Understanding is Overrated.

If you know me, you’ve probably noticed that I love analytical models.  During my career I’ve made a point of collecting these tools and applying them to the work I do for my clients.  I find these tools to be invaluable and rely upon them heavily – they make me a better researcher.  By drawing upon the thinking of very smart people, we can make ourselves that much smarter.  However, I’ve also learned to be judicious in how I use these tools.  As the statistician George Box famously observed, “All models are wrong, but some are useful.” The reason Box called them ‘wrong,’ is that models are – by their fundamental nature – oversimplifications.  And, while over-generalization can be powerful analytical technique, we must use it cautiously, and with full awareness of its limitations and risks.

I love a pretty diagram with colored boxes and circles and arrows as much as the next guy, but I regularly remind myself that such a diagram represents a simplistic version of reality, not reality itself.  The human world is a lot messier.  Take, for example, Jonathan Haidt’s Moral Foundations Theory model, depicted at the top of this article.

This is an indispensable tool that I use frequently—it’s great for helping us think about the role morality plays in forming perceptions and driving behaviors.  However, morality in the real world is a bewilderingly complex tangle of psychological motivators, evolved traits, cultural norms and social dynamics.  While Haidt’s model is valuable, we should never make the mistake of thinking morality can be reduced to six little colored circles.

Here’s a common tool marketers use: the consumer decision tree – also often called the purchase decision hierarchy.  This type of representation is essential for understanding consumer decision-making and buying behavior. A lot of shopper insights work I do is designed around this technique.  However, consumer decision-making is a not a neat process at all.  It’s not really a decision tree, it’s more like decision soup, and all we can realistically hope to do is identify commonalities and tendencies.  This isn’t to say that we shouldn’t engage in the exercise of creating the tree; we just have to be mindful of its limitations.

Another example: consumer segments.  When we do segmentation studies, once a segmentation solution has been identified, we often like to give the individual segments interesting names and even write up personifications.  This is a worthwhile exercise, but it’s valuable when looking at the results of a segmentation study to examine the underlying data for yourself, rather than just relying upon the segment names and descriptions.  You’ll probably see pretty quickly that the true picture is much more complicated, contradictory and nuanced.

The bottom line –  sometimes we work too hard to understand. We want our explanations to be too pat, and way too all-encompassing.  Be suspicious of descriptions that are extremely simple.  Instead, revel in the messiness.  Put another way, don’t let an analytical model become a surrogate for thought.  Models facilitate and organize our thinking, but they’re no substitute for careful analysis.

And remember what H.L. Mencken once wrote: “there is always a well-known solution to every human problem — neat, plausible, and wrong.”