They know where you live

Today's Wall Street Journal (subs. req'd) has a nice analysis of the factors that lie behind Wal-Mart's troubled entry to the UK supermarket business. They took over Asda a few years ago but, to cut a long story short, the long-time incumbent Tesco has run rings around them and is now planning a limited entry to the US market. While various things explain Tesco's success, one of the most important seems to be the unusually high coverage and scope for data-mining of its loyalty card, the Clubcard.

It has 12 million members and the card is used by 80 percent of customers. Up to 20 percent of coupons distributed to members are redeemed -- 10 times higher than the usual store coupon redemption rate, and 95 percent of cash-back vouchers are used. With so many purchases recorded and an apparently reliable set of addresses behind it, the company has been able to do vast amounts of background research and manage store placement and strategy accordingly. One example:

Clubcard records showed shoppers at a small store in the town of Slough weren't buying full meals. Many people in the town have South Asian or Arab roots.

Tesco decided to replace the store with a supercenter. Focus groups confirmed that people in Slough were buying some products at Tesco but turning to smaller markets for many staples -- large stacks of rice, big canisters of cooking oil and Asian brands. Many criticized the small plastic packages of herbs at Tesco and said they wanted loose bunches that they could touch and smell.

When the new Slough store opened in August 2005, it offered more than 800 foreign products, up from 150 in the previous store. It has a large halal butcher shop, the latest movies from India, newspapers in Arabic, Urdu, Punjabi and Bengali, and a jewelry counter with bangles in yellow 22-karat gold popular in India. The shopping carts are lower and flatter to fit big sacks of rice and flour.

Tesco wanted to know if the strategy was working, so it turned to Dunnhumby [their research firm]. The analysis found that 36% of Slough shoppers were buying goods from the World Foods line. That figure roughly matched the proportion of Slough's nonwhite population.

Dunnhumby then checked addresses of World Foods buyers against government census data that identify immigrant neighborhoods. It turned out that more than a quarter of the World Foods customers were coming from largely white neighborhoods. By examining the shopping baskets of these customers, Dunnhumby concluded that upscale white shoppers with an interest in non-European food were responsible for some of the success of the World Food line.

In the following days, executives huddled over big maps showing Britain's ethnic makeup. They outlined a plan to roll out the World Foods line to 300 stores in immigrant areas as well as to 25 stores in mostly white parts of the country.

A few weeks later, Tesco stores in places like Holland Park, a leafy part of west London, and Bar Hill, an affluent town near Cambridge, were stocking fragrant herbs and frozen samosas. Tesco says the Bar Hill store is selling World Foods better than any of its other stores.

What's noteworthy is not the general idea -- white people want their Asian herbs too -- but the combination of their vast proprietary dataset with government data to guide business decisions. We can't resist this example either:

Shoppers who buy diapers for the first time at a Tesco store can expect to receive coupons by mail for baby wipes, toys -- and beer. Tesco's analysis showed that new fathers tend to buy more beer because they are home with the baby and can't go to the pub.

Finally, there may be a deeper message about the UK being the right "size" of market for a company like Tesco: small enough that it's feasible to have a national database with some basic common structures across households, but big enough to have lots of interesting variation that the data miners can exploit. Which incidentally means that the supposed advantages of George Bush's database of most phone usage in America are likely exaggerated, since it lacks all the other information and has none of the structure that Tesco's database has.