NEW YORK -- Brand marketers can use data from retailers' frequent-shopper programs to test and evaluate consumer promotions.
By working with households in a single market and accessing data that the retailers are "willing to sell at very reasonable prices, a whole range of target-market testing, or simply enhanced matched-panel testing, has become a very real possibility," said Paul Taylor, a promotion researcher.
"It is possible, for example, to figure out if in-store delivery of coupons and samples is more effective than direct-mail delivery." Taylor, account supervisor with Cincinnati-based Promotion Decisions, spoke about "the capability and the power" of dealing with huge data bases at a conference sponsored by The Marketing Institute, a division of the Institute for International Research, New York.
"In any given market you could simultaneously test up to 10 different promotion variations with up to 25,000 shoppers in each cell," he said. "The cells can be derived by taking demographically balanced cross-sections of the market or balanced store samples. "Then, testing can be conducted at the individual store level with in-store activity, with targeted direct mail or with some support from the FSI distributors -- geographic splits of FSI vs. other types of promotions."
He distinguished between two types of data: longitudinal and transactional. Both are valuable for promotion evaluation, he said. Longitudinal data come from tracking specific households across time. Transactional data, though not by definition a frequent-shopper program, enable marketers to access the contents of all shopping baskets on an individual-transaction basis.
According to Taylor, about 25% of supermarket chains are currently involved in some form of frequent-shopper program.
There are currently seven major chains in the country that have made the leap from frequent-shopper programs to a research-quality data base, according to Taylor. They include Smitty's, Vons, Schnuck, Wegmans, Super Fresh, A&P and Safeway. This represents a new source of household purchasing data for the 1990s, according to Taylor. "Other chains may well have data bases of that quality at this time," said Taylor. "We simply have not worked with the data and validated it alongside store scanner sales and so forth. At least three additional chains will be coming on stream soon, so the total availability of households grows from 5 million to about 6 million."