B2C中的客户终生价值分析及其应用研究

华中科技大学硕士学位论文B2C中的客户终生价值分析及其应用研究姓名：陈建玲申请学位级别：硕士专业：管理科学与工程指导教师：马辉民20061025IB2CB2CB2CB2CB2CMarkovChainModelRFMRecency,Frequency,MonetaryMarkovChainRFMRFMB2CB2CK—meansB2CMarkovChainRFMK—meansIIAbstractWiththerapiddevelopmentoftheInternet,B2Cispronetobeamorepersonalizedservice.Inordertomeettheneedsofcustomers’individualization,theseB2Cprovidersneedtobalancethecustomerpromotingexpenditureandtherevenuesfrompromotingactivities.Thensomeproblemsthatwhoareprofitablecustomers,howtoevaluatetheprofitstheyhavebroughtintheirlifetimeandhowtosegmentcustomerbytheirprofits,andhowtomakepromotingpoliciesaccordingtothesegmentation,essentiallyaffectedthesesB2Cproviders’success.Inordertosolvetheseproblems,thisarticleemploystheCustomerLifetimeValueModelintotheB2CcontextandproposesaCustomerLifetimePredictionModel,andthenappliesitintocustomersegmentation.Afterthat,theauthorputforwardssomemarketingstrategiesinaninstance,accordingtodifferentsegmentation.CustomerLifetimeValuecanbeseenfromboththecustomers’perspectiveandcompanies’perspective,whilethisarticleanalyzestheCustomerLifetimeValuemodelsanditsapplicationsfromthecompany’sperspective.OnthebasisoftheflexibilityandpredictivefeaturesofMarkovChainModelandtheanalysisofcustomers’historicalbehaviorfeaturesinRFM(Recency,Frequency,Monetary),theauthorcombinesthesetwomodelstogether.ByrefiningtheconventionalRFMmodel,thepredictionofthecustomer’svaluebecomesmoreprecise.Andthen,thisarticleusestheK-meansarithmetictodothecustomersegmentation,andproposesdifferentmarketingstrategiesaccordingtodifferentcustomergroups.Suchmethodhelpstomakeeffectivemarketingpolices.Attheendofthearticle,theauthorconcludestheexistingresearchmethods,andbringsforwardhisownideasonhowtoperfecttheCustomerLifetimeEvaluationModelandCustomerSegmentationinB2Cmarket.KeyWords:Business-to-CustomerCustomerLifetimeValueCustomerMarketingCustomerSegmentationMarkovChainRFMK--Means11B2CInternetInternetB2CDELLB2CInternet20%20%B2C1.1B2C1998InternetB2BB2CB2GC2GC2CB-B-CB2GC2G2B2CCNNIC20066301230026%200019.4%30001/450%[1]2004B2C4220055633%2005201052%20104602008B2C3800[2]B2C0.02%B2C90%80%(1)CRM(2)B2CB2C31.2B2C1986Courtheaux[3]Kim,Mahajan&Srivastava[4]Jackson1996[5]SnuilGupta,DonaldR.Lehmann&JenniferAmesStuart2001AmazonAmeritradeeBayETradeB2C[6]B2C(1)[7](2)(3)B2CB2CB2CB2C[8]B2C[9]B2C4B2CB2C1.3CustomerSegmentation[10][11][12]ZeithamlKotlerHeskettBlattbergDeighton[12](CustomerPerceivedValue)()[14]Zeithaml(1988)[15]Kotler(1984)(CustomerDeliveredValue)[16]5(1)(CustomerLifetimeValue)Brealey&Myers1996[17]BlattbergDeighton(1996)(CustomerEquity)[18](CustomerLifecycle)TnnRCniV(2)[19]0[(Rn-Cn)*(1+i)-n]nTnV==∑(2)B2CB2C61.4B2CB2CB2CB2CMarkovChainB2CB2CB2CB2CB2CB2CFRMMarkovChainB2CB2C72B2CB2BB2CSchmittlein[19]NBD/Pareto(SMC)Schmittlein[20]SMCSharma[21]Reinartz[22]SMCPfeifer[23]MarkovChainB2BB2C(1)(2)(3)B2CB2C2.1B2CCRMInternetB2C8[24]InternetB2CwebCRMCRM[25]ECRMCRM(1)CRM(2)CRM9(3)CallCenterCRM[26]B2CCRM2.2B2CCustomerLifetimeValueBrealey&Myers1996[28]B2CB2CB2C2.2.1CLV(1)0.51()(1)niiiiRCCLVd−=−=+∑(3)10iiRiiCiiRnNPV(2)Pareto/NBDSchmittlein,MorrisonColomboPareto/NBD1111111111111αβsα+Tβ+Tβ+TP(Alive/r,s,α,β,x,T,t)=1+()()(,;;())()(,;;())r+x+sα+tα+tα+t11α-β()α+yrxssFabcZtFabcZTarxsbscrxsZy−+×−=++=+=+++=,,α,βrstT111(,;;)FabcZGaussxTT0Aliveα=βαβSchmittleinPareto/NBD11(3)Pareto/NBDReinatz&Kumar[21]Reinatz&KumarPareto/NBDAlive/AliveDeadAliveAliveReinatz&KumarReinatz&Kumar3(4)BarbaraJacksonAlways-a-ShareLost-for-GoodAlways-a-ShareLost-for-GoodDwyerLost-for-Good4[31]Always-a-ShareDwyer12(5)Pfeifer&CarrawayMarkovChainModelMCM[23]CLVMCMPfeifer&CarrawayMCMMCMMCMMCM2.2.2RFMCLVRFMCLVRFMBultWansbeekRFM[41](1)(Re)Rcency:13(2)()FFrequency:(3)()MMonetary:B2CRTB2CFTB2CB2CMTRFMB2CB2CRFRFMRFMB2C2.3B2CB2CB2C[26]2.3.1B2CB2C()(14),[28],,,.,,B2C(1)(2)[29](3)B2C152.3.2B2C(1)(2)20%80%(3)K-meansB2C[28](4):;;;B2CB2C16173MarkovChainB2CB2CSessionCustomerLifetimeValue,CLVRFM(Recency,Frequency,Monetary)B2CPhillipRobertCLVRFMMarkovChainBergerNasr[43]BlattbergDeighton[44]MarkovChainModelMCMCustomerMigrationCustomerRetentionMCMMCMRFMMarkovChainRFM(Recency,Frequency,Monetary)MarkovChainB2CB2CMarkovChain3.1MarkovChainMarkovChainMarkovChain[39][40](1)AANC(2)DwyerBerger&NasrA(3)d(4)AActiveM(5)Arecency18Arecency1(6)ArprAArecency15recency=5recency3-13-1MCM11223344100001000010000100001ppppPpppp−−=−−3-2MCMAPArecency5PttMarkovChainttPttP(,)ijArecencyitrecencyjtP19A3.2RFMMarkovChainRFMMarkovChainModel(,,)rfm(1)TTt(2)ABCostNCdRr=11rn0rNCCostRCost−=−4(3)RecencyFrequencyPt=0PttPPt,tijPt(,,)rfm(4)TTVVT10V[(1)]TttdPR−==+∑5T11limV{(1)}VIdPR−−==−+6(1)MCM20(2)MCM(3)MCM3.3MCMB2CB2C(1)(2)(3)ProfileProfile[44]3.1MCMB2C(,,,,)buysbuysstateRRFFM=3.1MCMbuyRT1buyR=0T2buyR=nTnsRTB2C1sR=0TB2C2sR=nT21n3.1MCMbuyFT1buyF=0T2buyF=nTnsFTB2C1sF=0TB2C2sF=nTnB2CMT1M=0T2M0T3.4MCMB2C(,,)rfm(,,,,)buysbuysRRFFM3-3(1)buyR0buyR=0T(2)buyR=nbuyR=n+1sR=msR,=m+1(3)buyRbuyR=nbuyR=1sRsR=msR=1buyFbuyF=xbuyF=x+1sFsF=ysF=y+1buyRbuyR=nbuyR=n+1sRsR=msR=m+1buyFsFsF=ysF=y+1(4)M22(5)sbuyRR≥FFsbuy≥3-3B2CMCM(1)P0(2)Aij(3