Localization Have Pursued Strategies of the Revolution in Consumer Markets Standardization

www.hbr.org

For a quarter century, the big winners in consumer markets Localization have pursued strategies of The Revolution in Consumer Markets standardization. But success for retailers and product manufacturers now hinges on by Darrell K. Rigby and Vijay Vishwanath their ability to cater to local differences—while maintaining scale efficiencies.

Reprint R0604E

For a quarter century, the big winners in consumer markets have pursued strategies of standardization. But success for retailers and product manufacturers now hinges on their ability to cater to local differences—while maintaining scale efficiencies.

Localization The Revolution in Consumer Markets

by Darrell K. Rigby and Vijay Vishwanath

We’re in the early stages of a quiet revolution sponse, smart retailers and consumer goods in consumer markets. For decades, the chains companies are starting to customize their of- that have dominated the landscape—titans ferings to local markets, rolling out different like Wal-Mart, Best Buy, and McDonald’s— types of stores, product lines, and alternative have pursued single-minded strategies of stan- approaches to pricing, marketing, staffing, and dardization. They’ve fine-tuned their store customer service. They’re moving from stan- formats, merchandise mixes, and operating dardization to localization. and marketing processes, and they’ve rolled Combining sophisticated data analysis with out their winning formulas internationally. innovative organizational structures, they’re They’ve demanded equally rigorous consis- gaining the efficiencies of centralized manage- tency from suppliers, pushing the standardiza- ment without losing the responsiveness of tion ethic deep into consumer product compa- local authority. The greatest benefit of moving nies and across the entire consumer supply from standardization to localization is strate- chain. gic. Standardized offerings discourage experi- But the era of standardization is ending. mentation and are easy for competitors to Consumer communities are growing more di- copy. (Sam Walton openly referred to Kmart as verse—in ethnicity, wealth, lifestyle, and val- the “laboratory” he copied while growing Wal- ues. Many areas, moreover, are now satu- Mart.) Customization encourages local experi- rated with big-box outlets, and customers mentation and is difficult for competitors to are rebelling against cookie-cutter chain stores track, let alone replicate. When well executed, that threaten the unique characteristics, such localization strategies can provide a durable as architectural styles and favored brands, competitive edge for retailers and product of their neighborhoods. When it comes to con- manufacturers alike. sumer markets, one size no longer fits all. In re- OPYRIGHT © 2006 HARVARD BUSINESS SCHOOL PUBLISHING CORPORATION. ALL RIGHTS RESERVED. BUSINESS SCHOOLOPYRIGHT © 2006 HARVARD PUBLISHING CORPORATION. C

harvard business review • april 2006 page 1

Localization

Reinventing the Big Box stores, products, and services with unprece- Although standardization has been a power- dented precision. (For an example of the new ful strategy in consumer markets, it’s reached insights technology can deliver, see the sidebar the point of diminishing returns. Customers “Mining the Internet.”) are becoming more diverse, according to Our analysis of 30 localization leaders, in- studies by geodemographers, people who cluding Best Buy, Tesco, and VF, documents study the population characteristics of spe- these benefits. Even Wal-Mart, the sultan of cific geographic areas. Measuring ethnicity, standardization, is moving toward localization. age, wealth, urbanization, housing styles, and The company has made customization the even family structures, the demographic cornerstone of its “store of the community” company Claritas determined in the 1970s strategy, announcing that it plans to tailor for- that 40 lifestyle segments were sufficient to mats and products to the local clientele in define the U.S. populace. Today, that number every store in its chain. has grown to 66, a 65% increase. Wal-Mart uses a rigorous process to ensure Diversity is not the only nail in standardiza- that customization does not undermine its tra- tion’s coffin. Many large chains have erected so ditional efficiency. That process begins when a many stores that they’re literally running out store is still on the drawing board. Company of room to expand. They can’t open new out- real-estate teams deeply research the local lets without cannibalizing old ones. Standard- customer base when scouting for locations. ized chains are also meeting with other con- Designers then create the store’s format by straints: Where attractive locations are still combining suitable templates—stores near of- available, attempts to build stores often face fice parks, for example, with prominent islands fierce resistance from community activists. featuring ready-made meals for busy workers. From California to Florida to New Jersey, Templates allow Wal-Mart to maintain consid- neighborhoods are passing ordinances that erable economies of scale. The company has dictate the sizes and even architectural styles also developed a sophisticated logistics system, of new shops. Building more of the same— encompassing 110 distribution centers in the long the cornerstone of retailer growth—has United States alone, to manage complex de- been tapped out as a strategy. livery schedules quickly and efficiently. Finally, standardization can do the most stra- Through its Retail Link program, Wal-Mart tegic damage by forcing products and practices works with suppliers to tailor store merchan- into molds. The resulting homogenization of dise with similar precision. Built on a vast data- business tends to undermine innovation, all base, Retail Link provides both local Wal-Mart the way up the supply chain. Managers become managers and vendors with a two-year history so focused on meeting tight operational tar- of every item’s daily sales in every Wal-Mart gets—and stamping out exceptions—that they store. Using the Retail Link Web portal, Wal- begin to consciously avoid the experimentation Mart and its suppliers can create maps of local that leads to attractive new products, services, customer demand, indicating which merchan- and processes. In the end, standardization dise should be stocked when and where. For erodes strategic differentiation and leads inexo- example, Wal-Mart stocks about 60 types of rably toward commoditization—and the lower canned chili but carries only three nationwide. growth and profitability that accompany it. The rest are allocated according to local tastes. The good news is that there’s a way out of Five years ago, Wal-Mart used just five plano- standardization’s dead end. Technological ad- grams (diagrams showing how and where Darrell K. Rigby is a Boston-based vances, from checkout scanners and data- products should be placed on retail shelves) to partner of Bain & Company and leads mining software to Internet stores and radio adapt its soup selection to local preferences. the firm’s global retail practice. A fre- frequency identification (a wireless technol- Today, with the help of Retail Link, Wal-Mart quent contributor to HBR, he coau- ogy that uses small electronic tags to identify and its suppliers use more than 200 finely thored “CRM Done Right” (November and track objects), are providing retailers and tuned planograms to match soup assortments 2004). Vijay Vishwanath, also a their suppliers with deep insight into local to each store’s demand patterns—raising partner in Boston, leads Bain’s glo- preferences and buying behaviors. For the first soup’s growth rate by several points in the pro- bal consumer products practice. He time, mismatches in supply and demand at in- cess. Product companies also use the system to coauthored “Expanding in China” dividual stores can be pinpointed immediately. track their sales and inventory levels in Wal- (March 2005). The new data make it possible to “localize” Mart’s stores and distribution centers and to harvard business review • april 2006 page 2

Localization

develop pricing and marketing programs to seen sales increase dramatically, according to boost sales. John Westling, senior vice president. Of course, customization has its limits. Even Thinking in Clusters with rich data, a company can’t customize As Wal-Mart and other leaders have discov- every element of its business in every location. ered, successful localization hinges on getting The sheer complexity would be overwhelming, the balance right. Too much localization can leading to spiraling costs, if not paralysis. corrupt the brand and lead to ballooning That’s why leading localizers have begun using costs. Too much standardization can bring clustering techniques to simplify and smooth stagnation, dooming a company to dwindling decision making, focusing their efforts on the market share and shrinking profit. relatively small number of variables that usu- Striking the right balance means under- ally drive the bulk of consumer purchases. standing which elements of a business should Rather than letting local managers’ decen- be considered for localization, how costly tralized decisions fragment economies of scale, they are to customize, and how much impact the pioneering companies have developed a they will have from one store to another. science of analyzing data on local buying pat- Far from being an all-or-nothing game, localiza- terns to identify communities that exhibit simi- tion can take place in myriad ways (see the larities in demand. For example, American exhibit “What, Where, and When Should We Eagle Outfitters, a retailer of fashionable casual Localize?”). For one retailer, it might make wear with 740 U.S. stores, found that custom- sense to have a highly localized staffing ap- ers in western Florida exhibited seasonal proach but a standardized product mix, while purchasing patterns and price elasticities that another retailer may warrant the opposite. closely matched those of certain communities Similarly, a manufacturer might localize prod- in Texas and California. By tailoring assort- uct features in one area and retailer incentives ments and promotions to such clusters of lo- in another. While it may be prohibitively ex- cations rather than to individual stores, com- pensive to customize a product to many loca- panies like American Eagle can benefit from tions, it may be possible to gain similar bene- customization while holding on to most of the fits by tailoring the product’s packaging or efficiencies of standardization. promotions at a far lower cost. Wal-Mart The customization-by-clusters strategy, which found that while ant and roach killer sells well Bain first applied to grocery stores in 1995, has in the southern United States, consumers in proven effective in drugstores, department the northern states are turned off by the word stores, mass merchants, big-box retailers, res- “roach.” After labeling the pesticide as “ant taurants, apparel companies, and a variety of killer” in northern states, the company has consumer goods manufacturers. Clustering sorts things into groups, or clusters, so that the associations are strong between members of the same cluster and weak between members of different clusters. Clusters enable manage- Mining the Internet able, modular operations—think again of Many retailers have opened online centralized merchandise pools to avoid Wal-Mart’s store templates—that capture most stores to complement their traditional local stock-outs, and excess demand can of the benefits of customization while also outlets. But the Web is not just a sales often be back-ordered for future deliv- simplifying decisions and protecting econo- channel; it’s also a powerful means of ery. By carefully tracking the home ad- mies of scale. Consider a merchandise man- collecting data on variations in local de- dresses of online buyers as well as the ager who has to decide how to stock 100,000 mand. Because online stores can offer products they’re buying (or avoiding), items in 1,500 stores for 365 days each year. extensive ranges of products to national, chains that maintain Internet stores can If she wanted to customize the mix, she or even global, customer bases, they can use online sales data to inform decisions would have to make about 54.8 billion deci- track consumer demand patterns much about what merchandise to stock in sions (100,000 x 1,500 x 365), many of which more broadly and precisely than physi- which store. And because the online would be based on such small sample sizes cal stores can. In a traditional store, after data can be collected in real time, shifts that the predictions of even sophisticated all, you never know what the demand in physical stores’ merchandise mixes models would be meaningless. If, however, the might have been for a product you don’t can be made quickly to respond to merchandise could be clustered into 2,500 clas- have on the shelves. Online stores use spikes in local demand. sifications, the stores could be clustered into 20

harvard business review • april 2006 page 3

Localization

What,Where, and When Should We Localize?

Many different elements of a company’s business can be customized, separately or in combination. In consumer markets, a useful way to think about the elements is to arrange them into three categories: what’s being sold (“offer”), where it’s being sold (“location”), and when it’s being sold (“time”). The table provides a generic overview of this organization.

WHAT: Offer Variables q Branding Good/better/best range q Vendor policies q Vendor services Store (banner names) Pack counts Information sharing Direct store delivery Product labels Packaging design Expense sharing Replenishment and stocking Vendor brands Product collaboration Customer education Proprietary (private brands) q Pricing Everyday low vs. q Marketing programs q Operating policies high-low policies q Store formats Spending levels Inventory levels Ranges Size and layout Media mix Sourcing strategies Points Store design type Major messages Shrink controls Matching policies Information sharing q Merchandise space q Store service levels q Promotions and assortment Store hours Types Division Labor quality and schedules Category Temporary price reduction levels Delivery policies Department In-store displays Checkout stations Classiﬁcation Markdown policies Special services Attributes (e.g., delivery, repair) Frequency Style and ﬂavor Depth Color Size

WHERE: Location Variables

q Consumer characteristics q Special Demand Drivers q Competitor Characteristics q Our Own Store Demand patterns School seasons Store saturation levels Characteristics Store purchase Hunting and Market share Our market share Area purchase ﬁshing seasons Store locations Our store locations Geodemographics Activities and sights Store formats Location characteristics and attitudes Ski resorts Pricing levels Site quality ratings Population density Beach towns Promotion policies Our store formats Age Athletic teams Marketing programs Sizes Income Tourist attractions Design types (models) Marital st atus Military bases Condition Ethnicity Special events Square footage allocation Religion Cinco de Mayo Special ﬁxtures and displays Lifestyle segment Pioneer Day Merchandise placement Psychographic Religious holidays zones Climate zone Stores of our sister Temperature divisions Precipitation Locations Potential weather events Merchandise mix

WHEN: Time Variables

q Hour q Week q Season q Day q Month q Year

harvard business review • april 2006 page 4

Localization

CHAID: Clustering by the Numbers One of many clustering techniques is combinations of characteristics that income areas can help SuperStuff to called CHAID, short for chi-squared au- best explain any variable we choose to profitably compete against KillerMart’s tomatic interaction detection. A statis- explore. In the example “Assessing offering. tical classification method proposed by Store Profitability,” we used CHAID Jumping to the right-hand side of the G.V. Kass in 1980, CHAID sorts items to understand what drives EBIT mar- CHAID tree, we learn about stores that into groups that are statistically differ- gins (earnings before interest and don’t face KillerMarts. In those areas, ent with respect to criterion or out- taxes) among SuperStuff’s 508 depart- the 76 large-format stores have an aver- come. For example, if we want to know ment stores. age EBIT margin of 9.1%, almost twice what groupings are associated with CHAID begins, at the top, by show- as much as the 122 small or midsize store profitability, CHAID might show ing us that the average EBIT margin is stores, which have a margin of only us that money-losing stores are in 4.2% for SuperStuff’s entire popula- 4.7%. Furthermore, the 60 small or mid- high-income neighborhoods with tion of stores. size stores that priced an average mar- multiple competitors, while the most CHAID then identifies the first differ- ket basket of groceries less than 3% profitable stores are in rural areas entiator of EBIT margins as the pres- above SuperStuff’s overall average had and have the capacity to carry the full ence of at least one KillerMart in each an EBIT margin of only 1.2%. However, product assortment. SuperStuff store’s trade area. The 198 the 62 small or midsize stores with A significant benefit of CHAID is that SuperStuff stores with no nearby Killer- prices more than 3% above SuperStuff’s it enables us to analyze the effects of Marts have an EBIT margin of 6.4%. average have a margin of 8.1%—almost characteristics in combination rather The 310 SuperStuff stores near Killer- seven times more than the 60 stores than independent from one another. Marts have an average EBIT margin of pricing less than 3% above the average. For example, adding playgrounds to only 2.8%. Sensible, but not terribly sur- It seems that small or midsize stores Burger King restaurants may have no prising so far. The next steps are where may do better by raising prices in less impact on average but could be very CHAID proves most valuable. competitive markets. profitable in suburban restaurants near For the 310 stores near a KillerMart, While CHAID certainly doesn’t high populations of young children and CHAID finds that household income provide all the answers, it can help to very unprofitable in downtown loca- levels drive significant profit differ- surface testable hypotheses such as tions with expensive real estate and few ences. The 188 stores in neighborhoods the following: children. with household incomes of more than • When opening new stores, avoid Let’s demonstrate the process with a $50,000 have average EBIT margins of locations near KillerMarts. department store chain we’ll call Su- 3.9%. The remaining 122 stores have • If there is a KillerMart in the area perStuff: margins of only 1.1%. (or one coming soon), position CHAID begins with a list of every The data also enables CHAID to stores in the highest-income store in the SuperStuff system and as generate remodeling ideas. Of the 188 neighborhoods. much information as possible about stores in higher-income neighbor- • When remodeling stores, each—including sales data by location, hoods near KillerMarts, the 113 that especially those near KillerMarts, time, and item. There is no need to have allocated more than 50% of their consider allocating more than worry about entering too much infor- square footage to apparel have EBIT 50% of the floor space to apparel. mation, since CHAID will highlight margins of 5.3%. The 75 stores with less • Smaller stores in areas without only the variables that create statisti- than 50% allocated to apparel have KillerMarts should test price cally significant differences. EBIT margins of only 1.8%. Apparently, increases. We can then use CHAID to find the plentiful apparel assortments in high-

harvard business review • april 2006 page 5

Localization

ASSESSING STORE PROFITABILITY

508 stores with 4.2% margin*

Presence of a Nearby KillerMart YES NO

310 stores 198 stores with 2.8% with 6.4% margin margin

Average Household Income Store Size >$50K <$50K LARGE SMALL/MEDIUM

188 stores 122 stores 76 stores 122 stores with 3.9% with 1.1% with 9.1% with 4.7% margin margin margin margin

Apparel Square FootConclusion: Prices Compared to Chain Average In areas without KillerMarts near- % % % <50% of total >50 of total by, build large <3 Above >3 Above square footage square footage stores if possible. Average Average

75 stores 113 stores 60 stores 62 stores with 1.8% with 5.3% with 1.2% with 8.1% margin margin margin margin

Conclusion: Conclusion: In areas with KillerMarts In areas without nearby, locate stores in KillerMarts nearby, high-income neighborhoods smaller stores and give apparel more than should keep prices half of the store's square at least 3% above footage. SuperStuff's average. *Margin based on EBIT.

harvard business review • april 2006 page 6

Localization

similar types (for example, Latino border loca- there’s a set of stores with specially trained tions or upscale suburban places), and the tim- staffs, extensive displays of office equipment, ing (back to school, winter holidays) could be and mobile “Geek Squads” of service techni- broken into 52 weeks, the number of decisions cians. would be reduced to 2.6 million, which a mod- While the chain plans to phase out these in- ern computer model can optimize fairly easily. dividual names beneath its banner, the termi- (For a discussion of a particularly powerful nology helped Best Buy crystallize the vision of statistical technique used in sorting through each target customer for each cluster of stores. many variables, see the sidebar “CHAID: Clus- By customizing stores in clusters, rather tering by the Numbers.”) than individually, Best Buy has been able to Best Buy is using clustering to move away maintain many of the scale economies that from a standardized big-box strategy. It has re- have long underpinned its success. So far, the vamped close to 300 of its 700 U.S. stores, in- new strategy is delivering strong results. The 85 troducing “customer-centric” formats to ap- Best Buy stores that had been localized as of peal to local shoppers. The company identified early 2005 posted sales gains two times the five representative types of customers. First, company’s average. Encouraged, the company there’s “Jill,” a busy mother who is the chief is accelerating the conversion, with plans to buyer for her household and wants quick, change over all its U.S. stores in three years personalized help navigating the world of and localize outlets in other countries as well. technology. In Eden Prairie, Minnesota, the So how do you get started with clustering? company designed a store that caters to the Begin by collecting as many data as possible on needs of this busy suburban moms segment. key elements of your business for each store. The company found that this group of previ- (Use the exhibit “What, Where, and When The era of ously untapped consumers offered the best Should We Localize?”) If some information is opportunity for expansion in the region. To at- missing or hard to get, don’t wait for it to be standardization is tract this group, the store has an uncluttered collected. Use what’s readily available to ending. Consumer layout with wider aisles and warmer lighting, launch the analysis, recognizing that clustering and technology-related toys for children. Per- always gets better over time. Use the data to communities are growing sonal shopping assistants educate technology develop clusters and identify customization neophytes about products, and there’s more opportunities. Then estimate the economics more diverse—in floor space allocated to household appliances. (including both sales and costs) of localizing ethnicity, wealth, Although the store still serves other, more tra- the most promising elements of the customer ditional electronics shoppers, the company offering—using as few clusters as possible. A lifestyle, and values. hopes the store can boost its sales by attracting clothing retailer, for example, might find that a set of local customers that have felt over- localized markdown policies offer attractive re- whelmed inside a Best Buy store. turns and that climate is the key variable influ- Other stores are being designed around the encing markdown decisions. Further analysis remaining four types of customers and are may determine that a small number of store based on local demand patterns. For example, clusters—three, say—will be sufficient to gain there’s “Buzz,” a technology junkie who wants the optimum economic benefit. For merchan- the latest gear for entertainment and gaming. dise mix, by contrast, the key variable might be Stores catering to Buzz have lots of interactive customer lifestyle, which may require a dozen displays that allow shoppers to try out new clusters to get the maximum payoff. equipment and media. Then there is “Barry,” an affluent, time-pressed professional who is Diversity in the Product Line looking for high-end equipment and personal- As big retailers shift away from standardiza- ized service. Stores tailored to his needs fea- tion, the ripple effects will reshape the entire ture a store-within-a-store for pricey home-the- consumer supply chain. Consumer goods com- ater setups. Stores made with “Ray” in mind panies will need to introduce more variations emphasize moderately priced merchandise into their lines, collaborating closely with re- with attractive financing plans and loyalty pro- tailers to put the right products in the right grams for the family man on a budget who places at the right times with the right pricing wants technology that can enhance his home and promotion programs. Manufacturers in life. Finally, for small-business customers, general have been slow to make this change.

harvard business review • april 2006 page 7

Localization

Although they conduct extensive consumer re- filled chocolate Cadbury Kiwi Royale in New search to develop specialized products for Zealand. Kraft developed Post’s Fiesta Fruity unique segments, they have little confidence Pebbles ready-to-eat cereal especially for His- that rigid retailers will sort, merchandise, and panics. Coca-Cola has developed four canned, market custom products to the right customer ready-to-drink coffees for Japan, each formu- clusters. Products developed for senior citi- lated for a specific region. Procter & Gamble zens will pile up in college communities— introduced Curry Pringles in England and, slowing inventory turns, forcing costly mark- later, Spanish Salsa flavor in England and other downs, and often leading retailers to drop po- parts of Europe and Funky Soy Sauce Pringles tentially profitable niche products. in Asia. Frito-Lay developed Nori Seaweed Nevertheless, as growing numbers of re- Lay’s potato chips for Thailand and A la Turca tailers are rolling out their own versions of corn chips with poppy seeds and a dried to- Wal-Mart’s Retail Link—including Lowe’s mato flavor for Turkey. (LowesLink) and Target (Partners Online)—a One of the leading localizers is consumer handful of consumer product companies are products giant VF, a $6 billion apparel maker seizing the advantage by learning to localize. that owns such popular jeans brands as Lee When one food company introduced low- and Wrangler as well as upscale labels includ- calorie versions of some of its snack foods, it ing Nautica and North Face. VF integrates shipped additional cases to stores near many data sources to identify customization Weight Watchers clinics. Cadbury added kiwi- opportunities—to the delight of retailers and

Extreme Localization While localizers typically customize 5%–25% buffs a place to hang out; and Studio D, a provide sandwiches at lunchtime, then create of a standardized format, extreme localizers cozy, neighborhood technology store for the prepared meals for customers to pick up on are developing a range of new—but closely suburban mom who stocks up for the family their way home for dinner. related—shopping formats to give targeted at Best Buy’s large formats but fills her per- >>TREND: Multiformat customers are customers more convenient purchasing op- sonal technology needs closer to home. generating higher profits and deeper be- tions. This is not conventional segment- >>TREND: Technological advances allow havioral insights. based expansion, where retailers build port- for more meaningful sharing of customer Bain’s research shows that multiformat cus- folios of brands to serve different sets of cus- knowledge and supply costs when chain tomers—those, for example, that buy from a tomers (think Talbots for women, Talbots for stores are selling the same items through chain’s superstore, catalog, Web site, and men, and Talbots for kids). Rather, this is so- multiple formats. neighborhood store—typically spend two to phisticated localization based on insights By capitalizing on common information sys- six times as much with a retailer as single- into three emerging trends in consumer mar- tems, supply chain logistics, and purchasing format customers do. Each positive experi- kets: processes, Tesco has embarked on extreme lo- ence builds scale and loyalty, making custom- >>TREND: Consumer purchasing pat- calization in the grocery sector—and is in- ers more profitable to the retailer and less terns vary not just by segment but also by creasing margins and service levels in the likely to be seduced by competitors at vulner- purchase occasion. process. Through its loyalty cards, Tesco sees able decision points. Additional sales gener- Cross shopping is increasing. The same con- what, where, and when customers buy across ate additional insights into consumer behav- sumers who buy their computers at a big-box the full range of store formats. On the basis of iors under a wide variety of shopping electronics store are heading to a neighbor- that knowledge, Tesco has built five special- conditions. They provide greater opportuni- hood electronics shop to pick up one-off pe- ized food formats in the UK: Tesco Super- ties to test innovative approaches. ripherals (accessories such as mice, printer store, a traditional grocery store for weekly Small-scale retailers used to count on local cartridges, and cables). By way of response, suburban shopping; Tesco Extra, a one-stop knowledge and scarce real estate to protect Best Buy is turning insights from its cus- hypermarket for large shopping trips; Tesco them from the big boys. But those barriers are tomer-centric stores into new store formats Metro, a smaller supermarket for customers crumbling as sophisticated chains stretch in- that draw targeted segments of customers in high-density urban areas; Tesco Express, a formation technology and creative formats. who don’t always want to slog through the tiny convenience store tailored to quick trips Extreme localization pioneers are building big box. They are testing out smaller, more in local neighborhoods; and Tesco.com for powerful platforms for innovation. Better yet, convenient stand-alone formats with the Web shoppers. Each of these formats is, of they are finding space for new growth in launch of Geek Squad stores; Escape, a store course, clustered and localized to meet spe- crowded landscapes and improving their eco- that provides 25- to 29-year-old technology cific needs. Metro stores, for example, often nomics and customer loyalty in the process. harvard business review • april 2006 page 8

Localization

consumers. “It is not unusual for localization to ness. Indeed, our research found that large improve sales by 40% to 50% while simulta- manufacturers are less willing to collaborate neously reducing store inventories and mark- with, or offer their best terms to, highly decen- downs,” says Boyd Rogers, VF’s president for tralized retailers. supply chain. “We consider our localization J.C. Penney discovered this the hard way in capabilities to be one of our most powerful the late 1990s, when it ran into problems by al- competitive advantages.” lowing store managers to determine order VF combines third-party geodemographic quantities. Local managers turned out to be and lifestyle data with daily store-level sales too conservative. Seeking to minimize risk, data, extensive consumer research, and com- they would buy a wide variety of goods rather petitor analysis to develop localization strate- than concentrate on hot items. As a result, the gies with retailers, such as Kohl’s. VF has found, stores ran out of popular products quickly and for instance, that while many buyers now de- were left with swollen stocks of slow sellers. sire lighter-weight denim, male Hispanics still And because headquarters lacked information prefer heavier weights. Women in southern on what was in each store, central managers California tend to buy shorter denim skirts couldn’t even see the problems. Between mid- than those in northern California. Even stores 1998 and the end of 2000, Penney’s stock price in the same metropolitan area can exhibit very plummeted from $54 to $8. different demand patterns for jeans and other Then, in 2000, Penney’s embarked on a suc- clothes. A store in a community with a large cessful turnaround program under the direc- immigrant population, for example, will tend tion of its then-new CEO, Allen Questrom. to have greater demand for smaller-size cloth- Penney’s went from a decentralized company ing than a store surrounded by nonimmigrant whose buying and markdown decisions were Americans—a subtle testament to America’s made at the stores to a centralized, data-driven obesity problem. organization. The management team classified For one U.S. chain, VF created 40 clusters, stores into seven clusters on the basis of size based largely on consumer lifestyle segments and customer demand patterns, developed and purchasing patterns. Product assortments, merchandise and fixture modules, and con- marketing strategies, and supply chain sys- solidated purchase orders. It also developed tems are tailored to each cluster. VF uses rapid demand-based optimization techniques— data exchanges to study each store’s daily allowing product and price ranges, replenish- point-of-sales data—not just to replenish ment policies, as well as the timing and depth shelves but also to discover new demand of markdowns, to be tailored to store clusters. trends in colors and styles and foster innova- Over the next five years, Penney’s stock price tion. Through such efforts, VF and its retailers more than tripled. Comparable department are boosting sales substantially while also store sales (sales of stores open for 12 consecu- avoiding markdowns and returns. tive months), having eroded 2.3% in 2000, rose 3.4% in 2001 and 5% in 2004. Central Control, Local Touch As Penney’s discovered, efficient localization A shift to localization raises big management requires that most decisions be coordinated and organizational challenges. The early mov- centrally, by managers with a broad view of de- ers are, in fact, breaking through the old “cen- mand patterns and sufficient store-level data to tralization/decentralization compromise.” But distinguish real insights from random noise. To it’s tricky. Executives’ first instinct is often to support headquarters decision makers, leading empower local managers, giving them control retailers are building sophisticated information over, say, the selection of products on store systems that draw from many sources—census shelves or major promotional programs. and other demographic research; data from Such decentralization often backfires, for store scanners and loyalty cards; consumer sur- two simple reasons. First, local managers lack veys and unsolicited comments; Internet sales the depth of data, and often the skill, to make data; data from third-party syndicators like AC- consistently smart decisions about buying, Nielsen; and intelligence on competitors. Local merchandising, and operations. Second, giving managers and personnel are also critical local managers too much leeway can introduce sources of information—often picking up sig- costly complexity and inconsistency into a busi- nals that computerized systems can’t see.

harvard business review • april 2006 page 9

Localization

When Wal-Mart, for example, introduced ko- A World of Difference sher food to its store in Berryville, Arkansas, it Localization isn’t free. The shift requires was acting on a recommendation from the greater investment in data collection and anal- store manager. The company’s other data ysis. And however sophisticated the clustering sources had not uncovered the nearby Jewish effort, some economies of scale will need to be community. sacrificed—in purchasing, marketing, manu- Central coordination is also essential to forg- facturing, and store construction. Most coming close relationships between retailers and panies will want to focus their initial efforts product suppliers. Product manufacturers on areas offering the greatest and quickest re- have deep knowledge about how goods sell turn. For example, the investment is typically across all stores in a region. Retailers have lower and the payback faster on localizing equally deep knowledge about how products markdowns (typically less than one year) than sell across their networks of stores. Combining localizing base prices (often two years or those two troves of information allows for a more). But as localization skills grow, so do lo- much more comprehensive understanding of calization opportunities. The systems, data, both local demand patterns and the way they and organizational processes that first enable may cluster across regions. a company’s leap to localized markdown Leading from the center does not mean strategies greatly ease subsequent steps to the that local managers become unthinking ro- localization of pricing, promotion, and mar- bots. In fact, by centralizing data-intensive keting programs. (For examples of retailers and scale-sensitive functions such as store de- pushing the frontiers of localization, see the sign, merchandise assorting, buying, and sup- sidebar “Extreme Localization.”) ply chain management, localization liberates Ultimately, all companies serving consumers store personnel to do what they do best: Test will face the challenge of local customization. innovative solutions to local challenges, en- It’s often been assumed that globalization im- gage with store guests, and forge strong bonds plies ever-greater homogenization of busi- with their communities. Wal-Mart’s store nesses and their products and services. The managers are legendary for highlighting hot world, in this view, will be packed with indis- items and responding to local pricing chal- tinguishable big boxes selling the same goods lenges. Best Buy encourages store employees and services to everyone. But a look at the to create and test hypotheses and share what emerging localization strategies of the leading they have learned throughout the chain. One companies in consumer markets—companies Best Buy employee recently hypothesized that once shunned customization but now em- that she could raise store sales by making brace it—reveals how mistaken this assump- iPods easier to find. She moved a display to tion is. We are advancing to a world where the the front of the store, created a shirt that said, strategies of the most successful businesses will “iPods here,” and raised the store’s sales rank- be as diverse as the communities they serve. ing from 240th to 69th. 7-Eleven knows that corporate headquarters could never predict a Reprint R0604E busload of football players arriving on a Fri- Harvard Business Review OnPoint 4109 day night, but the store manager can. Combin- To order, see the next page ing the efficiencies of a national chain with the or call 800-988-0886 or 617-783-7500 entrepreneurial touches of a mom-and-pop or go to www.hbr.org convenience store, 7-Eleven has created a system that it calls “centrally decentralized.”

harvard business review • april 2006 page 10

Further Reading This article is also available in an enhanced Harvard Business Review OnPoint edition, (Product no. 4109), which includes a summary Harvard Business Review OnPoint of its key points and company examples to help articles enhance the full-text article you put the ideas to work. The OnPoint edition with a summary of its key points and also includes the following suggestions for a selection of its company examples further reading: to help you quickly absorb and apply the concepts. Harvard Business The Triple-A Supply Chain Review OnPoint collections include Hau L. Lee three OnPoint articles and an Harvard Business Review overview comparing the various October 2004 perspectives on a specific topic. Product no. 8096

Designing High-Performance Jobs Robert Simons Harvard Business Review July–August 2005 Product no. 1517

The Four Faces of Mass Customization James H. Gilmore and B. Joseph Pine II Harvard Business Review January–February 1997 Product no. 97103

To Order

For reprints, Harvard Business Review OnPoint orders, and subscriptions to Harvard Business Review: Call 800-988-0886 or 617-783-7500. Go to www.hbr.org

For customized and quantity orders of reprints and Harvard Business Review OnPoint products: Call Rich Gravelin at 617-783-7626, or e-mail him at [email protected]

page 11