That was one finding of an IBM/Safeway data mining trial based on a pilot study of remote shopping at Safeway's Basingstoke branch.
David McKenzie, IBM's European director of pervasive computing, used the supermarket trial as an example to illustrate the capabilities of deep data mining techniques. It took three to five weeks to determine each customer's personal shopping profile, he told the conference.
For the pilot, customers were given Palm handheld devices loaded with Safeway product details, which enabled them to draw up shopping lists anywhere.
To place orders they simply plugged the Palm into a cradle at home for a direct link to the mainframe.
Following transactional analysis, the mainframe limited the download of products to their personal profile of 300 products, plus 100 products that they did not buy, but which were best related to those they did buy.
The end result was increased purchasing of both products the customer would normally buy and also of those they did not previously buy. As a result the pilot is being rolled out to six other Safeway stores.