MITSloan：從品牌對(duì)話中創(chuàng)造價(jià)值 Deriving Value From Conversations About Your Brand 2024

BradFayEdKeller

RickLarkinKoenPauwels

DerivingValue

From

Conversations

AboutYourBrand

Researchshowsthatbothonlineandoff-linecustomer

conversationsdrivepurchasedecisions—buttheyrequireseparatemarketingstrategies.

Reprint#60210

/2zEc7M2

MITSLOANMANAGEMENTREVIEW

MARKETING

DerivingValueFromConversationsAboutYourBrand

BRADFAY,EDKELLER,RICKLARKIN,ANDKOENPAUWELS

Researchshowsthatbothonlineandoff-linecustomerconversationsdrivepurchasedecisions—buttheyrequireseparatemarketingstrategies.

Nordstrom,theSeattle-basedretailer,hadamemorable2017.InearlyFebruary,DonaldTrump,thenthenewly

electedU.S.president,tooktoTwittertoberate

NordstromfordroppingtheIvankaTrumpclothingline,complainingthatthecompanyhadtreatedhisdaughter“sounfairly…terrible!”Thetweetsetoffapowerful

reactioninsocialmedia.Ourresearchshowedthe

numberofweeklymentionsoftheNordstrombrandonTwitterandothersitessurgedby1,700%,whilethetone

ofthoseconversations(asmeasuredusingnatural-

languageprocessing,whichinterpretsmeaningfrom

adjacentwordsandcontext)swungsharplyfrompositivetonegative.

1

However,inof-lineconversations

(measuredviasurveys),thesentimentstayedpositive.Amidstthesemixedsignals,Nordstromrolledthroughthe2017holidayseasonwitha2.5%salesincreaseovertheprioryear.

Divergentconversationsaboutbrandsarefairlycommon

—andnotonlyforbrandsthatgetcaughtupin

controversies.

2

Indeed,westudiedmorethan500

leadingconsumerbrandsandfoundthatinmostcasestherewaslittlecorrelationbetweenwhatconsumerssaidaboutthebrandsonlineandwhattheysaidoff-line,eventhoughbothstreamsofconversationcanhavebigeffectsonacompany’ssales.

3

Marketershavelongrecognizedwordofmouthasa

powerfulforceaffectinghowwellproductsperform.

SincetheadventofTwitterandFacebook,somepeople

nowthinkofsocialmediaas“wordofmouthonsteroids”

—theconversationthatrepresentswhatconsumersaresaying.

4

Yetwefoundthatonlineandoff-line

conversationsmatterfordifferentreasons.

Copyright?MassachusettsInstituteofTechnology,2018.Allrightsreserved.Reprint#60210

/2zEc7M2

MITSLOANMANAGEMENTREVIEW

MARKETING

Moststudiesonsocialmediamarketingeffectiveness

havelookedathowbrandengagementonspecific

platformssuchasFacebookorTwitter(forexample,thelikes,shares,retweets,andcomments)respondsto

marketinginitiativesasopposedtoconsideringthesocialecosystemasawhole.Thereislittleresearchlookingat

off-lineconversations—thosethatoccurface-to-faceattheofficewatercooler,overthekitchentable,orata

healthclub—becauseofthedifficultyandcostof

measuringthem.However,weaddressedthatchallengebyaskingselectedconsumerstorecalltheproductand

servicecategoriesandbrandstheytalkedabouttheday

before,includingwhetherthebrandconversationswerepositiveornegative.Weexaminedthissurveydatafor

off-lineconversationsalongwithsocialmediadataso

thatwecouldcomparethetwotypesofconversationsandidentifytrendsinboth.Wealsolookedatweeklyad

expendituresforspecificbrandsfromNumerator,an

advertisingtrackingcompany,andsalesdatatocreateacomprehensivepictureofthefactorsthatleadconsumers

tobuycertainbrands.(See“AbouttheResearch”)

AbouttheResearch

Forourresearch,wedevelopedaproprietarydata

platformtoincorporateonlineandoff-lineconversationdataon501U.S.brands.Fortheanalysispresentedhere,wecollectedonlinedatafor2015and2016andoff-linedatafor2013through2016.Onlinedatawascollected

throughkey-wordsearchesofTwitter,publicFacebookposts,blogs,forums,andconsumerreviewsites.Usingnatural-languageprocessing,weanalyzedwhetherthe

conversationswerepositiveornegative.Ourcontinuoussurveyresearchprogramyieldeddataonbrandsfromanaverageof7,000off-lineconversationsperweekwith

consumersages13to69.Respondentswereaskedto

reportonwhethertheirconversationabouteachbrand

waspositive,negative,neutral,or“mixed”O(jiān)urinitialstepwastocorrelatetheonlineandoff-linedatastreamsfor

allbrands.Wethendidaregressionanalysistolinktheonlineandoff-lineconversationstothird-partyweeklysalesdatathatweacquiredfor175brands,andtoweeklyadexpendituredataforasubsetof21ofthosebrands

usingamethodknownasmarketmixmodeling.

i

Weusedthisapproachtostudytherelationshipbetweenonlineandoff-lineconversationsin15industries,

includingelectronics,packagedfoodsandbeverages,

telecommunications,finance,andtravel.Formanyofthe500brands,wewereabletoobtainthird-partysalesdata,andwepaidparticularattentiontoasubsetof21brands—includingApple,Intel,A&W,Budweiser,Campbell’s,Lay’s,Pepsi,RedBull,andRevlon—forwhichwewerealsoabletoobtainadvertisingdata.

Ouranalysisshowsthateventhoughonlineandoff-lineconversationsbothdrivesales,theyoperate

independentlyfromeachother,sotheyneedtobe

measuredandmanagedseparately.Indeed,managers

can’trelysolelyonsocialmediatorepresenttheentiresocialecosystemthataffectsbrandsuccess.Wedescribeourfindingsingreaterdetailbelowandexplainthe

implicationsforcompanies’marketingefforts.

HowCustomer

ConversationsAffect

Sales

Knowinghowpeoplemakebrandandpurchasedecisionscanbetremendouslyvaluableforcompanies,particularlythosethatrelyheavilyonnew-customeracquisition.U.S.consumersspendapproximately$51trillioneachyearon

Copyright?MassachusettsInstituteofTechnology,2018.Allrightsreserved.Reprint#60210

/2zEc7M2

MARKETING

allmannerofgoodsandservices,fromsoftdrinksandmobilephonestoairlineticketsandautoinsurance.

5

Giventhepotentialpayoff,companiesarealwayslookingforwaystoinfluencethosechoices—andconsumer

conversationsandrecommendationsrepresentamajoropportunity.

Overall,wefoundthatoff-lineandonlineconversationshadsimilarimpactsonsales.Forthe21brandswecloselystudied,wefoundthat9%ofpurchasedecisionscouldbetracedbacktopublicconversationsandengagementthatoccurredinsocialmedia(includingFacebook,Twitter,

Instagram,YouTube,blogs,andcustomerforums).A

slightlylargershare—10%ofsales—wasrelatedtooff-lineconversationsasmeasuredthroughourcontinuoussurveys.Thatmeanssome19%ofU.S.consumer

purchasescouldbetracedtoconsumerstalkingto

friends,relatives,colleagues,orothers(someofwhomtheyknewonlythroughsocialmedia)aboutbrands.

Yetwesawnotabledifferencesacrossproductandservicecategories.(See“WhatYou’reSellingMakesa

Difference”)Forexample,wehadassumedthat

technology-orientedofferingswouldskewmoretowardonlineconversationsthaneverydayconsumergoods,andthatproductslikefood(whichareoftenconsumedin

socialsituations)mightbehighlyinfluencedbyoff-line

conversations.However,wefoundthatsalesoftech

brandslikeAppleandIntelweredrivenmorebyoff-lineconversationsthanonline,whilesalesofgroceryandfoodbrandssuchasCampbell’stendedtobedrivenmoreby

socialmediathanbyoff-linerecommendationsfromfriends.

WhatYou’reSellingMakesaDifferenceAlthoughmanymarketersarefocusedonconsumers’

socialmediaactivity,insomecategoriesoff-line

conversationscanbeevenmoreinfluentialindrivingpurchases.

Source:EngagementLabseconometricmodelfor21leadingbrands

Thepricepointofaproductorservicewasoftenakey

factor.Higher-pricedgoodsandservicesweremoreapttobeinfluencedbyoff-lineconversations,perhapsbecausethestakeswerehigherandoff-linediscussionspermitteddeeperexplorationofabrand’sprosandconsthanonline.Significantly,though,foreveryproductcategorywe

studied,theratiobetweenonlineandoff-linewasnevermorelopsidedthan60-40ineitherdirection,meaningthatbothtypesofconversationsmatteredalot.

Whileonlineconversationsaremorevisible,off-line

conversationsaremoreplentiful.Oursurveyrevealsthattwo-thirdsofpeopletalkaboutbrandswithatleastonefriend,relative,orneighboronanygivenday,whereas

only7%post,tweet,write,orcommentaboutthe

productstheyuse.Onlineconversationstendtobeabout“socialsignaling”toone’snetwork,atermacademicsusetodescribethemotivationbehindpostsabouthightech

Copyright?MassachusettsInstituteofTechnology,2018.Allrightsreserved.Reprint#60210

/2zEc7M2

MITSLOANMANAGEMENTREVIEW

MARKETING

andhighfashion.

6

Particularlywhenindividualsaretryingtoappealtoalargegroupoffriendsand

acquaintances,theycrafttheironlinemessagestoshowtheyareontheleadingedgeofatrend.Off-line

conversations,bycontrast,focusononepersonandare

aboutvariousproductsandservices,manyofwhicharen’t“sexy”enoughtotweetaboutormentiononFacebook.

TheMetricsThatMatter

Most

Forthebrandswestudied,themostinfluentialmetric

wasoff-lineconversationvolumeasrepresentedbythenumberofconversationspeoplehaveaboutthebrandin

aweek.Themoreface-to-faceconversationsand

recommendations,thebetteritisfordrivingabrand’s

sales.Forexample,eachAugust,Lay’srunsapromotionalcampaigncalledDoUsaFlavorinwhichconsumersareaskedtovoteonanewpotatochipflavor.Thecampaigntypicallygeneratesasurgeinoff-lineconversationsthatcontinueslongafterthecampaignends.Theactivity

drivesbrandengagementandpurchasesatthestartoftheU.S.professionalandcollegefootballseason,when

friendsandfamilywatchtelevisedgamestogetherandchipconsumptionrises.

Thesecondmostinfluentialmetricwas“onlinenet

sentimentwhichwecalculatedbysubtractingthe

percentageofnegativesocialmediaconversationsaboutthebrandfromthepercentageofpositiveones.Thiswasfollowedcloselyby“off-linenetsentimentwhichwe

calculatedsimilarly,usingdatafromourconsumer

survey.Theimportanceofthetwonetsentimentmetricssuggeststhatcompaniesshouldpaycloseattentionto

whethertheirbrandsarebeingtalkedaboutpositivelyornegatively,whetheronlineoroff-line.

Howcancompaniesmanagethesemetrics?RedBull,theenergydrinkmaker,offersagoodexample.Wefound

thatiteffectivelydrovepositiveconversationsonsocialmediabycreatinghighlyentertainingandshareable

videosofathleticachievement.Butthebrandhasbeenlesssuccessfulatproducingpositiveoff-line

conversations.Identifyingwaystoencouragemorepositiveface-to-faceconversationsmayprovidenewopportunitiesforRedBulltodrivesalesgrowth.(See“WhatHastheBiggestImpactonSales?”)

WhatHastheBiggestImpactonSales?Foroff-lineconversations,themostimportantmetricwasthevolume,orquantity,ofconversations.Foronline

conversations,sentimentmatteredmore.

Source:EngagementLabseconometricmodelfor21leadingbrands

Onlineandoff-linesentimentoftenmoveinopposite

directions.InthewakeoftheFebruary2018school

shootinginParkland,Florida,Dick’sSportingGoods,thelargesportinggoodsretailchain,tighteneditsgunsale

policiesandceasedsellingassaultweapons.Theonline

reactionwasextremelynegative—peopleconcerned

withgunrightsdenouncedthecompanyonsocialmedia.Buttheoff-linesentimentwaspositive.Infact,the

company’srevenuesrose,andthestockjumpedmore

Copyright?MassachusettsInstituteofTechnology,2018.Allrightsreserved.Reprint#60210

/2zEc7M2

MITSLOANMANAGEMENTREVIEW

MARKETING

than20%upontheannouncementoffirst-quarterresultsin2018.Thisunderlinedtheimportanceofmonitoringbothformsofsentiment—onlineandoff-line.

7

Yetanothermetricworthtrackingistheextenttowhichthebrandisbeingtalkedaboutoff-lineby“influencers”—peoplewhoregularlygiveconsumeradvice.Brands

canleveragetheirmarketpositionbyencouraging

influencerstosharewhattheylearnaboutproductswithfriendsandfamily,expandingthereachofmarketingandthespeedofadoption.

8

Nintendoemployedthis

strategywhenittargetedandcultivated“alphamoms”tointroducetheoriginalWiigamingconsolesthroughtheirreal-worldsocialnetworks.

9

Onemetricthatdidn’tseemtohavemuchinfluence,at

leastonthesurface,was“brandcontentsharingwhichmeasuresthedegreetowhichconsumerstellustheyaretalkingaboutbrandadvertising(off-line)orhavepressedthesharebuttonforbrandcontent(online),astheydid

withRedBull’sshareablevideos.Althoughwefoundthatthemetricrankedlowinitsimmediateimpactonsales,itseffectwaslarger:Ourmodelshowedthatadvertisingexpendituresdroveconversations,which,inturn,ledtosales.Indeed,conversationsamongpeoplewhoknowandgenerallytrustoneanotheraddpersuasivepowertotheadvertisingthatsparksthoseconversations.

10

Implicationsfor

Marketing

Marketershaveknownforyearsthatcustomer

conversationsandrecommendationsarepowerfulformsofbrandengagement.

11

Buthowcancompanies

leveragethoseconversationsonbehalfoftheirbrands?

Broadlyspeaking,managersshouldlookforwaystodrivemorepositiveconversationsbothonlineandoff-line.Inmanycases,thiswillmeangoingbacktomarketing

fundamentals—rethinkingproductdesign,market

segmentation,customerservice,messaging,andchannelselection—withsocialstrategyinmind.Toaccomplishthisefficientlyandeffectively,wesuggestfoursteps.

1.Determinewhichconversationdrivershavethe

greatestpotentialforyourbusiness.Largecompanies

withanalyticsdepartmentsmaywanttobuildastatisticalmodelaswedid,tolinkconversationdatatobusiness

results.Butcompaniescanuseotherapproachestofigureoutwhichmetricstheyshouldfocuson.Forexample,youcanlearnalotfromyourcompany’sonlineconsumer

reviews,aswellasthoseofkeycompetitors.Ifthereviewsarealreadylargelypositive(forexample,4.5outof5

stars),youcantrytostimulatemoreconversationvolume—particularlythroughoff-linerecommendations.Ifyouronlinereviewsarelesspositivethanthoseofcompetitors,trytoimprovethem.Reachouttocustomerswhogave

youso-soratings,askingthemwhatyoucoulddobetter,andinvitesatisfiedcustomerstosharetheirexperiencesthroughonlinecommentsortweets.Youcanalso

conductinexpensiveonlinesurveystolearnabout

customers’off-linerecommendingbehavior(forexample,howoftentheyrecommendproductsandwhattheysay).Inaddition,itmightbeworthwhiletoexaminecall-

centerdataforthemesthatcorrelatewithcustomersatisfactionanddissatisfaction.Iftherootcausesofdissatisfactionaresubstantial,trytofixthem,andencouragethosewhoaresatisfiedtosharetheir

experienceswithothers.Manytravelandrestaurant

businessesdothisbygivingfriendlyremindersto

recommendthemtoothersrightafteragreatexperience

Copyright?MassachusettsInstituteofTechnology,2018.Allrightsreserved.Reprint#60210

/2zEc7M2

MARKETING

orbyprovidingcustomerswithincentivesforreferrals.Companiesinotherindustrycategoriescandothesame.

2.Identifytheconsumersegmentsthataremostlikelytoenhancesalesperformance.Onceagain,thereareseveralwaystodothis.WorkingwithamajorfinancialcompanyintheUnitedKingdom,forexample,weminedlarge

databasesforinsightsandfoundthatfocusingonthe

needsofaffluentwomenandtargetingtheminmarketingwerethekeystogeneratingtheconversationsthatledtonewaccounts.Alternatively,somecompanieshavefounditusefultoconductsmall-scalesurveysofexisting

customerstoidentifywhoisrecommendingproductsorservicesmostoften,andwhy.Havingthisinformationcanhelpyoudeepenrelationshipswithinfluencers

throughevents,customercareofferings,andotherinitiativesandreachotherprospectsthroughthem.

Prioritizingcustomerswhosimplyhavelargesocial

networkscanalsobeworthwhile;theymightbeabletorecruitothercustomerstoyourbrandbothonlineandoff-line.Wehavefoundthatahighpercentageof

consumerswhocontactabrandviaitswebsite,socialmedia,orcallcentersareinfluencers.

12

3.Refineyourmessagingandstimulateconversation.

Developingshareablemarketingcontentrequires

creativityaswellasstatisticalanalysis.Oneapproachistomonitorsocialmediadiscussionsaboutyourbrandto

identifythemes,andeventhelanguageconsumersare

usingwhentalkingaboutyourproductsandthemarketingeneral,andthenconductasmallsurveytolearnhowthoseideasresonateoff-line.Thenyoucanusethemostcompellingthemesandlanguagewhendeveloping

creativemessaging.Anothermethodistofindawaytosparkaconversationthatislikelytogoviral.A2010

televisioncommercialforOldSpice,thepersonalcare

brand,providesagoodexample.Inthead,the

spokesman,afitex-NFLplayer,tellsviewersthatwhileeverymancan’tlooklikehim,theycansmelllikehim.TheadwassharedmillionsoftimesoverYouTubeandcontributedtoan11%increaseinOldSpicesales.

4.Optimizeyourconsumertouchpointstosupportyourconversationstrategy.Companiesoftenassumethatthebestavenueforincreasingconversationabouttheir

offeringsisthroughsocialmedia.Butaswehaveshown,conversationsspringfromavarietyoftouchpointsa

consumermighthavewithabrand,socompaniescandomanyotherthings.Forexample,in-storeproductdisplayscaninviteconsumerstotakeselfiestosharethroughtextmessageswithfriends.Couponscanbedesignedto

generatebonussavingsforcustomersshoppingwith

friends.Emailmarketingcampaignscanencourage

peopletoforwardmessagestofriendsorfamilymembers.Thekeyistoencouragetalkingandsharing,andtodoitthroughthevariouschannelsyoucanusetointeractwithyourconsumers.

Mostbrandsthatattempttofollowconsumer

conversationschoosetoconcentrateonsocialmedia.

However,aswehavenoted,thiscanpointyouinthe

wrongdirection.HadNordstromexecutivestakenthe

onlineconversationaboutIvankaTrump’sclothinglineasgospel,forexample,itmighthavemadedecisionsthat

wouldhavehurtsalesinsteadofboostingthem.Bothoff-lineandonlineconversationscanhaveasignificant

impactonacompany’stopline.Understandingthevaluethateachtypeofconversationmayprovide—andhow—canhelpbusinessesdevelopsmartermarketingstrategiesandmaketargetedinvestmentsthatleadtogrowth.

Copyright?MassachusettsInstituteofTechnology,2018.Allrightsreserved.Reprint#60210

/2zEc7M2

MITSLOANMANAGEMENTREVIEW

MARKETING

AbouttheAuthors

BradFayisthechiefcommercialofficerandEdKelleristheCEOatEngagementLabs,adataand

analyticsfirminNewBrunswick,NewJersey,whereRickLarkinisvicepresidentforanalytics.KoenPauwelsisaprofessorofmarketingatNortheasternUniversity’s

D’Amore-McKimSchoolofBusinessinBoston.

References

1.C.ManningandH.Schütze,

“FoundationsofStatisticalNatural

LanguageProcessing”(Cambridge,MA:MITPress,1999).

2.B.FayandR.Larkin,“WhyOnline

Word-of-MouthMeasuresCannotPredictOfflineOutcomesJournalofAdvertisingResearch57,no.2(June2017):132-143;B.

Fay,“Dick’sSportingGoodsProvesthe‘Noise’ofSocialMediaCanGivean

IncompleteSignalJune14,2018,

.

3.J.Morrissey,“BrandsCloselyMonitorSocialMedia,butOfflineChatterIsJustas

ImportantTheNewYorkTimes,Nov.27,

2017;and“

ReturnonWordofMouth

workingpaper,WordofMouthMarketingAssociation,September2015.

4.L.Geller,“WhyWordofMouthWorks

May13,2013,

;H.

Conick,“‘WordofMouthonSteroids’:

BrandsFindSuccessinPeer

Endorsements,StudyFindsMarketingInsights,April5,2016;andA.Lane,

“

WordofMouthonSteroids—

UnderstandingtheMotivesofSharing

Content

Nov.27,2017,

.au

.

5.TradingEconomicsandU.S.BureauofEconomicAnalysis,totalforfourquartersendingJuly2018,

.

6.M.J.Lovett,R.Peres,andR.Shachar,

“OnBrandsandWordofMouthJournalofMarketingResearch50,no.4(August2013):427-444;andA.BaraschandJ.

Berger,“BroadcastingandNarrowcasting:HowAudienceSizeAffectsWhatPeople

ShareJournalofMarketingResearch51,no.3(June2014):286-299.

7.B.Fay,“Dick’sSportingGoodsProvesthe‘Noise’ofSocialMediaCanGivean

IncompleteSignalJune14,2018;andW.Duggan,“GunRestrictionsDon’tDampenDick’sStockMay30,2018,132-143,

.

8.B.Libai,E.Muller,andR.Peres,

“DecomposingtheValueofWord-of-

MouthSeedingPrograms:AccelerationVersusExpansionJournalofMarketingResearch50,no.3(April2013):161-176.

9.D.Chmielewski,“MarketingMoms:

NintendoReachesOuttoaRelatively

UntappedSegmentofPotentialUsersinanEfforttoPromoteItsNewConsole

LosAngelesTimes,Dec.25,2006;andL.Richwine,“Disney’sPowerfulMarketing

Force:SocialMediaMomsReuters,June

15,2015.

10.This“two-stepflow”isconsistentwithworkthatgoesbacktothe1950s,whenitwasdevisedbyresearchersatColumbiaUniversityandtheUniversityof

Pennsylvania.SeeP.LazarsfeldandE.

Katz,“PersonalInfluence:ThePartPlayed

byPeopleintheFlowofMass

Communications