1.HBR.ORG DECemBER 2011 reprint R1112ESpotlight on Reinventing RetailKnow What YourCustomers WantBefore They DoRetailers need to target customers with the rightdeal at the right time. Heres how to nail the nextbest offer. by Thomas H. Davenport, Leandro DalleMule, and John Lucker

2. Spotlight on Reinventing RetailSpotlight Artwork Rachel Perry WeltyLost in My Life (Playmobil)2010, pigment printKnow WhatYour CustomersWant BeforeThey Do hoppers once relied onRetailers need to target customers with a familiar salespersonthe right deal at the right time. Heressuch as the proprietorhow to nail the next best offer.of their neighborhoodgeneral storeto helpby Thomas H. Davenport, Leandro Dallethem find just what theyMule, and John Lucker wanted. Drawing on what he knew or could quicklydeduce about the cus- tomer, he would locate the perfect product and, often, suggest additional items the customer hadnt even thought of. Its a quaint scenario. Todays dis-Photography: Rachel Perry Welty and Yancey Richardson Gallery, NEW YORK tracted consumers, bombarded with information and options, often struggle to find the products or services that will best meet their needs. The shorthanded and often poorly informed floor staff at many retailing sites cant begin to replicate the personal touch that shoppers once depended onand consumers are still largely on their own when they shop online. This sorry state of affairs is changing. Advances in information technology, data gathering, and analytics are making it possible to deliver something likeor per- haps even better thanthe proprietors advice. Using increasingly granular data, from detailed demographics and psychographics to consumers clickstreams on the web, businesses are starting to create highly customized offers that steer consumers to the right merchandise or servicesat the right moment, at the right price, and in the right channel. These are called next best offers.2 Harvard Business ReviewDecember 2011 3. For article reprints call 800-988-0886 or 617-783-7500, or visit hbr.org 4. Spotlight on Reinventing RetailConsider Microsofts success with e-mail offers forits search engine Bing. Those emails are tailoredWhat Makes to adjust merchandise for local tastes and to custom- ize offerings at the individual level across a varietyto the recipient at the moment theyre opened. In An NBO?of store formats, from hypermarts to neighborhood200 millisecondsa lag imperceptible to the re- Next best offer is shops. For example, Clubcard shoppers who buyincreasingly used tocipientadvanced analytics software assembles an diapers for the first time at a Tesco store are mailedrefer to a proposaloffer based on real-time information about him or customized on thecoupons not only for baby wipes and toys but also forher: data including location, age, gender, and online basis of beer. (Data analysis revealed that new fathers tend toactivity both historical and immediately preceding,the consumersbuy more beer, because they are spending less time attributes andalong with the most recent responses of other cus-behaviorsat the pub.) More recently, Tesco has experimentedtomers. These ads have lifted conversion rates by as(demographics, with flash sales that as much as triple the redemp-much as 70%dramatically more than similar butshopping history)tion value of certain Clubcard couponsin essenceuncustomized marketing efforts.the purchase making its best offer even better for selected custom-contextThe technologies and strategies for crafting next (bricks and mortar,ers. A countdown mechanism shows how quicklybest offers are evolving, but businesses that waitonline)time or products are running out, building tensionto exploit them will see their customers defect top roduct or serviceand driving responses. Some of these offers havecharacteristicscompetitors that take the lead. Microsoft is justsold out in 90 minutes.(shoe style, type ofone example; other companies, too, are revealingmortgage)Tescos NBO strategy seeks to expand the range ofthe business potential of well-crafted NBOs. Butt he organizationscustomers purchases, but it also targets regular cus-in our research on NBO strategies in dozens of re-strategic goalstomers with deals on products they usually buy. As a(increase sales, buildtail, software, financial services, and other compa-customer loyalty)result of its carefully crafted, creatively executed of-nies, which included interviews with executives at fers, Tesco and its in-house consultant dunnhumby15 firms in the vanguard, we found that if NBOs are NBOs are most oftenachieve redemption rates ranging from 8% to 14%designed to inspiredone at all, theyre often done poorly. Most are indis- a purchase, drivefar higher than the 1% or 2% seen elsewhere in thecriminate or ill-targetedpitches to customers wholoyalty, or both.grocery industry. Microsoft had a very different sethave already bought the offering, for example. OneThey can consist ofof objectives for its Bing NBO: getting new customersretail bank discovered that its NBOs were more likely products to try the service, download it to their smartphones,(a coupon for diapers)to create ill will than to increase sales. install the Bing search bar in their browsers, andservicesCompanies can pursue myriad good goals us-(a discount on a spa make it their default search engine.ing customer analytics, but NBO programs providevisit) Starting with a clear objective is essential. So isperhaps the greatest value in terms of both potential information being flexible about modifying it as needed. TheROI and enhanced competitiveness. In this article (Google ads to click on) low-cost DVD rental company Redbox initially madewe provide a framework for crafting NBOs. You may relationships e-mail and internet coupon site offers intended tonot be able to undertake all the steps right away, but(LinkedIn and Facebook familiarize consumers with its kiosks. Redbox ki-recommendations)progress on each will be necessary at some point toosks were a new retail concept, but in time peopleimprove your offers.Despite the name, an became accustomed to automated movie rentals.NBO may in fact be anAs the business grew, the companys executives re-initial engagement. AndDefine Objectives whether the customer alized that to increase profits while maintaining theMany organizations flounder in their NBO efforts notrelationship is new or low-cost model, they needed to persuade customersbecause they lack analytics capability but becauseongoing, the NBO isto rent more than one DVD per visit. So they shiftedintended to be a bestthey lack clear objectives. So the first question is, offer.the emphasis of their NBO strategy from attractingWhat do you want to achieve? Increased revenues? new customers to discounting multiple rentals.Increased customer loyalty? A greater share of wal-let? New customers?Gather DataThe UK-based retailer Tesco has focused its NBOTo create an effective NBO, you must collect and in-strategy on increasing sales to regular customers andtegrate detailed data about your customers, your of-enhancing loyalty with targeted coupon offers deliv- ferings, and the circumstances in which purchasesered through its Clubcard program. As Roland Rustare made.and colleagues have described (Rethinking Market-Know your customers. Information valu-ing, HBR JanuaryFebruary 2010), Tesco uses Club- able for tailoring NBOs can be relatively basic andcard to track which stores customers visit, what theyeasily acquired or derived: age, gender, number ofbuy, and how they pay. This has enabled the retailer children, residential address, income or assets, and4 Harvard Business ReviewDecember 2011 5. For article reprints call 800-988-0886 or 617-783-7500, or visit hbr.org Idea in BriefTargeting individuals withPerfecting these nextIts hard to perfect all fourperfectly customized of- best offers (NBOs) involvessteps at once, but progressfers at the right moment four steps: defining objec- on each is essential to com-across the right channel istives; gathering data about petitiveness. As the amountmarketings holy grail. As your customers, your of-of data that can be capturedcompanies ability to captureferings, and the contexts ingrows and the number ofand analyze highly granularwhich customers buy; usingchannels for interaction pro-customer data improves,data analytics and business liferates, companies that aresuch offers are possibleyet rules to devise and execute not rapidly improving theirmost companies make them offers; and, finally, applyingoffers will only fall furtherpoorly, if at all. lessons learned.behind.psychographic lifestyle and behavior data. Previousabout 600 billion geospatially tagged data feedspurchases are often the single best guide to what aback to telecommunications providers every day.customer will buy next, but that information may be An application developed by the software analyt-harder to capture, particularly from offline channels. ics company Sense Networks can compare a con-Loyalty programs like Tescos can be a powerful tool sumers movements with billions of data points onfor tracking consumers buying patterns. the movements and attributes of others. Using thisEven as companies work (and sometimes strug- location history, it can estimate the consumers age,gle) to acquire these familiar kinds of customer data, travel style, level of wealth, and next likely location,the growing availability of social, mobile, and loca- among other things. The implications for creatingtion (SoMoLo) information creates major new data highly customized NBOs are clear.sets to be mined. Companies are beginning to craft Know your offerings. Unless a company hasoffers based on where a customer is at any given detailed information about its own products ormoment, what his social media posts say about hisservices, it will have trouble determining which of-interests, and even what his friends are buying or ferings might appeal most to a customer. For somediscussing online. products, such as movies, third-party databasesOne example is Foursquare, which makes cus- supply product attributes, and companies that renttomized offers according to how many times con- or sell movies can surmise that if you liked onesumers have checked in to a certain retail store. movie with a particular actor or plot type, you willAnother is Walmart, which acquired the social mediaprobably like another. But in other retail industries,technology start-up Kosmix to join its newly formedsuch as apparel and groceries, compiling product at-digital strategy unit, @WalmartLabs, in capitalizing tributes is much more difficult. Manufacturers donton consumer SoMoLo data for its offers. Amonguniformly classify a sweater as fashion forward orthe units projects is finding ways to predict shop- traditional, for example. They dont even have clearpers Walmart.com purchases on the basis of theirand standardized color categories. So retailers mustsocial media interests. Walmart is also looking into spend a lot of time and effort capturing product at-location-based technologies that will help custom- tributes on their own. Zappos has three departmentsers find products in its cavernous stores. The apparel working to optimize customers searches and createretailer HM has partnered with the online gamethe most effective offers for its shoes. Even whenMyTown to gather and use information on customer the attributes are narrowed down to product type,location. If potential customers are playing the gamestyle, color, brand, and price, a shoe might have anyon a mobile device near an HM store and check in, of more than 40 material patternspearlized, patch-HM rewards them with virtual clothing and points; work, pebbled, pinstripe, paisley, polka dot, or plaid,if they scan promoted products in the store, it enters to name just those beginning with p. Without athem in a sweepstakes. Early results show that ofsystem for such detailed classification of product at-700,000 customers who checked in online, 300,000 tributes, Zappos wouldnt know that a customer hadwent into the store and scanned an item. often bought paisley in the past, so it wouldnt knowMany retailers focus on how to use customers that it should include paisley products in NBOs tolocation information in real time; where the cus- that customer.tomers have been can also reveal a lot about them. Similarly, without good classification systems,In the United States alone, mobile devices sendgrocers cant easily determine what products will December 2011Harvard Business Review5 6. Spotlight on Reinventing Retaillure adventurous, health-conscious, or penny-customer contact dont work well because custom-pinching customers. When Tesco wants to identify ers have neither the time nor the inclination to en-products that appeal to adventurous palates, it will gage with them, whereas they might be receptive tostart with something that is widely agreed to be a the same offers during a walk-in. Likewise, someonedaring choice in a given countryThai green currywho calls customer service with a complaint is un-paste in the UK, perhapsand then analyze the otherlikely to respond to a product offer, though he or shepurchases that buyers of the daring choice make. Ifmight welcome it by e-mail at another time.customers who buy curry paste also frequently buyOther contextual factors that may affect thesquid or wild rocket (arugula) pesto, these products design of an NBOand a consumers response tohave a high relationship coefficient.itinclude the weather, the time of day or the day Know the purchase context. Finally, NBOsof the week, and whether a customer is alone or ac-must take into account factors such as the channel companied. Although clickstream or recent onlinethrough which a customer is making contact withpurchase data are often the most relevant in guidinga business (face-to-face, on the phone, by e-mail, an online NBO strategy, in some cases, such as air-on the web), the reason for contact and its circum-travel ticket pricing, time and day are important: Air-stances, and even voice volume and pitch, indicating lines can hike prices on a Sunday evening, becausewhether the customer is calm or upset. (Emotion- more people search then than, say, midday duringdetection software is proving valuable for the lastthe week. A Chinese shoe retailer we studied is test-factor.) Bank of America has learned that mortgage ing offers that target primary buyers companions.offers presented through an ATM at the moment of When a woman walks into one of its stores with herBuilding THE NEXT BEST OFFERExemplary companies build or sharpen an1 2 3 4NBO strategy through four broad activities: DefiningGathering Analyzing and Learning and objectivesdataexecuting evolving Craft NBOs to achieve Collect detailed data Use statistical analysis, Think of every offer as a specific goals, such as about customers (demo-predictive modeling,test. Incorporate data on attracting new custom-graphics and psycho-and other tools to matchcustomers responses in ers or increasing sales,graphics; purchase his- customers and offers. Use follow-on offers. Formu- loyalty, or share of wallet.tory; social, mobile, and business rules to guide late rules of thumb for Be ready to modify your location information),what offers are made un-designing new offers that objectives to exploit your offerings (product der what circumstances. are based on the perfor- changing circumstances. attributes, profitability,Carefully match offersmance of previous ones. availability), and purchase and channels. Make of- context (customers con-fers sparingly, time them tact channel, proximity,deliberately, and monitor the time of day or week). contact frequency.6 Harvard Business ReviewDecember 2011 7. For article reprints call 800-988-0886 or 617-783-7500, or visit hbr.org he weather, the time of day or day of the week, and whether or not a customer is accompanied may affect the design of an offer.husband, she is usually the primary buyer, and theoffers a robust overview of key analytical, quantita-retailers NBO is usually a relatively inexpensive item tive, and computer modeling techniques.)for the husband. The choice of what to offer himAlthough such analytics can yield a profusion ofarises from the insight that men who accompanypotentially effective offers, business rules govern thetheir wives shopping but are not actively shoppingnext step. When an analysis shows that a customerthemselves are more price sensitive than solo hus-is equally likely to purchase any of several products,bands who are searching for a specific product. a rule might determine which offer is made. Or it Of course, countless other contextual factors de-might limit the overall contact frequency for a cus-pend on the nature of the business and its customers. tomer if analyses have shown that too much contactreduces response rates. These rules tend to go be-Analyze and Execute yond the logic of predictive models to serve broad The earliest predictive NBOs were created by Ama-strategic goalssuch as putting increasing customer zon and other online companies that developedloyalty above maximizing purchases.people who bought this also bought that offersA carefully crafted NBO is only as good as its de- based on relatively simple cross-purchase correla- livery. Put another way, a brilliant e-mail NBO that tions; they didnt depend on substantial knowledge never gets opened might as well not exist. Should of the customer or product attributes, and thus were the NBO be delivered face-to-face? Presented at rather a blunt instrument. Somewhat more targetedan in-store kiosk? Sent to a mobile device? Printed offers are based on a customers own past purchase on a register receipt? Often the answer is relatively behavior, but even those are famously indiscrimi-straightforward: The channel through which the nate. If you buy a book or a CD for a friend who customer made contact is the appropriate channel doesnt share your tastes, that can easily skew thefor delivering the NBO. For example, a CVS customer future offers you receive. who scans her ExtraCare loyalty card at an in-store Companies that have systematically gathered in-kiosk can instantly receive customized coupons. formation about their customers, product attributes, There are times, however, when the inbound and and purchase contexts can make much more sophis- outbound channels should differ. A complex offer ticated and effective offers. Statistical analysis and shouldnt be delivered through a simple channel. predictive modeling can create a treasure trove of Recall Bank of Americas experience with mortgage synthetic data from these raw information sourcesoffers: The inbound channelthe ATMwas quickly to, for example, gauge a customers likelihood offound to be a poor outbound channel, because responding to a discounted cross-sell offer deliv- mortgages are just too complicated for that setting. ered on her mobile device. Behavioral segmentation Similarly, many call-center reps dont understand and other advanced data analytics that simultane-customer needs and product details well enough to ously account for customer demographics, attitudes,make effective offersparticularly when the reps buying patterns, and related factors can help iden-primary purpose is to complete simple sales or ser- tify those customers who are most likely to defect.vice transactions. Armed with this information and a customers ex- Companies often test offers through multiple pected customer lifetime value, an organization canchannels to find the most efficient one. At CVS, Ex- determine whether its NBO to that customer shouldtraCare offers are delivered not only through kiosks encourage or discourage defection. (A detailed dis-but also on register receipts, by e-mail and targeted cussion of marketing data analytics is beyond thecirculars, and, recently, via coupons sent directly to scope of this article, but the 2002 book Marketing En- customers mobile phones. Qdoba Mexican Grill, a gineering, by Gary L. Lilien and Arvind Rangaswamy,quick-serve franchise, is expanding its loyalty pro-December 2011Harvard Business Review7 8. Spotlight on Reinventing Retail For article reprints call 800-988-0886 or 617-783-7500, or visit hbr.org gram by delivering coupons to customers smart-Learn and Evolve phones at certain times of the day or week to increase Creating NBOs is an inexact but constantly improv- sales and smooth demand. Late-night campaignsing science. Like any science, it requires experimen- near universities have seen a nearly 40% redemp- tation. Some offers will work better than others; tion rate, whereas redemption rates average 16%companies must measure the performance of each for Qdobas overall program. Starbucks uses at least and apply the resulting lessons. As one CVS execu- 10 online channels to deliver targeted offers, gauge tive said to us, Think of every offer as a test. customer satisfaction and reaction, develop prod-Companies can develop rules of thumb from ucts, and enhance brand advocacy. For example, its their NBOs performance to guide the creation of fu- smartphone app allows customers to receive tailoredture offersuntil new data require a modification ofthe rules. These rules will differ from one companyto the next. In our research we identified some thatleading companies use:Footlocker: Promote only fashion-forwardshoes through social media. pscale retailers and CVS: Provide discounts on things a customer hasbought previously.financial services firms find thatSams Club: Provide individually relevant offersfor categories in which a customer has not yet pur-a human being is often the best chased, and reward customer loyalty.Nordstrom: Provide offers through sales associ-channel for delivering offers.ates in face-to-face customer interactions.Rules of thumb should be derived from data-driven and fact-based analyses, not convention or promotions for food, drinks, and merchandise based lore. The rules above have been tested, but they will on their SoLoMo information. need to be challenged and retested over time to en- Nordstrom and other upscale retailers, and fi- sure continued effectiveness. nancial services firms with wealthy clients, investMeanwhile, legal, ethical, and regulatory issues heavily in their salespeoples product knowledge associated with NBO strategies are evolving fast, as and ability to understand customers needs and the collection and use of customer data become in- build relationships. For these businesses, a human creasingly sophisticated. When companies enthusi- being is often the best channel for delivering offers. astically experiment with NBOs, they should be wary Many organizations devise multiple offers and sort of unwittingly crossing legal or ethical boundaries. them according to predictive models that rank aIt would be hard for any company to incorporate customers propensity to accept them on the basisevery possible customer, product, and context vari- of previous purchases or other data. Salespeople orable into an NBO model, but no retailer should fail customer service reps can select from among theseto gather basic demographics, psychographics, and offers in real time, guided by their dialogue with the customer purchase histories. Most retailers need to customer, the customers perceived appetite for aaccelerate their work in this area: Their customers given offer, and even the comfort level between theare not impressed by the quality or the value of of- customer and the salesperson. Combining humanfers thus far. Variables and available delivery chan- judgment with predictive models can be more ef-nels will only grow in number; companies that arent fective than simply following a models recommen-rapidly improving their offers will just fall further dations. For example, insisting that a rep deliver a behind. HBR Reprint R1112E specific offer in every case may actually reduce both customers likelihood of accepting the offer and Thomas H. Davenport is the Presidents DistinguishedProfessor of Information Technology and Management their postpurchase satisfaction. The investment firm at Babson College, a senior adviser to Deloitte Analytics, and T. Rowe Price provides call-center representatives the research director of the International Institute for Analyt- with targeted offers, but it has concluded that if a rep ics. Leandro Dalle Mule is the global analytics director atCitibank. John Lucker is a principal at Deloitte Consulting delivers the offers in more than 50% of interactions,LLP, where he is a leader of Deloitte Analytics in the U.S. and he or she probably isnt tuning in to customers needs.of advanced analytics and modeling globally.8 Harvard Business ReviewDecember 2011