The veil of statistics

When most Marketers hear the word “Customer Data”, he or she likes to look at the Statistician in the room, to ask him what the data means. If there is no statistician in the room, then the conversation stops and is redirected into safer waters. That’s where the problem starts. It’s sad but most marketers have only two attitudes towards the Marketing Statistician: Blind Ignorance for any value of this field or blind admiration for the mathematical skills of deciphering hidden treasures from stream of data. Both is not just plain wrong but also dangerous.

Wikipedia calls Statistics a “mathematical science pertaining to collection, analysis, interpretation and presentation of data.” Please notice it doesn’t talk about truth, action, or meaning; it talks mathematics. Mathematics is an attempt to build a parallel universe of connection and correlations that mirrors as closely as possible the world that we experience everyday. Mathematics is a descriptive tool set to better understand our world, so is applied statistics in our marketing world. Therefore, every marketer needs to have a basic understanding of how statistics work, what its limitations are, and how it can be used as a parallel marketing universe of understanding customer behavior.

Here are my simple and short guidelines of how Marketers can deal with Statistics and the analysts who love talking about it.

• No tech speak: Don’t get intimidated by the statistical jargon that a lot of marketing analysts use. Tell them to explain the problem, their approach, and the insight in basic terms without hiding behind meaningless technicalities. Example: If the analyst talks about regression analysis, let him/her explain what it is until you understand it.
• No Analytics without hypothesis: Start every analytical problem with a clear definition of the problem and what you expect as the outcome. The projected outcome should be used as a guiding hypothesis that can be verified or falsified, not as a description of what the analytical exercise should demonstrate. Example: Don’t tell the analyst to build a behavioral segmentation. Draw your best guess of this segmentation and give it to the analyst as a guiding principle in his/her data mining practice
• Visuals, Visuals, Visuals: Guide the Analyst to show the analytical outcome in a visual manner, not in a 30 folder large Excel file. It’s not about oversimplification but about being intellectually rigorous to distill the results into an understandable mode
• Ask, what’s the 60 seconds story?: Ask for the 60 second version of the statistical outcome and its relevance for marketing. If there is not a 60 second story, then there is no story at all. You can even tell the story of Dostoyevsky’s “The Idiot” in 60 seconds.

The successful application of statistics into the field of marketing is about pushing aside its veil which is too often very comfortable for both the statistician and the marketer. The statistician can hide behind it to be perceived as the highly sophisticated data geek. The marketer can hide in front of it, so he does not have to deal with all these more complex data issues. The solution resides in inviting the Analytical Geek out of his dark room into the bright light, where he or she can be part of a productive marketing discussion. It will be quite a bright light for most of us.