CORE Metadata, citation and similar papers at core.ac.uk

Provided by Research Papers in Economics

Making Category Management More Practical Ron Larson

The historical roots of the category-management supermarkets reduced duplicate items a little, they concept can be traced back more than 20 years. could boost category sales. Stores that eliminated The concept came to the forefront in the early too many “duplicate” items experienced category 1990s as part of the Effi cient Consumer Response sales declines. Therefore, an optimal level of va- (ECR) initiative. One of the early diffi culties was riety exists that may be slightly below what many to defi ne category management, and many, some- supermarkets are offering. Category management what confl icting defi nitions have been presented attempts to identify the optimal level of variety for (Dussart 1998). Harris and McPartland (1993) a retailer. said it consists of three interrelated elements: a Early success stories encouraged retailers and philosophy for strategically managing a business manufacturers to adopt category management. In that recognizes categories as strategic business 1997, Giant (Landover) said they were pleased with units, a process through which retailers and sup- the concept and were expanding their efforts (Chain pliers jointly develop category plans, and an or- Store Age 1997). It helped Supervalu reduce items ganizational concept that dictates the integration in test stores by 12 percent and increase category of buying with merchandising. This paper uses a sales by 6.5 percent (Purpura 1998). In 2003, Spar- simplifi ed “process” defi nition: category manage- tan claimed category management was the key to ment is the use of sales data, buyer profi les, and rebuilding their retail business (Ghitelman 2003). store characteristics to make product-assortment As category management became more popular in and shelf-arrangement decisions. The logic behind the food industry, it also spread to other industries the concept is that faster-selling products deserve including automobile parts, apparel, hardware and more shelf space and better shelf positions. This home supply, and bookstores. research looks at how the concept has evolved and Even after 10 years, category management suggests a practical approach that seems to address is being employed in new ways. An ACNielsen many of the criticisms of category management. survey found greater use of many category-man- agement practices (Mass Market Retailers 2004). Contributions from Category Management Cannondale Associates reported that more retailers and manufacturers rated category management as One reason category management is important is highly important in 2004 than in other years (Con- that consumers like having choices. Ratner, Kahn, venience Store News Online 2004). Another survey and Kahneman (1999) found that people choose found that category management was the area with less-favorite options on some occasions to experi- the most collaborative progress in 2004 (Veiders ence variety. Stores that did not offer consumers 2004). A majority of food-industry members seem some variety probably would lose customers to to believe that category management offers much stores that gave people more choices. In addition, potential. the assortments people are exposed to and how products are organized effect their choices (Kahn Criticisms of Category Management and Wasink 2004). Therefore, shelf arrangement can be an important tool for infl uencing sales. Un- As category management was being embraced by fortunately, retailers sometimes provide too much many retailers and manufacturers, there was grow- variety. Research by Willard Bishop Consulting and ing criticism about how the concept was being used. Information Resources, Inc. (1993) found that if Glen Terbeek (1993) of Andersen Consulting argued that if category management is not selectively used Larson is associate professor, Department of Marketing, as a tactic within an overall marketing strategy, it Haworth College of Business, Western Michigan University, will cause serious problems. He was concerned Kalamazoo. that category management treats each category 102 March 2005 Journal of Food Distribution Research 36(1) separately and does not consider how changes will assortment developed using sales data from this affect complementary or competitive categories or period would give the product less shelf space and shopper experiences in the store. In addition, prac- less on-shelf inventory, so the out-of-stocks would titioners of the concept were using chain-level, not continue. New products present another problem. store-level, data and were measuring performance If a store chooses not to stock a new item that its with sales or gross-margin dollars instead of direct- customers would want, a category-management margin dollars per square foot of space or customer analysis of that store’s sales would not identify loyalty. Terbeek believed that tailoring each store to the mistake. The process also depends on accurate the needs of the customers who shopped that store sales information. Given that stores sometimes was essential for improving store profi tability. have scanning problems (e.g., scanned prices too After experimenting with category manage- high or too low, items missing from the scanner ment, some retailers decided that it was not useful. database, etc.), analyses that rely on this data could An executive from a large Southern supermarket be fl awed. A similar problem is that a manufacturer chain declared, “Category management is dead” could focus a brand’s promotions on the period used (Wellman 1997). Sainsbury, a large chain in the in a category-management analysis, gain additional United Kingdom, eliminated its entire category- shelf space based on the infl ated sales rates, and management department because the benefi ts did then stop promoting the brand. The development not appear to exceed the costs (Storey and Benady of category captains (i.e., manufacturers who are 2001). In a survey of retail executives by Ernst and selected by retailers to counsel them on assortment Young, 40 percent thought the benefi ts of category and shelf-arrangement decisions) has raised some management had been exaggerated and 73 percent legal issues because the advice given may be biased thought the industry had shown little improvement (Desrochers, Gundlach, and Foer 2003). with its use (Supermarket Business 1997). Stanton Many category management analyses work with (1999) agreed that category management should die aggregate data. This could be at the chain level, so because it is a bad practice. store variations may be ignored, or they could be The list of specifi c criticisms is fairly long. Ini- at the store level, so preference differences across tially, fi rms lacked an understanding of category customer segments may not be considered. The pro- management and needed considerable training. cess does not look at customer satisfaction. A study Because the focus is on individual categories, the by Clayton/Curtis/Cottrell found that when many defi nition of a category affects what products are stores started practicing category management, their considered in each analysis. Often the category shoppers did not notice any differences (Progressive defi nitions that manufacturers use may not make Grocer 1995). In general, the process provides few much business sense from a retailer’s perspective. consumer benefi ts. When one retailer started using it Another problem was that category sales drivers and in one category, prices increased and sales decreased the supply-chain costs were not usually considered. (Basuroy, Mantrala, and Walters 2001). Although The typical category-management analysis does not this may have increased retailer profi t, the change address when to expand the space allocated to one could also have encouraged some consumers to category at the expense of another category. Distri- shop at competing stores. Most category manage- bution- and handling-cost differences between items ment analyses also do not consider the benefi ts of were generally not incorporated into the analysis. consistently carrying the brands that a customer seg- As fi rms started using category management, ment wants to buy (Krishnan, Koelemeijer, and Rao they experienced diminishing returns from the 2002). Dropping slow sellers could lead people to process. A survey of manufacturers by Silvermine switch stores because a key item on their shopping Consulting Group revealed that 41.7 percent es- lists is no longer available. timated that it took between 100 and 500 hours There are also three technical problems with cat- to complete a category-management project and egory management analyses. First, category man- 19.4 percent thought it took more than 500 hours agement usually emphasizes the number of facings (Kelly 1996). Besides being time- and resource-in- for each brand, not shelf placement. Dreze, Hoch, tensive, the analysis can perpetuate in-store errors. and Purk (1994) concluded that shelf placement If a product is out-of-stock, its sales are reduced is more important than facings. Second, the role and competing brands may gain volume. A shelf of each category in a store’s marketing strategy is Larson Making Category Management Practical 103 undefi ned. Some analyses treat every category the because their buyers tend to make large transac- same. Dhar, Hoch, and Kumar (2001) suggest that tions. price, promotion, and assortment should vary by The second approach attempted to model sales category role. They recommended four category trends using time-series data and identify the effects roles: staples, variety enhancers, niches, and fi ll- of promotions and other variables. This has proven ins. Changing a category’s role would change the to be very diffi cult because promotion effects vary tactical recommendations from an analysis. Finally, and some may have long-term impacts on sales. average sales data is used to make incremental as- Jiang et. al. (2004) developed a Bayesian vector sortment and facing decisions. Lee and Brown error-correction model that attempted to minimize (2001) showed that average revenue is not a good the problem of nonstationary time series. This ap- measure for shelf-space reallocations in the juice proach could help identify sales gains from assort- category. The marginal effect of each facing needs ment changes. to be estimated to make better decisions on shelf The next method involves constrained optimiza- space. Unfortunately, most category-management tion. Several researchers have proposed maximizing analyses do not estimate the marginal gains from a store’s revenue or profi ts by selecting the best facings. prices and quantities sold for each item subject to demand and shelf space constraints (e.g., Borin Addressing the Concerns and Farris 1995; Urban 1998; Lim, Rodrigues, and Zhang 2004). Items with zero quantity would not Some people who work in the category-manage- be stocked on the shelf. There are several varia- ment fi eld have tried to address these issues. One tions of this method, but a common challenge is the idea is to move from category to aisle analysis (i.e., problem’s complexity. Standard algorithms cannot aisle management) or to develop new “supercat- solve it, so the researchers proposed alternatives egories.” This might reduce the cost to analyze the (e.g., simulated annealing, greedy search heuristics, entire assortment offered by a department because it genetics algorithms, multi-stage experiments, tabu would decrease the number of separate studies. The search, squeaky-wheel optimization etc.). Besides traditional category-management process used in an the challenge of using these sophisticated algo- analysis has eight steps. ACNielsen recommended rithms, a major diffi culty is that the demand that is changing the process to include nine steps (Kent relevant for retailers includes many nonlinearities 2004). Their new process included clustering stores that cannot be described with a smooth function, based on the characteristics of consumers in their such as price-ending effects (Schindler and Kibar- neighborhoods to reduce some of the aggregation ian 2001). Therefore, the results may not be readily problems. Cannondale Associates developed their applicable to a retailer’s needs. own new category-management process with fi ve The fi nal method from academic researchers steps (Cosgrove 2003). The focus of the fi ve steps involves testing assortment and shelf-arrangement was to speed up the analysis and help differentiate options with virtual shopping simulations. In these retailers. Although these proposals provided some studies, shoppers navigate a store environment on minor improvements, many of the criticisms of a computer screen, examine products on the store category management were not addressed. shelves, and make selections. This approach has Academic researchers have also considered the been used to test store and shelf arrangements in problem and have developed four approaches that several industries (Burke 1996; Needel 1998). could be helpful for product assortment decisions. Tests on the method’s validity have shown that The fi rst approach involves studying and model- purchases in virtual stores are similar to purchases ing the contents of shopper market baskets. For in real stores with the same confi guration (Burke et. example, Russell and Kamakura (1997) divided al. 1992). This method could be used to test several consumers into segments based on their purchases, candidate category assortments and arrangements estimated the category structure, and noted syner- to identify which generated most store profi ts. This gies between categories using a probabilistic model. method would probably be very useful for develop- This approach would include complementary and ing and testing principles that category managers competing categories in the analysis and identify could use for designing assortments and shelf ar- slow-moving items that are profi table for a store rangement for stores. Testing each category plan for 104 March 2005 Journal of Food Distribution Research 36(1) every store layout would be cost-prohibitive given References the wide variation in store confi gurations. Basuroy, S., M. Mantrala, and R. Walters. 2001. A Practical Category-Management Approach “The Impact of Category Management on Re- tailer Prices and Performance: Theory and Evi- If each item’s retail price is fi xed at a level that a dence.” Journals of Marketing 65(4):16–32. store might use, the constrained-optimization ap- Borin, N. and P. Farris. 1995. “A Sensitivit Analysis proach becomes much simpler to solve. The objec- of Retailer Shelf Management Models.” Journal tive can be expanded to examine the profi t generated of Retailing 71(2): 153–171. from three shelf regions: the prime location (about Burke, R. 1996. “Virtual Shopping: Breakthrough in 51 to 53 inches from the fl oor), above the prime lo- Marketing Research.” Harvard Business Review cation, and below the prime location. Each of these 74(2):120–131. areas would have a space-available constraint and Burke, R., B. Harlam, B. Kahn, and L. Lodish. 1992. a function linking quantity with facings for each “Comparing Dynamic Consumer Choice in Real item in that area. Distribution and handling costs and Computer-Simulated Environments.” Jour- for each product can be incorporated into the profi t nal of Consumer Research 19(1):71–82. function. The functions linking quantity with fac- Chain Store Age. 1997. “Category Management is ings should incorporate important complementary Working at Giant.” 73(3):56,62. and substitute relationships and should have a nega- Convenience Store News Online. 2004. “Study tive second derivative over the range of reasonable Shows Continued Support for Category Man- facing levels. agement.” March 17. www.csnews.com. Stores could be grouped using the characteristics Cosgrove, J. 2003. “The Evolution of Category of the people living around them, as ACNielsen Management.” Beverage Industry 94(2):50. suggests. Stores in the same cluster are likely to Desrochers, D., G. Gundlach, and A. Foer. 2003. have similar responses to marketing stimuli. The “Analysis of Antitrust Challenges to Category quantity responses to different facing levels in each Captain Arrangements.” Journal of Public Policy shelf region could be developed using virtual store and Marketing 22(2):201–215. tests on typical stores from each cluster. Expanding Dhar, S., S. Hoch, and N. Kumar. 2001. “Effective the space constraints to the aisle level and adding Category Management Depends on the Role other categories from the same aisle to the objec- of the Category.” Journal of Retailing 77(2): tive function could help allocate category space. A 165–184. benefi t of this approach is that it is less dependent Dreze, X., S. Hoch, and M. Purk. 1994. “Shelf on having accurate sales data. Sales information Management and Space Elasticity.” Journal of could be used to set the constants in the quantity- Retailing 70(4):301–326. facings functions. Dussart, C. 1998. “Category Management: This optimization problem probably could be Strengths, Limits and Developments.” European solved using standard algorithms, perhaps even Management Journal 16(1):50–62. those in spreadsheet software. Although there Ghitelman, D. 2003. “Spartan: Category Man- continues to be a risk of not fi nding the optimum, agement Key to Rebuilding Retail Business.” the simplicity of the objective function and the Supermarket News 51(26):6. constraints makes this less likely. The method Harris, B. and M. McPartland. 1993. “Category could be repeated for several combinations of item Management Defi ned: What it is and Why it prices. This practical approach appears to address Works.” Progressive Grocer 72(9):5–8. many criticisms of category management and may Jiang, J., M. Zhong, G. Klein, and H. Chen. 2004. produce the type of useful results that were hoped “Nonstationary Brand Variables in Category for when ECR designers included the concept in Management: A Cointegration Perspective.” the initiative. Decision Sciences 35(1):101–128. Kahn, B. and B. Wansink. 2004. “The Infl uence of Assortment Structure on Perceived Variety and Consumption Quantities.” Journal of Consumer Research 30(4):519–533. Larson Making Category Management Practical 105

Kelly, P. 1996. “Category Mismanagement.” Food Sake of Variety.” Journal of Consumer Research and Beverage Marketing, May:16–17. 26(1):1–15. Kent, S. 2004. “Bringing the Consumer into Cat- Russell, G. and W. Kamakura. 1997. “Model- egory Management: Making an Old Process ing Multiple Category Brand Preference with New Again for Today’s Retail Environment.” Household Basket Data.” Journal of Retailing Consumer Insight 6(2): 17-19,34–35. 73(4):439–461. Krishnan, T., K. Koelemeijer, and R. Rao. 2002. Schindler, R. and T Kibarian. 2001. “Image Com- “Consistent Assortment Provision and Service municated by the Use of 99 Endings in Adver- Provision in a Retail Environment.” Marketing tised Prices.” Journal of Advertising 30(4): Science 21(1):54–73. 95–99. Lee, J. and M. Brown. 2001. “The Effect of Shelf Stanton, J. 1999. “Category Management is Dead!” Space on Sales of Fruit Beverages.” Journal of Food Processing 60(1):28–30. Food Products Marketing 7(1/2):3–13. Storey, J. and D. Benady. 2001. “The Final Cur- Lim, A., B. Rodrigues, and X. Zhang. 2004. “Me- tain?” Marketing Week 23(50):28–20. taheuristics with Local Search Techniques for Supermarket Business. 1997. “Category Manage- Retail Shelf-Space Optimization.” Management ment ‘Disappointing.’ ” 52(6):9. Science 50(1):117–131. Terbeek, G. 1993. “Why Category Management will Mass Market Retailers. 2004. “Category Manage- Fail.” Progressive Grocer 72(9):11–15. ment in Vogue.” 21(10):135. Urban, T. 1998. “An Inventory-Theoretic Approach Needel, S. 1998. “Understanding Consumer Re- to Product Assortment and Shelf-Space Alloca- sponse to Category Management through Vir- tion.” Journal of Retailing 74(1):15–35. tual Reality. Journal of Advertising Research Veiders, C. 2004. “Beyond Zero-Sum Relation- 38(4):61–67. ships.” Supermarket News 52(24):18–26. Progressive Grocer. 1995. “Will Category Man- Wellman, D. 1997. “A Categorical Rejection.” agement Kill the Supermarket?” 74(August Supermarket Business 52(9):69–73. Supplement):5–9. Willard Bishop Consulting, Ltd. and Information Purpura, L. 1998. “Supervalu Reporting SKU Drop, Resources, Inc. 1993. Variety or Duplication: A Sales Rise.” Supermarket News 48(13):39–40. Process to Know Where You Stand. Food Market- Ratner, R., B. Kahn, and D. Kahneman. 1999. ing Institute: Washington, D.C. “Choosing Less-Preferred Experiences for the