DELIGHTFUL OR DEPENDABLE? VARIABILITY OF CUSTOMER

DELIGHTFULORDEPENDABLE?

VARIABILITYOFCUSTOMEREXPERIENCESASAPREDICTOROFCUSTOMERVALUE



YanliuHuang

GeorgeKnox

DanielKorschun

*



WCAIProposal



December2012



Abstract

Isitpreferableforacompanytooccasionallysurpriseitscustomerswithexceptionalservice(i.e.,delight

customers)ortostriveforconsistencyinserviceacrosstimeandspace(i.e.,bedependable)?This

proposalinvestigatestheextenttowhichvariabilityinserviceencountersaffects

futurecustomer

satisfaction,likelihoodofupselling,andcustomervalue.Drawingfromtheexpectationsandservice

qualityliteratures,wedevelopamodelthatincludesthreepotenti altypesofvariabilityinacustomer‐

companyrelationship:(1)eachcustomer’shistoricalsatisfactionwithHertz,(2)employee’ssatisfaction

ataHertzsite,and(3)other

customers’recentsatisfactionlevelsataHertzsite.Byexaminingboth

satisfactionandvariabilityofsatisfaction,thisresearchcontributestotheliteraturebyprovidinga

nuancedpictureofthelinkbetweensatisfactionandcustomervalue.Formanagers,theresearchwill

helpdeterminewhethertoallocatescarceresourcestoexceptionalorconsistent

servicequality;for

example,shouldincentivesbegiventothebest(i.e.,exceptionalemployees)ortothosewhoarethe

mostconsistentwithinthesiteandcompany?





Yanliu Huang ([email protected]), George Knox ([email protected]), and Daniel Korschun (dek4[email protected]du)

are assistant professors of marketing at the LeBow College of Business, Drexel University. Please address

correspondence on this proposal to George Knox.

DELIGHTFULORDEPENDABLE?

VARIABILITYOFCUSTOMEREXPERIENCESASAPREDICTOROFCUSTOMERVALUE



Introduction

Thecurrentmantraatmanycompaniesisto“delight”customersbysurprisingthemwithagreat

experience.Suchaviewwouldpredictthatpunctuatingaseriesofexperienceswithafewexceptional

encounterswillleadcustomerstodeepentheirrelationshipwiththecompany.Thiswilloccurbecause

thedelightful

experiencewillmakeservicequalitymoresalientandgreatlyexceedpriorexpectations

(RustandOliver2000).Incontrast,somebrandmanagementtheorists(BarwiseandMeehan2010)

pointtotheimportanceofreliablydeliveringonabrand’spromisestocustomers.Theypredictthata

companythatachievessuchconsistencyisbestpositioned

tohavecustomersdeepenthecustomer‐

companyrelationship.

Thisresearchaskswhetheritispreferable–fromHertz’perspective–forcustomerstohaveafew

exceptionalexperiences(i.e.,delight)ortohavemorepredictableexperienceswiththebrand(i.e.,

dependability).Asaresult,itprovidesanuancedpictureof

thelinkbetweensatisfactionandcustomer

value.TheprojectwillalsoprovideguidancetoHertzmanagersonwhetheritisbettertoconcentrate

ondelightfulexperiences(bysay,rewardingemployeesthatprovidetrulyexceptionalservice)or

consistentservicequality(bysay,concentratingonpoorlyperformingemployeesandmakingsurethat

highly

motivatedemployeesactconsistentlywiththebrand).

Inthefollowingsections,weexplainthelogicbehindourtheoreticalmodelandprovideanoverviewof

ourmethodfortestingthemodel.



DetailedProjectProposal



Thepost‐encountersatisfactionofacustomerisoftenexplainedbycitingfactorssuchasthe

qualityof

servicedelivered,thequalityofserviceatonesiteversusanother,ortheoverallfriendlinessof

employees.Asaresult,manycompaniesstrivetoenhancecustomersatisfactionandcustomervalueby

improvingtheaveragelevelofserviceandtheaveragelevelofemployeesatisfactionateachsite.

Butadvances

inservicetheorysuggestthattheremaybemoretothestory.Researchoncustomer

expectationspointsoutthatsatisfactionisconstructedwhencustomerscomparecurrentservicequality

withthatwhichwasexpectedbasedonpriorexperi ences(Bouldingetal.1993;Parasuramanetal.

1994).Whencurrentservicequalityexceeds

expectations,expectationsaresaidtobeconfirmed,but

whencurrentservicefallsshort,expectationsaresaidtobedisconfirmed.Thisdynamicviewreveals

thatvariabilityislikelytoleadacustomertoreassesstherelationshipbecausevariabilityissomewhat 

destabilizingforexpectations.

Thereareatleastthreetypesofvariabilitythatcaninfluencesatisfactionandcustomervalue.Time

variabilityreferstohowvariedasinglecustomer’sencounterswiththecompanyhavebeenovera

periodoftime.Employeevariabilityinvolvesvariabilitystemmingfromthefactthatcustomerscan

interactwithmultipleemployeesin

asingleencounterwiththecompany.Sitevariabilityinvolveshow

consistentlycustomersaresatisfiedatthesite.

Wenowexplaineachofthesetypesofvariabilityinmoredepth,providingthetheoreticallinkbetween

variabilityandcustomersatisfactionandcustomervalueineachcase.

Timevariability

Somecustomersmayhave

veryconsistentexperiencesovertime,whileothersmayhavegreater

variability,withsomepositiveandsomenegativeexperiences.Forexample,twocustomersmayhavea

seriesofencountersthatonaverageareindistinguishable,yetonehashadveryconsistentencounters,

whiletheotherhashadaseriesofeitherverypositive

orverynegativeencounters.Weasktowhat

extentthevariabilityofprio r experiencesaffectscustomersatisfactionandcustomervaluein

subsequentperiods.

Ourexpectationisthatcustomersatisfactionandcustomervaluewilldecreaseasvariabilityinprior

customerexperiencesincreases.Thiswilloccurbecause,asnotedabove,priorexperiences 

thatdiffer

fromeachotherandfromthecurrentexperiencewillleadthecustomertocontinuallyreassessthe

customer‐companyrelationship.Suchreassessmentsmayencouragethecustomertoconsider

alternativeofferingsateachencounter.Asaresult,weexpectahigherdefectionrateasvariability

increases.Thispredictionisalsoconsistent

withdecisiontheory,whichholdsthatcustomersmake

purchasedecisionsbasedontheextenttowhichthosedecisionswillreduceuncertainty(e.g.,Rust,

Inman,JiaandZahorik1999).

Employeevariability

Theinternalmarketingliteraturehasfordecadesstudiedthelinkbetweenemployeesatisfactionand

customersatisfaction(George1990;Grönroos1981).

Theevidencetendstosupportapositiveand

casuallinkwherebysatisfiedemployeesaremorelikelytodeliverexcellentservicetocustomersand

createhighcustomersatisfaction.However,mostofthisresearchexaminestheeffectofaverage

employeesatisfactiononcustomersatisfaction.

Inpractice,employeesatisfactionmayvarysubstantiallyevenwithin

asingleworksite,andasingle

customermayinteractwithmultipleemployeesinasingletransa ction.Wepredictthatallotherthings

equal,increasesinvariabilityinemployeesatisfactionwillharmcustomersatisfactionandcustomer

value.Thiswilloccurforatleasttworeasons.First,acustomerthatencounterstwo

employeeswith

verydifferentlevelsofsatisfactionmayreceivedifferentservicequality.Thisvariabilityinservicewill

makethetransactionlesscertainandtherefore,moreriskytothecustomer.Second,highlysatisfiedand

dissatisfiedworkersmaybelesslikelytocooperatewithoneanotherbecausetheirgoalsmaydiffer

substantially.This

willcreatearelativelynegativeserviceclimateoremployees’sharedperceptionsof

thepractices,which,inturn,willleadtodecreasedcustomersatisfaction(e.g.,Salanova,Agut,andPeiro

2005).

Sitevariability

Wefinallyproposethatasitewithmorevariedcustomersatisfactionwillcreatelowercustomervalues

thanasitewithmoreconsistentcustomersatisfaction,althoughthetwosites mighthavesimilar

average

customersatisfaction.Thisisbecausewithservicevariability,thosedissatisfiedcustomersare

morelikelytoengageinwordofmouth,which,inturn,willresultinloweredprofitability(Anderson

1998).Furthermore,mixedcustomerexperiencemightreflectvariedemployeesatisfactionlevels,

which,asdiscussedearlier,alsoinfluencecustomervaluesnegatively.

‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐

Insert

Figure1AboutHere

‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐

ModelSketch

Weplantoanswerthisquestionbyexaminingtheeffectsofbothaveragesatisfactionaswellas

uncertaintyinsatisfactionlevelsonfuturecustomersatisfaction,customerupselling,andcustomer

value.Theuniquedatasetallowsustomeasurebothaveragesatisfactionlevelsandvariancein

satisfactionusingthreedifferencesources:(1)eachcustomer’shistoricalsatisfactionwithHertz,(2)

othercustomers’recentsatisfactionlevelswiththesamesite,and(3)employee’ssatisfactionwithHertz

atthechosensite.Figure1summarizesourproposedmodel.

Wefocusonpredictingfuturetransactionsandcustomersatisfactionforthe

Hertzloyaltycardmembers

(forwhichthereexistlongitudinaldata).Thesettingisnon‐contractual:atsomepointintimethatis

unobservedbythefirm,customersmaybecomeinactive.Hence,theattritionprocessislatentandmust

beinferredfromprioractivity(KumarandReinartz2006).Theintuitionisthat

alapseintransactions

maybedueeithertoanactivebutinfrequentcustomer,oracustomerwhohasirreversiblydefected.

Westartwithadiscretetimelatentattritionmodel(e.g.,Fader,HardieandShang2010)that

probabilisticallycharacterizescustomersaseitheractiveorinactiveandextenditintwo

ways.

First,wespecifythetime‐v a ryingtransitionprobabilitythatcustomeriremainsactiveaftermontht(p

)

asafunctionoftheindependentvariablesofinterest–theaverageandvarianceofpreviously

experiencedsatisfactionlevelsofcustomeri,theaverageandvarianceofemployeesatisfactionlevelsat

thesitemostrecentlyvisitedbycustomeri,andtheaverageandvarianceofsatisfactionlevelsatthe



sitereportedbycustomersotherthani.





][][][][

,,,,,

tistististisititit

VarCSATAvgCSATVarESATAvgESATVarSATAvgSATfp 



Wemayspecifysomeoftheseindependentvariablesasstockvariablestoaccountforthestronger

weightplacedonmorerecentexperiences(e.g.,Bolton1998).Furthermore,wemayspecifythat

responsesdifferbyvalenceorlevels,toaccommodateasymmetriceffectsforpositivevs.negative

satisfactionshocks,orpossible“zonesof

indifference”(e.g.,Oliver1997).Toincorporateunobserved

heterogeneityintheresponseofremainingactivetothesevariablesweintendtouselatentclassesora

hierarchicalBayesprocedure.

Secondly,foractivecustomers,wemodelseveraleventsrelatedtothe“standard”probabilityof

conductingatransaction.Conditionalonacustomermakingatransaction,wecanmodelthe

transactionamount,

whethertherewasany“upselling”(definedinthepresentation),whetherthe

customerdecidedtogiveasatisfactionscore,and,conditionalonprovidingthescore,thescoreitself.It

isdesirabletoallowthesedecisionstoberelated;forexampleupsellingmaybeassociatedwithlower

satisfactionratings,or,notproviding

asatisfactionresponse.Weintendtolinkallthesedecisionsis

usingaGaussiancopulaapproach(Danaherand Smith2011),whichlinksthemarginaldistributions.

References:

Anderson,EugeneW.(1998),“CustomerSatisfactionandWordofMouth,”JournalofServiceResearch,

1(August),5‐17.

Barwise,PatrickandSeanMeehan(2010),

"TheOneThingYou MustGetRightWhenBuildingaBrand,"

HarvardBusinessReview,88(12),80‐84.

Boulding,William,AjayKalra,RichardStaelin,andValarieA.Zeithaml(1993),"ADynamicProcessModel

ofServiceQuality:FromExpectationstoBehavioralIntentions,"Journalof MarketingResearch,30,7‐27.

BoltonR.

N.,LemonK.N.&BramlettM.D.(2006),“TheEffectofServiceExperiencesoverTimeona

Supplier’sRetentionofBusinessCustomers”,ManagementScience,52(December),1811‐1823.

Danaher,PeterJ.andMichaelS.Smith (2011),“ModelingMultivariateDistributionsUsingCopulas:

ApplicationsinMarketing,”MarketingScience,30(1),

4‐21.

Fader,PeterS.,BruceG.S.HardieandJenShang(2010),“Customer‐BaseAnalysisinaDiscrete‐Time

NoncontractualSetting,”MarketingScience,29(6),1086‐1108.

George,WilliamR.(1990),"InternalMarketingandOrganizationalBehavior:APartnershipinDeveloping

Customer‐ConsciousEmployeesatEveryLevel,"Journalof BusinessResearch,

20(1),63‐70.

Grönroos,C.(1981),"InternalMarketing–anIntegralPartofMarketingTheory,"inMarketingof

Services,AmericanMarketingAssociation,Chicago,IL,J.H.DonnellyandW.E.George,eds.

Kumar,V.andWernerJ.Reinartz,CustomerRelationshipManagement:ADatabaseApproach.John

WileyandSons,NewYork.

Parasuraman,A.,Valarie

A.Zeithaml,andLeonardL.Berry(1994),"ReassessmentofExpectationsasa

ComparisonStandardinMeasuringServiceQuality:ImplicationsforFurtherResearch,"Journalof

Marketing,58(1),111‐24.

Rust,RolandT.andRichardL.Oliver(2000),“ShouldWeDelighttheCustomer?”JournaloftheAcademy

ofMarketingScience,

28(1),86–94.

Rust,R.T.,J.J.Inman,J.JiaandA.Zahorik(1999),“WhatYouDon’tKnowAboutCustomer‐Perceived

Quality:TheRoleofCustomerExpectationDistributions.”MarketingScience,18(1):77‐92.

Salanova,Marisa,SoniaAgut,andJoséMaríaPeiró(2005),“LinkingOrganizationalResourcesandWork

EngagementtoEmployeePerformance

andCustomerLoyalty:TheMediationofServiceClimate,”

JournalofAppliedPsychology,90(6),1217‐1267.



Biographies



YanliuHuangisAssistantProfessorofMarketingatLeBowCollegeofBusiness,DrexelUniversity.Her

researchinterestsfocusonconsumerdecisionmakingincludingin‐storeshoppermarketing,consumer

planning/learning/memory,andhealthmarketing.Shewillbringtothisprojecttheexpertiseon

consumerbehaviorandconsumerdecisionmakingprocess.



GeorgeKnox

isAssistantProfessorofMarketingatLeBowCollegeofBusiness,DrexelUniversity.His

researchinterestsarecustomerrelationshipmanagement,marke tingeffectiveness,andin‐store

retailing.Hewillbringthemodelinganddataanalysisexpertisetotheproject.



DanielKorschunisAssistantProfessorofMarketingatLeBowCollegeofBusiness,Drexel

University.His

researchisoncorporatesocialresponsibility(CSR),witharecentemphasisonhowCSRmotivates

employeestoservecustomerneeds.Hebringsexpertiseoninternalmarketingandcustomer

orientation.



Figure1

ConceptualFramework



users  listeners  fans  donors  viewers  households  shoppers  sellers  readers  browsers  friends  followers

travelers  patients  contributors  attendees  readers  subscribers  buyers  clients  visitors  guests  customers

Page ‹#›

DELIGHTFUL OR DEPENDABLE?

VARIABILITY OF CUSTOMER EXPERIENCES AS A PREDICTOR OF CUSTOMER VALUE

OVERVIEW

Is it preferable for Hertz to occasionally surprise its

customers with exceptional service (i.e., delight customers)

or to strive for consistency in service across time and space

(i.e., be dependable)?

Professors Yanliu Huang, George Knox, and Daniel

Korschun (Drexel University) quantify the effects of variability

on customer satisfaction, upsell potential, and customer

lifetime value.

This unique dataset enables the researchers to examine

three types of variability:

• Customer variability: variability of a customer’s

historical satisfaction with Hertz

• Employee variability: variability of employees’

satisfaction at a Hertz site

• Site Variability: variability of other customers’ recent

satisfaction level at a Hertz site

MANAGERIAL IMPLICATIONS

• Reach out to customers with variable (but on average

good) service, as well as those who received poor service

• Reward employees for consistency versus exceptional

service

• Track consistency at each site (in addition to overall

satisfaction)

MODEL MECHANICS

We propose a latent attrition model where customers may

become irreversibly inactive at some point in time. We make

the transition probability from being an active customer to an

inactive one as a function of our six main drivers of interest

(see Figure above):

Behaviors (and perceptions) while active are linked via a

Gaussian copula.

VariabilityOfCustomerExperiencesAsAPredictorOfCustomerValue





][][][][

,,,,,

tistististisititit

VarCSATAvgCSATVarESATAvgESATVarSATAvgSATfp 