Research is Exploration

Jul 20, 2022 / Mind Cart AI

Market research seeks to answer a simple question: What do people want?

Market research seeks to answer a simple question:  What do people want?  In doing so, research is in many ways like exploration.  When we want to understand a consumer, we map consumer preferences along multiple dimensions.  Much like a cartographer who maps new land, measuring the topographical area and deducing the nature of the structure below, we seek to map the mind of the customer on any given topic—whether it be taste in chocolate, vaccine protocols, human sexuality, or otherwise. 

At the heart of exploration is the search for the unknown–the discovery of something that no one else has found.  It requires audacity and vision.  As the great British explorer Ernest Shackleton, who led three British expeditions to the Antarctic, once described his own journey to cross Antarctica on foot, it was to be “a greater journey than the journey to the [South] Pole and back” and the “largest and most striking of all journeys.”  [See Alfred Lansing, Endurance 9]  At the same time, though, exploration is fraught with danger.  It requires planning, yet also an open mind and adaptability to change.  

Market research is exploration in both of these senses.  Our goal is to map a previously un-mapped area.  It is the discovery of something altogether new and not yet understood.  Yet one can easily be led astray and into danger.  And sometimes, one learns something that requires one to revise course—or more radical yet, requires one to abandon ship altogether.  

The metaphor, of course, only goes so far.  But keeping the connection between research and exploration in mind may help us avoid many of the common trappings of market research.  Too often, market research is used simply to validate what one either already knows or worse, what one believes one knows.  Nearly fifty years ago, Howard Moskowitz, the pioneer of the methods underlying MindCart AI’s technology, encountered this very problem when, as a young consultant, he began work for PepsiCo.  In 1965, in a stroke of serendipity, scientists had discovered aspartame–the artificial sweetener now used ubiquitously in diet sodas.  Pepsi’s task for Dr. Moskowitz: Find the optimal amount of aspartame to create the “perfect” diet Pepsi.  Pepsi explained that it knew that 8% was the lower bound—not sweet enough—and 12% was the upper bound—too sweet.  So, Dr. Moskowitz sought to find what Pepsi had told him to find.  He conducted experiments, varying the amount of aspartame in different variations of Diet Pepsi.  But far from finding an optimum, he found a jumble.  The data was inconclusive; there were preferences all over the spectrum.  The problem, Dr. Moskowitz realized, was that Pepsi had misunderstood its consumers, and simultaneously, artificially limited its research range.  It had treated research as validation–confirm what we know–rather than research as exploration–find what our customers want.

Having had the problem framed in this narrow way, it took some time before Dr. Moskowitz realized that the problem hadn’t been posited properly.  Only by taking a step back–revising one’s course, if you will–did Dr. Moskowitz realize that the issue was that there wasn’t one perfect Diet Pepsi, but rather many.  In other words, different kinds of customers enjoyed different things–and along different dimensions.  By taking this step back, treating the problem not as to merely validate what Pepsi had told him to validate, and ultimately, exploring the problem from first principles–finding what customers wanted–he was able to explain the jumbled data. 

To be sure, validation has its role–but at a much later stage than it is typically conducted.  Knowing the lay of the land is critical.  Once one can see the entire landscape, it becomes easier to deduce the underlying features of any particular sub-problem.  

That’s our goal here at MindCart AI:  Through pattern analysis, we help map the problem you are trying to solve along multiple dimensions.  Thus, rather than inductively test one issue at a time, pattern analysis is deductive.  Pattern analysis ensures that each respondent sees a different set of twenty-four screens, allowing the researchers to measure interactions within and across combinations of messages, rather than repeatedly testing the same combinations.  We do all this knowing that we are not merely validating what someone believes he or she already knows.  We do it as an explorer–with a plan, a vision, and open-mindedness.  We hope to find out something new.