LAS VEGAS -- Though speaking at a technology vendor's user conference, an executive from H.E. Butt Grocery Co. said retailers can't rely strictly on systems to make important merchandising decisions. The art of retailing also plays a role, and it often comes from the stores.
"I really think it becomes the ability of your merchants and operators to collaborate on what are the best items for a particular store, customer base and merchant grouping, and that's a lot of art," said Mark Beaty, director of merchandising and business intelligence for H-E-B, based in San Antonio. "You start with a scientific base, but there's a lot of art blended on the top."
Beaty spoke at JDA Software Group's Focus 2004 conference here in late April at a panel discussion on analytics, optimization and data mining.
"We know intuitively that we want a better science base -- we know that better data breeds better decisions," Beaty said. "But I would hate to think that we would ever get to the point where we let science rule completely. I still want that merchant who knows my products, my customers [and] my market area to have the ability to go out and take a risk for me once in a while."
Beaty did acknowledge that data mining has helped H-E-B in advertising and in knowing where to place items in stores. For example, when the chain advertised imported beer, it used data mining to put limes next to the beer, which created a phenomenal jump in lime sales. Other moves seemed less intuitively right, such as placing toothbrushes next to tuna fish, yet it led to an increase in toothbrush sales.
While data mining helped in these cases, he still maintained that stores come up with the best information. "They are front line with the customers," he said. "They get the info, and they know better how to service the customers."
Beaty used the example where in one region, it might take 58 stockkeeping units to get to 90% of sales potential, whereas in another region, it only takes seven SKUs to reach 90%. "I think that that's where you have to make sure that you challenge your folks to step out on that limb to look for the undetected patterns and selling potential you may not have tried before," he said.
While always stocking the basics, stores "need to look for differentiation, to look for value-added products, [and] to look for things your customers want that you didn't know they wanted. I don't know if there is ever a strong science base for determining whether you ought to carry a new item or not."
At H-E-B, decisions are made at both the corporate side and at the merchant side, he said. Recognizing the power of individual stores to know what is right for their customers, corporate will offer them recommendations in a "menu format," but let the stores decide. Beaty distinguished between products that performed the same way throughout the chain and those that did not. "You're really looking at product performance across tight groups of stores and wide groups of stores," he said. "If there's no difference, then it ought to be an enterprise decision. If you see variance, it ought to be based around what's best for this group of 15 stores or that group of nine stores."
He again stressed that it was important to allow individual stores to assert themselves. "Even if you have one store that's so far out on the edge that it doesn't match any group, let it be its own group," he said. "A cluster can be one store."
