I know, I’ve already lost half of you. And the other half that’s still hanging on is waiting patiently for me to start throwing formulae at you. But that’s not going to happen. I do want to point out, though, that while statistics are monotonous, boring, and somewhat difficult to understand they are one of the most important facets of marketing.

After all, marketing is all about studying a group of people, modeling their behavior, and creating a product that most effectively uses that behavior to drive profit. If not for statistics, creating these abstract consumer models would be next to impossible!

Unfortunately, few of us use “real” statistics in our day-to-day work. We’ll instead make lists of assumptions regarding our market and go out to test them with a live product or beta test. If proven accurate, we move forward. If proven otherwise, we re-evaluate our assumptions and try again.

The key phrase here: ** if proven**. How exactly do we “prove” an assumption. Let’s assume consumers will pay $10 for a particular widget with a particular function. We put it on a shelf and stand back for a week. Seven days later, we’ve sold 100 widgets. Does this prove our assumption, or do we need more information? If we know that only 100 people saw the display, does this prove our assumption? What if 1,000,000 saw it and only 100 bought it?

Our analysis never happens in isolation – we’re always pulling in other information to judge whether or not a particular test is relevant. This is the basis of statistical marketing, whether you crunch numbers or just eyeball the results of a market study.

Most statistic relevance is readily apparent. In the hypothetical example above, you can clearly see the difference between a 100% conversion rate and a .01% rate. No fancy theorems or formulae here. But other elements of marketing can have more subtle statistical underpinnings. How well-fit is your feature set to the market’s psychographic tendencies? How confident are you in your model of the market?

A lot of this has to do with sample size – if your market is 20 vastly different individuals, I would have a hard time placing any faith in your model. On the other hand, if your market is 2,000 individuals with similar interest, nearly identical demographics, and easily predictable purchase behavior I might be more willing to buy in to it.

At the end of the day, marketers need to ask themselves one question about their strategic plans: are my assumptions based on statistically relevant and reliable data? If the answer is “no” (or if you can’t even answer the question) then you might want to rethink where you’re spending your budget this month.