMS Principles:
ommended. We were also able to get an idea of the financial LinkingStrategyto Performance.New York:The Free Press.
implications of potential quality expenditures. Cadotte, Ernest R., Robert B. Woodruff, and Roger L. Jenkins
Researchers in the future should validate this approach (1987), "Expectationsand Norms in Models of CustomerSatis-
over time by adopting the quality improvement expenditures faction,"Journal of MarketingResearch, 24 (August), 353-64.
the system recommends and then tracking whether the out- Campanella,Jack (1990), Principles of Quality Costs, 2nd ed. Mil-
waukee:ASQC Quality Press.
comes are as predicted (e.g., see Fudge and Lodish 1977). Carman,J. (1990), "ConsumerPerceptionsof Service Quality:An
The system should be applied across several industries to Assessment of SERVQUALDimensions,"Journalof Retailing,
confirm that the approach is sufficiently generalizable. By 66 (Spring), 33-55.
its nature, this follow-up research will take years to com- Carr,LawrenceP. (1992), "ApplyingCost of Quality to a Service
Business,"Sloan ManagementReview, (Summer),72-77.
plete, but it is a necessary next step.
Chandler,Colby H. (1989), "Quality:Beyond CustomerSatisfac-
To summarize, we have presented a financial approach
tion,"Quality Progress, 22 (February),30-32.
to quality improvement, which we term the "return on qual- Churchill,GilbertA., Jr. and Carol Surprenant(1982), "An Inves-
ity (ROQ)" approach. This approach treats quality improve- tigation Into the Determinantsof CustomerSatisfaction,"Jour-
ment efforts as investments and assumes that these efforts nal of MarketingResearch, 19 (November),491-504.
must be made financially accountable. We note that it is pos- Cronin,J. Josephand StevenA. Taylor(1992), "MeasuringService
sible to spend too much on quality and that not all quality Quality:A Reexaminationand Extension,"Journal of Market-
ing, 56 (July), 55-68.
expenditures are equally valid. The ROQ approach enables and (1994), "SERVPERF Versus
managers to determine where to spend on service quality, SERVQUAL: Reconciling Performance-Based and Percep-
how much to spend, and the likelyfinancial impact from ser- tions-Minus-ExpectationsMeasurement of Service Quality,"
vice expenditures, in terms of revenues, profits, and return Journal of Marketing,58 (January),125-31.
on investments in quality improvement-the "return on Crosby,Philip B. (1979), Qualityis Free. New York:McGrawHill.
Dawkins, Peter and FrederickF. Reichheld (1990), "CustomerRe-
quality." tention as a Competitive Weapon,"Directors and Boards, 14
(Summer),41-47.
Day, RalphS. (1984), "Towarda Process Model of ConsumerSat-
REFERENCES isfaction,"in Conceptualizationand Measurementof Consumer
Aaker, David A. and Robert Jacobson (1994), "The Financial In- Satisfactionand Dissatisfaction, H. Hunt, ed. Cambridge,MA:
formationContent of Perceived Quality,"Journal of Marketing MarketingScience Institute.
Research, 31 (May), 191-201. Dean, James W., Jr. and James R. Evans (1994), Total Quality
American Productivity& Quality Center (1994), "A Compilation Management, Organization,and Strategy. Minneapolis: West
of CustomerSatisfactionSurveys and Process Flows."Houston: PublishingCompany.
American Productivity& Quality Center. Deming, W. Edwards(1986), Out of the Crisis. Boston: MIT Cen-
Anderson, Eugene W., Claes Fornell, and Donald R. Lehmann ter for Advanced EngineeringStudy.
(1994), "CustomerSatisfaction, MarketShare, and Profitabili- DeSarbo, Wayne (1993), "Consumer Heterogeneity in Service
ty,"Journal of Marketing,58 (July), 53-66. Quality Perceptionand AttributeImportance,"presentedat the
and RolandT. Rust (1994), "CustomerSatis- AMA Frontiersin Services Conference,Nashville, TN.
faction, Productivity, and Profitability: Differences Between , LeonardHuff, Marcello M. Rolandelli, and Jungwhan
Goods and Services,"working paper,Universityof Michigan. Choi (1994), "On Measurementof Perceived Service Quality:
and Mary W. Sullivan (1993), "The Antecedents and A Conjoint MeasurementApproach,"in Service Quality, R.T.
Rust and R.W. Oliver, eds. Thousand Oaks, CA: Sage Press,
Consequences of Customer Satisfaction for Firms,"Marketing 199-220.
Science, 12 (Spring), 125-43.
Arnold,WilliamW. and JeanneM. Plas (1993), TheHuman Touch. Devlin, Susan J., H.K. Dong, and MarbueBrown (1993), "Select-
New York:John Wiley & Sons. ing a Scale for MeasuringQuality,"MarketingResearch, 5 (3),
12-17.
Babakus, Emin and GregoryW. Boller (1992), "An EmpiricalAs-
sessment of the SERVQUAL Scale," Journal of Business Re- Erevelles, S. and Clark Leavitt (1992), "A Comparisonof Current
Models of Consumer Satisfaction/Dissatisfaction,"Journal of
search, 24, 253-68. ConsumerSatisfaction, Dissatisfaction, and ComplainingBe-
Bohan, George P. and Nicholas F. Homey (1991), "Pinpointingthe havior, 5, 104-14.
Real Cost of Quality in a Service Company,"National Produc-
Fornell,Claes (1992), "A National CustomerSatisfactionBarome-
tivity Review (Summer), 309-17. ter: The Swedish Experience,"Journal of Marketing,56 (Janu-
Bolton, Ruth and James Drew (1991a), "A LongitudinalAnalysis ary), 6-21.
of the Impactof Service Changeson CustomerAttitudes,"Jour- , Lenard Huff, and Eugene W. Anderson (1994), "Are
nal of Marketing,55 (January),1-9. There Trade-offsBetween Productivityand Quality?"working

68 / Journalof Marketing,April1995

This content downloaded from 128.6.218.72 on Sun, 25 Oct 2015 11:47:44 UTC
All use subject to JSTOR Terms and Conditions
 paper, National Quality Research Center, University of Lord, FrederickM. (1958), "The Utilization of Unreliable Differ-
Michigan. ence Scores," Journal of Educational Psychology, 19 (3),
and Birger Wernerfelt (1987), "Defensive Marketing 150-52.
Strategy by Customer Complaint Management:A Theoretical Nandakumar,P., S.M. Datar, and R. Akella (1993), "Models for
Analysis," Journal of Marketing Research, 24 (November), Measuringand Accounting for Cost of ConformanceQuality,"
337-46. ManagementScience, 39 (January),1-16.
and (1988), "A Model for Customer Com- Nelson, Eugene, RolandT. Rust,AnthonyZahorik,Robin L. Rose,
plaint Management,"MarketingScience, 7 (Summer),271-86. Paul Batalden,and Beth A. Siemanski (1992), "Do PatientPer-
Fudge, W.K. and LeonardM. Lodish (1977), "Evaluationof the Ef- ceptions of Quality Relate to HospitalFinancialPerformance?"
fectiveness of a Model-Based Salesman's Planning System by Journal of Health Care Marketing,13 (December), 1-13.
Field Experimentation,"Interfaces, 8, Part 2 (November), Oliva,TerenceA., RichardL. Oliver,and Ian C. MacMillan(1992),
97-106. "A Catastrophe Model for Developing Service Satisfaction
Garvin, David A. (1984), "What Does 'Product Quality' Really Strategies,"Journal of Marketing,56 (July), 83-95.
Mean?"Sloan ManagementReview (Fall), 25-48. Oliver,RichardL. (1980), "A Cognitive Model of the Antecedents
and Consequencesof SatisfactionDecisions," Journal of Mar-
Grant, Robert W., Rami Shani, and R. Krishnan(1994), "TQM's
ketingResearch,42 (November),460-69.
Challenge to ManagementTheory and Practice,"Sloan Man-
(1989), "Processing of the Satisfaction Response in
agement Review (Winter),25-35.
Green, Paul and V. Srinivasan(1978), "ConjointAnalysis in Con- Consumption:A Suggested Frameworkand ResearchProposi-
sumerResearch:Issues and Outlook,"Journalof ConsumerRe- tions,"Journal of ConsumerSatisfaction, Dissatisfaction, and
ComplainingBehavior, 2, 1-16.
search, 5 (September), 103-23.
and (1993), "A Conceptual Model of Service Quality and
(1990), "ConsumerResearchin Market- Service Satisfaction:CompatibleGoals, DifferentConcepts,"in
ing: New Developments with Implications for Research and Advances in Services Marketing and Management: Research
Practice,"Journal of Marketing,54 (October),3-19. and Practice, TeresaA. Swartz, David E. Bowen, and Stephen
Greising, David (1994), "Quality:How to Make It Pay,"Business W. Brown, eds., 2, 65-85.
Week(August 8), 54-59. andWayneDeSarbo (1988), "ResponseDeterminantsin
Griffin,Abbie and John R. Hauser (1993), "TheVoice of the Cus- Satisfaction Judgments,"Journal of Consumer Research, 14
tomer,"MarketingScience, 12 (Winter), 1-25. (March),495-507.
Griliches, Ziv (1971), Price Indices and Quality Change. Cam- and John Swan (1989), "Equity and Disconfirmation
bridge, MA: HarvardUniversity Press. Perceptions as Influences on Merchantand Product Dissatis-
Gryna, Frank M. (1988), "Quality Costs," in Quality Control faction," Journal of Consumer Research, 16 (December),
Handbook,4th ed. New York:McGraw-Hill,4.1-4.30. 372-83.
Hauser,John R., Duncan I. Simester,and BirgerWernerfelt(1994), Olshavsky, R. and J. Miller (1972), "Consumer Expectations,
"Customer Satisfaction Incentives," Marketing Science, 13 ProductPerformanceand Perceived ProductQuality,"Journal
(Fall), 327-50. of MarketingResearch,9 (February),19-21.
Heskett, James L., W. Earl Sasser, Jr., and ChristopherW. L. Hart Olson, JerryC. and Philip Dover (1976), "Disconfirmationof Con-
(1990), Service Breakthroughs.New York:The Free Press. sumerExpectationsThroughProductTrial,"Journalof Applied
Hill, R. Carterand P.A. Cartwright(1994), "The StatisticalProper- Psychology, 64 (April), 179-89.
ties of the Equity Estimator,"Journal of Business and Econom- Parasuraman,A., ValarieA. Zeithaml, and LeonardBerry (1985),
ic Statistics, 12 (April), 141-47. "A Conceptual Model of Service Quality and its Implications
Hill, RobertC. (1993), "Whenthe Going Gets Rough:A Baldrige for FurtherResearch,"Journal of Marketing,48 (Fall), 41-50.
AwardWinnerOn the Line,"Academy of ManagementExecu- , , and (1988), "SERVQUAL:A Multiple-Item
tive, 7 (August), 75-79. Scale for MeasuringConsumerPerceptionsof Service Quality,"
Hoerl, A.E. and R.W. Kennard(1970), "Ridge Regression:Biased Journal of Retailing, 64 (1), 12-40.
Estimationfor Non-OrthogonalProblems,"Technometrics,12, Payne, Adrianand John Rickard(1993), "RelationshipMarketing,
55-67. Customer Retention and Service Firm Profitability,"working
Honomichl, Jack (1993), "Spendingon SatisfactionMeasurement paper,CranfieldUniversity.
Continues to Rise," MarketingNews (April 12), 17-18. Peter,J. Paul, GilbertA. Churchill,Jr.,and Tom H. Brown (1993),
Juran,J.M. and FrankM. Gryna,Jr. (1980), Quality Planning and "Caution in the Use of Difference Scores in Consumer Re-
search,"Journal of ConsumerResearch, 19 (March),655-62.
Analysis: FromProductDevelopmentThroughUse. New York:
McGraw-Hill. Peters,ThomasJ. (1988), Thrivingon Chaos. New York:AlfredA.
Kaarre,Jim (1994), "PerceivedValue, Customer Satisfaction, and Knopf.
Peterson, Robert A. and William R. Wilson (1992), "Measuring
Profitability:An Actionable Perspective,"presentation at the CustomerSatisfaction:Fact and Artifact,"Journal of the Acad-
Third Annual AMA Frontiers in Services Conference,
emy of MarketingScience, 10 (Winter),61-71.
Nashville, TN.
Rangaswamy,Arvind and LakshmanKrishnamurthi(1991), "Re-
Kordupleski,Ray, RolandT. Rust, and AnthonyJ. Zahorik(1993),
sponse FunctionEstimationUsing the Equity Estimator,"Jour-
"Why ImprovingQuality Doesn't ImproveQuality,"California nal of MarketingResearch, 28 (February),72-83.
ManagementReview, 35 (Spring), 82-95. Reichheld, FrederickF and W. Earl Sasser (1990), "Zero Defec-
Krishnamurthi,Lakshmanand Arvind Rangaswamy(1987), "The tions: Quality Comes to Service," Harvard Business Review,
Equity Estimatorfor MarketingResearch,"MarketingScience, (September-October),105-11.
6 (Fall), 336-57.
Rice, Faye (1990), "How to Deal With TougherCustomers,"For-
and (1994), "The StatisticalPropertiesof the tune (December 3), 38-48.
Equity Estimator:A Reply,"Journal of Business and Economic Rose, S. (1990), "The Coming Revolution in CreditCards,"Jour-
Statistics, 12 (April), 149-53. nal of Retail Banking, 12 (Summer), 17-19.
Little, John D.C. (1970), "Models and Managers:The Concept of Rust, RolandT., Bala Subramanian,and MarkWells (1992), "Mak-
a Decision Calculus," Management Science, 16 (April), ing Complaintsa ManagementTool,"MarketingManagement,
B466-85. 3, 40-45.

Returnon Quality/ 69

This content downloaded from 128.6.218.72 on Sun, 25 Oct 2015 11:47:44 UTC
All use subject to JSTOR Terms and Conditions
 andAnthonyJ. Zahorik(1993), "CustomerSatisfaction, Whittaker, Barrie (1991), "The Path to Excellence," Canadian
Customer Retention and Market Share,"Journal of Retailing, Business Review (Winter), 18-21.
69 (Summer), 193-215. Wiesendanger,Betsy (1993), "Deming's Luster Dims at Florida
Spitzer, RichardD. (1993), "ValuingTQM ThroughRigorous Fi- Power & Light" Journal of Business Strategy, 14 (September-
nancialAnalysis,"QualityProgress (July), 49-54. October),60-61.
Teas, R. Kenneth (1993), "Expectations,PerformanceEvaluation, Wind, Jerry, Paul E. Green, Douglas Shifflet, and Marsha Scar-
and Consumer Perceptionsof Quality,"Journal of Marketing, brough (1989), "Courtyardby Marriott:Designing A Hotel Fa-
57 (October), 18-34. cility with Consumer-BasedMarketingModels,"Interfaces, 19
(January-February), 25-47.
Tse, David K. and Peter C. Wilton (1988), "Models of Consumer
Zahorik,AnthonyJ. and RolandT. Rust (1992), "Modelingthe Im-
Satisfaction Formation:An Extension,"Journal of Marketing
pact of Service Qualityon Profitability:A Review,"in Advances
Research, 25 (May), 204-12. in Service Marketingand Management,T. Swartz, ed. Green-
Training (1991), "The Post-Deming Diet: Dismantling a Quality wich, CT: JAI Press, 247-76.
Bureaucracy,"28 (February),41-43. Zeithaml, ValarieA., LeonardBerry, and A. Parasuraman(1993),
Westbrook,RobertA. and RichardL. Oliver (1991), "The Dimen- "The Nature and Determinantsof Customer Expectations of
sionality of ConsumptionEmotion Patternsand ConsumerSat- Service," Journal of the Academy of Marketing Science, 21
isfaction,"Journal of ConsumerResearch, 18 (June), 84-91. (Winter), 1-12.

lir NE ll

When Marketersneed a solution they begin here:

The 1996 American Marketing Association
Yellow Pages & International
Membership Directory
Ifreachingan audience of literallythousands of marketingprofessionalsis an
importantpart of yourfirm'scommunicationsgoals, the AmericanMarketing
AssociationYellowPages is the place to be. Placinga listingor display
advertisementinthis annualbuyers'guideensures thatyourcompany's
message is deliveredto an influential
and highlytargetedreadership.

By actingnowyou can be includedin our1996 publication and save money

at the same time.Justcall KenPfeiferat 312-831-2729foran orderform.
Ifyou reservespace foryourad and/orlistingby May5, 1995,you
can takeadvantageof ourcurrentlowerrates.

A few copies of the 1995 YellowPages are stillavailable.Ifyou'dliketo

hi
'N ordera copyof this publication forjust$125, call 1-800-AMA-1 150.
A Po,
^-o

70 / Journalof Marketing,April1995

This content downloaded from 128.6.218.72 on Sun, 25 Oct 2015 11:47:44 UTC
All use subject to JSTOR Terms and Conditions