DATA ON THE 4PS FOR A MULTI-CHANNEL/MULTI-BRAND RETAILER WCAI Research Opprtunity 27 September 2012 1 http://wcai.whartn.upenn.edu @WhartnCAI
WHY SHOULD YOU SUBMIT A PROPOSAL? Access t unique data Simple, efficient prpsal prcess Prpsals due 2-3 weeks after the webinar Data awarded abut 4-6 weeks after the webinar Assistance with data cleaning and preparatin Telecnference Q&A sessins with the crprate spnsr Invitatin t a private sympsium (typically at Whartn) t share yur research findings with the crprate partner abut a year frm nw Prmtin f yur research paper thrugh the WCAI SSRN Research Paper Series 2
THE COMPANY 3
ABOUT THE COMPANY Majr specialty retailer Three channels Three distinctive retail brands Independently managed with little crssprmtin One brand has a lyalty prgram Unique prduct lines Each brand has its wn stres; stres are nt c-lcated Stre Online Catalg Stres are primarily in Nrth America The cmpany wishes t be annymus during the prpsal prcess. Researchers awarded the data will be tld which specific brands are included in the data, but will nt be permitted t reveal thse brands in any publicatins. 4
THE CRM SYSTEM Recently, the cmpany has implemented a state-f-the-art system t track and merge key custmer tuch pints ( events ) with the firm Direct Mail e.g., pstcards E-Mails Catalg Mailings Web Visits In-Stre Purchases Direct Telephne/Online Purchases Prduct Reviews 5
THE CRM SYSTEM: TRACKING EVENTS As custmer events ccur, they are matched t existing custmers in the CRM system Fr sme events, the custmer id is knwn A catalg was mailed t a specific custmer id When the custmer id is unknwn, the cmpany matches t an existing custmer based n infrmatin gathered during the event Fr an nline purchase, the ship-t name and address match Fr an in-stre purchase, the lyalty card number matches Fr an in-stre purchase, an e-mail prvided t a stre clerk matches an existing custmer Sme events can nt be matched, but are maintained in the database under annymus custmer ids Cash purchase at a physical stre E-mail sign up withut name 6
ABOUT THE DATA Fr each f three brands, the data includes 14,000 custmers randmly selected frm amng Nrth American custmers wh were active with that brand Active means the user had at least ne marketing cntact, website visit, r purchase between 1 July 2010 and 30 June 2012 with that brand Includes bth users wh s name and address are knwn (89%) as well as users wh are tracked by e-mail address r web-user id nly (11%) Fr thse 42,000 custmers, the data includes all custmer event with any f the three brands between1 July 2010 and 30 June 2012 7
THE 6 P S FOR THIS DATA Peple Purchases Place Physical Stres Online Stres Prduct Price Prmtins 8
PEOPLE: DEMOGRAPHICS Amng 42,000 Active Custmers Average Age 39.7 (39.1% unknwn) Gender 75.6% Female (13.3% unknwn) Has Purchased Brand A 61.8% Has Purchased Brand B 71.0% Has Purchased Brand C 33.5% Lyalty Prgram Participatin (Amng thse active with brand A) 57.7% Custmer data includes: Age (demgraphic append) Gender (demgraphic append) Gegraphic lcatin (census blck) First purchase date (fr each brand) Distance t nearest stre (fr each brand) Membership in the lyalty prgram (Brand A nly) Whether the custmer has ever created a wish list Custmer preferences fr catalg and e-mail mailings (fr each brand) 9
PEOPLE: TENURE WITH THE BRAND The data includes first purchase date fr each brand Even if that date is befre the bservatin windw Tenure with the brand varies widely, with Brand C having a relatively newer custmer base 10
PURCHASES The data includes all purchases made by this set f custmers ver the tw-year windw Purchases are seasnal with a substantial amunt f direct business Direct In-Stre 11
PURCHASES: DETAILS Fr each transactin, the data includes: Brand Date Channel (direct r in-stre) Methd f payment Prducts purchased (identified by SKU) Amunt paid Whether it was returned Stre lcatin 12
PLACE (PHYSICAL) Brand Percentage f Stres Brand A 41% Brand B 42% Brand C 18% Each purchase is tracked t a stre. Data n 300+ physical stres includes: Brand Lcatin (zip, lat & lng) Cmpany-defined regin/district Date stre pened/clsed Stre surrundings (mall, city street, etc.) Climate categry (cld, mild, warm, ht) Stre size 13
PLACE (PHYSICAL): STORE SIZE There is a great deal f variatin in the size f stres Brand B has very wide variatin in the size and layut f stres 14
PLACE (PHYSICAL): DISTANCE TO CUSTOMER Data includes the distance between each custmer wh has made a purchase frm a particular brand and her nearest stre 15
PLACE (DIRECT) Direct prvides an alternative channel accessible t all either via catalg r nline Unlike physical stres, visits t the nline stre are tracked in detail 16
PLACE(DIRECT): CLICKSTREAM Each nline visit recrd includes: Time and date f visit Hw many pages were viewed Hw many prducts were viewed Full text fr any n-site searches Hw many items were placed in the cart Hw many prducts were purchased Number f visits with at least: One page view 2,362,473 Ten page views 712,430 Fifty page views 119,300 One n-site prduct search 138,816 One prduct view 1,138,339 Five prduct views 449,099 One item purchased 25,074 17
Prduct PRODUCT Prduct selectin changes frequently 18
PRODUCT: DETAILS Because prduct ffering is an imprtant decisin fr this retailer, the spnsr has prvided rich data n the prducts including: Brand Psitin in prduct hierarchy Divisin(25) / department(60) / class(271) / grup(347) / item / sku Sets f prducts that are related (e.g., different clrs f the same basic prduct r crdinating prducts) are identified Prduct vendr 19
PRODUCT: ONLINE MERCHANDISING Fr prducts that are sld nline, we als have infrmatin abut hw the prduct was merchandised nline Categry the prduct was displayed in n the website Full text f the nline prduct descriptin Which clr was the first ne displayed nline We als have infrmatin abut hw many times each prduct was viewed nline 20
PRODUCT: ONLINE REVIEWS Fr thse prducts that are sld nline, the data includes the average user rating Additinally, full text f reviews is available fr thse custmers included in the sample WHARTON CUSTOMER ANALYTICS INITIATIVE PAGE 21
PRICE The cmpany des nt practice regular price prmtins 22
PROMOTIONS: CATALOGS, E -MAIL AND DIRECT MAIL The retailer prmtes its stres and prducts primarily thrugh direct marketing with little mass media, s much f the marketing activity can be tracked at the individual levels Nte that the cmpany des little price prmtin, s this cmmunicatin is pure advertising Fr each custmer we bserve: When each direct cntact ccurred Channel (mail, e-mail) Campaign descriptin 23
Q&A 24
RESEARCH QUESTIONS Multi-channel/multi-brand advertising attributin: Hw d nline and ffline marketing cmmunicatins affect bth nline and ffline sales? What is the rle f utbund marketing fr a nn-prmtinal brand? What is the relatinship between prduct attributes and sales? Can prduct infrmatin be used t predict what individual custmers will buy next r t make prduct recmmendatins? Hw can this CRM type data be used t quantify retail saturatin & crss-channel cannibalizatin? 25
RESEARCH QUESTIONS What effect d lyalty prgrams have n subsequent purchase behavir? Hw shuld this CRM data be used t predict custmer lifetime value (CLV)? Hw shuld nline and catalg buying behavir be used t identify new ptential physical stre lcatins? Are there custmers wh are trendsetters, wh can be used t identify what t sell next? 26
PROPOSAL FORMAT The prpsals shuld be n mre than 2,000 wrds and shuld include the fllwing sectins: Title Authr(s) name, title, affiliatin and e-mail address. Please designate a crrespnding authr. Abstract Intrductin. Describe expected cntributin(s), cvering bth the academic and practical aspects. Please keep it cncise, and cite relevant wrk as necessary t explain yur academic cntributin. There is n need t include a lengthy literature review. Detailed Prject Prpsal. Describe the prpsed research. Please include enugh detail that we can assess the feasibility and merit f the prpsed apprach. Fr example, mdeling prjects shuld include at least a sketch f the mdel. Experimental prjects shuld include an utline f the research design. Please als include a rugh estimate f hw lng the prject will take. Bigraphies. Include up t a paragraph-lng bigraphy highlighting what each team member will cntribute t the prject. 27
PROPOSAL SUBMISSION PROCESS Read mre abut the prpsal frmat at http://www.whartn.upenn.edu/wcai/prpsals.cfm Submit prpsals at http://www.whartn.upenn.edu/wcai/prpsalfrm.cfm n later than Thursday, Octber 18, 2012. Prpsals will nly be accepted nline. Prpsals will be evaluated bth n academic cntributin and ptential t significantly imprve the research spnsr s marketing practice. Eric Bradlw (WCAI), Pete Fader (WCAI), Elea Feit (WCAI), and representatives f the spnsr. Cntact wcai-research@whartn.upenn.edu, if yu have questins prir t submitting yur prpsal. 28
OTHER WCAI OPPORTUNITIES FOR RESEARCHERS If yu registered fr this webinar, yu will receive regular annuncements abut: Research Opprtunities like this ne SmartMeter Data: Octber 2012 Lyalty data fr a travel services prvider: Nvember 2012 Online Ratings: January 2013 General Mtrs: February 2013 Grant/funding pprtunities Als find us at: SSRN Research Paper series: http://www.ssrn.cm/link/whartn- Cust-Analytics-Initiative-RES.html Annuncements: http://wcai.whartn.upenn.edu 29