What Needs to be Fixed in the Market Research Profession?  (from my perspective)

I think this question and the relating need should be asked frequently if not at least annually. Those who have heard me speak on "Researching Research" know what I am about to say, but for those who have not heard the presentation, here are my thoughts:

We, as market researchers, supply management with data encouraging the expenditure of millions of dollars to put approximately 20,000 new products on the market each year. Within 12 months, nine out of every ten of these new products will fail. At the same time, we offer data to reject approximately 190,000 new ideas each year (and how many of these ideas were really good ones?). Bottom line is we need a more effective way of predetermining the success and failure of new ideas.

Let us look at the primary tool currently used to determine the market viability of a new idea, the Simulated Test Market, its potential weaknesses, and the attributes of an alternative method that just might improve our success rate.

The key ingredients of the Simulated Test Market are: the population sample, the measure of acceptance, the "black box" predictive model and the benchmarks for the predictive model.

Regardless of our target population sample, random or specific, less than one-third of eligible people will agree to participate and in many cases depending on the protocol, it is less than ten percent (10%). We can balance our sample by such factors as age, family composition, etc. However, we must ask how the volunteers differ from the non-volunteers in the way they think and act and how this might affect the assessment of new ideas. The test sample is a very big problem.

The measure of product acceptance is usually some stated purchase intent scale. These types of scales will frequently show purchase intent interest in the 75% range, much higher than we see in actual practice, requiring drastic adjustments to reflect true purchase frequencies.

These rough estimates, purchase intent scores, are then plugged into the "black box" and adjusted based on average historical results. We must ask ourselves tough questions about historical databases such as: age of the data, percentage of successful predictions, both market success and failure, number of results from the same product category, etc.

Having said all this, what information do I think we should have to help reduce the number of failures entering the marketplace? Basically, I would like to have real purchase data from real stores by people who are not in a test environment. I do not want to have to make projections about sales from purchase intent scores that must be assessed, based on some historical database. I want to use a minimum of product, have minimal disruption to the market and at a reasonable cost (comparable to an STM).

The research method that does all of the above has been available under the name of "Disposable Test Market" and now there is an enhanced version about to be released. This new version will be called RealTest which is about to be released by Elrick & Lavidge. This new version is a result of the efforts of Dr. Richard Fox, University of Georgia "Masters of Marketing" program, Dr. Frank Bossu, Vice President, Elrick & Lavidge, Mr. Gary Dispensa, Elrick & Lavidge and myself.