Econometric models, often referred to as marketing mix models when applied to sales and marketing data, evaluate the contribution each element of a marketing program makes to improve sales or share. This statistical technique brings marketers closer to uncovering such mysteries as which part of their ad budget is being wasted, what their optimal spending level is and what minimum marketing exposure levels should be. The models have even been used to show which creative advertising is superior. Soon, models will help marketers study consumer spending.
"When times are tough, people want to know that every dollar they spend is working as hard as it can for them," says David Meer, worldwide chairman of the Advanced Techniques Group of WPP Group's MindShare, New York.
Indeed, divisions within agencies offering market research and planning tools, such as Mr. Meer's group, report greater interest from marketers in their services over the past year. One of their hottest offerings now-even though it has been in use in less-refined forms for decades-is econometric modeling.
Econometric models require clients to invest large amounts of data and time. Modelers need to learn what each client's business challenges are and which questions the model should answer. Modelers then form a theory, build a model that processes data from the client's historical marketing programs and sales results, and use statistical methods to find patterns.
The more data modelers have-across many media channels and over long periods of time-the more accurate the models are, practitioners say.
"It becomes more effective when it is an evolved process, so don't just build a model and walk away from it," says Michele Madansky, director of marketing intelligence at Grey Global Group's MediaCom, New York.
Slim-Fast Foods Co. began using MediaCom's marketing mix modeling tool 18 months ago, when the package-foods company extended its product message and started using new media channels. According to Michiel Kruyt, Slim-Fast's VP-marketing, the service has already met his expectations and will become a fixed part of the media planning. "We're happy with it because it found our ideal weekly [gross rating points] and has been able to show us the effect on sales over time of the advertising dollars we spend. From that we can adapt our media plans accordingly," Mr. Kruyt says.
Thanks to scanner data, package-goods companies have been able to run marketing mix models for decades. Now the technique is gaining appeal among marketers in other categories too. For example, teams at WPP's Mediaedge:CIA build models for AT&T Corp., Payless Shoe Stores and Yum Brands' Pizza Hut.
Econometric modeling still has its skeptics, with some practitioners conceding that models are costly and fail on product launches.
However, Ms. Madansky says, "If you can explain 80% to 90% of variants in sales by using this technique, I don't know how you can argue with that."