Millennium Marketing Research®

Tom Schori DBA Millennium Marketing Research®, 808 Ironwood, Normal IL 61761, 309-532-8466

Starting Page    Millennium Marketing Research?    Types of Research    Principals    Articles   Quote

A modeling approach to better predicting the 'cannibalization' effect.

By Thomas R. Schori, Ph.D., and Michael L. Garee, Principals,  Millennium Marketing Research, 808 E. Ironwood, Normal, IL 61761-5239. Tel. 309-532-8466 -

Every successful franchise operation is ultimately confronted by the issue of "cannibalization," the tendency for new franchise operations to become successful, in part or in whole, merely by "stealing" business from existing franchises in the same market. It’s an issue of vital concern to both the franchise parent company and to existing franchisees.

One effective way to better predict the "cannibalization" effect with a rather high degree of validity would be to design a statistical model that incorporates both consumer behavioral data and pertinent demographic and geographic data. We contend that such a well-designed statistical model could help franchisers better anticipate the probable "cannibalization" effect, resulting in far better public relations with existing franchisees as well as improved profitability for all outlets in any given market. Let’s look at an example of how such a model might work.

Let’s assume that ABC Restaurant, a very successful franchise operation, currently has three restaurants in City A, population 125,000. Each restaurant currently averages gross annual sales of, say, $1 million. The parent company, Acme Holding Company, is considering adding an additional restaurant in the market because it perceives that there is additional "capacity" in the market for its ABC Restaurants. But, being the type of company that judiciously wants to avoid doing anything that might alienate existing franchisees, before deciding whether or not to launch the new, additional restaurant, Acme Holding Company wants to be able to predict how much, if any, "cannibalization" might be expected to occur among existing restaurants when the new restaurant is introduced into the market.

In order to obtain consumer data to inform the statistical model, which we’ll simply call the Cannibalization Potential Model, it would first be necessary to administer a brief "one-on-one" survey to, say, 100 randomly selected customers as they are leaving each existing ABC Restaurant in the market, in order to determine, among other things, the likelihood of their switching to the proposed new ABC Restaurant location, their beliefs about the ABC Restaurant they’ve just patronized (quality of food, service, cleanliness of facilities, etc.), plus their beliefs about those same characteristics vis-à-vis the proposed new ABC Restaurant. In addition, the survey would determine customers’ bases of travel, i.e., did they come to the restaurant from work? Home? etc., as well as other pertinent demographic information.

These consumer data would then be incorporated into the Cannibalization Potential Model. In addition, the model would also take other pertinent factors into consideration, such as natural barriers (rivers, boulevards, irregular travel patterns, etc.) and psychological barriers (proximity of undesirable areas, highly congested thoroughfares, etc.) to customers’ willingness or potential to switch their patronage from existing ABC Restaurants to the proposed new ABC Restaurant. (These latter data, of course, would have to be obtained on a market-by-market basis.)

If the Cannibalization Potential Model is correctly designed, and is indeed shown to have incorporated the most pertinent, most important factors in determining customers’ likelihood of switching to the proposed new ABC Restaurant (and thereby "cannibalizing" business from existing ABC Restaurants), one could logically assume that in future markets it might not even be necessary to conduct further customer surveys. But, of course, that wouldn’t be known with certainty until the predictive validity of the model alone could be adequately measured.

In order to determine the predictive validity of the Cannibalization Potential Model, it would be necessary to determine, say, one year down the road, how accurate the model itself was in predicting "cannibalization," as well as how accurate the model used in conjunction with the customer inputs was in making that same prediction. For example, using the model and the customer inputs might prove to be 90% accurate in predicting switching behavior, i.e., the "cannibalization" effect, while using just the model itself might result in, say, 70% accuracy. At that point it would merely be a management decision as to whether or not it made sense, both from an economic and operational standpoint, to continue to incorporate the customer inputs in future market studies or merely to rely exclusively upon the Cannibalization Potential Model.

In any case, applied correctly, we believe that this approach could indeed result in the Acme Holding Company (or any similar business organization) being able to rather precisely predict the degree of "cannibalization" that might occur if they add additional outlets to markets with existing franchises. Such a tool would allow the franchiser to anticipate problems and deal with them before they occur, rather than forcing the franchiser into a reactive position, where it likely will be trying to "put out fires" for several years after the introduction of a new franchise outlet in markets with existing franchisees.