As a result, pricing is simplistic and formulaic at many companies. They either peg their prices to competitors' prices, called pricing to the market, or they add up production costs and tack on a standard markup, called cost-plus pricing.

A third and better approach, value pricing, involves segmenting buyers and setting prices based on the characteristics of each segment. For example, airlines guess that the business executive buying a ticket for a flight across the country next week will pay $1,000 for his seat, but the man planning a vacation for his family next summer won't pay more than $250 for the same seat. Value pricing requires good data mining and modeling capabilities as well as good analytic tools for understanding the results of pricing decisions.

Rich Farrell, vice president of information services at Piggly Wiggly Carolina Co., says that until 2005, the 115-store grocery chain used old-fashioned, ad hoc methods to set prices. "We knew the cost of product from suppliers. We had designated markups. We kind of knew what our competitors were offering, and it was a manual process," he says.

Now Piggly Wiggly uses the Consumer Demand Management service from DemandTec Inc. in San Carlos, Calif. Piggly Wiggly provides point-of-sale transaction histories and information about its 20,000 products to the model, which is run by DemandTec. It also specifies three objectives: Increase sales x percent, increase margins y percent, and maintain a favorable "price image" (neither too high nor too low compared with competitors). Suggested prices from the model, by product and store, go back to a pricing group at Piggly Wiggly.

The four-person pricing group is new, Farrell says. It has taken over the work previously done by buyers and others throughout the company while also maintaining the models run by DemandTec. "You need some dedicated people to manipulate and operate the model," Farrell says.