We have posted in the past about Macy’s attempts to tailor its selections in each store to local tastes. That is, after spending all that time, effort, and cash to roll up regional department store chains and rebrand then all as Macy’s, they are trying to make sure that customers around the country experience their own unique Macy’s. Now both BusinessWeek (With Stores Nationwide, Macy’s Goes Local, Sept 30 and the source of the above graphic) and the New York Times (Atlanta Hats? Seattle Socks? Macy’s Goes Local, Oct 2) are offering their take on these efforts.
The new information in these articles that is surprising is just how Macy’s is going about tailoring its assortments. I would have assumed that in a world of data mining and cloud computing, this would have been a data-driven exercise. Rocket scientists would be pouring over point of sales data and picking up on trends from disparate stores. That apparently is not how they have implemented this (from the NYT article).
In essence, Macy’s requires sales clerks and store managers to examine the local population almost like anthropologists — studying, for example, what churchgoing black women here in Atlanta shop for compared with the shopping habits of Microsoft wives, as employees call one segment of shoppers in the store in Bellevue, Wash.
At the same time, the retailer doubled its staff overseeing store assortments and decreased the stores that staff members dealt with. It required the people responsible for merchandise assortment to visit stores daily, added log books at each register where sales clerks entered suggestions from shoppers, and introduced a review process so the staff visiting stores could make recommendations to buyers.
Remarkably, this human-driven system works pretty quickly (from the BW article).
Salesclerks record local requests from shoppers in log books to pass on to the district managers, who then electronically make about 1,000 requests to the main office each week for a customized mix of merchandise for their stores. Headquarters approves about 90 percent of the pitches, usually within seven days, Sluzewski says. Initially, top-level buyers resisted what district managers wanted to buy, Lundgren says. He also faced skepticism that a single buying office could supply everyone’s needs. “This was a major cultural shift for the company,” Lundgren says. “The hardest part was that not everyone could come along.” (Implementing the divisional consolidation last year led to about 1,900 job cuts.)
It seems that some of the data they are aiming for cannot be captured in a POS system. Think of this as the purple-pants problem. American men may be desperate to wear purple pants but no point of sales system will ever capture this because no department store carries purple pants for men. You only capture that information if sales clerks write down that middle age men are all asking for purple pants. That’s an extreme example, but women in Atlanta wanting white outfits for church well into the fall apparently is a reality.
The real power to this approach would seem to come from combining sales staff requests with data. Corporate is approving 90% of pitches. At some point, they should have a good way of judging the brilliant from the marginal. That could make tailoring assortments even more efficient.