Marketers, Beware the Decimal Point
In CMA’s Weekly Watching Brief February 5th edition (accessible to CMA members), there was reference to a study from the US-based CMO council regarding the value of loyalty programs. I found this posting very interesting for many reasons, but mostly because it illustrates how easy it is to potentially mislead people, whether intentionally or not, by including a few choice numbers. In the classic 1950s book called “How to Lie with Statistics”, the author Darrell Huff describes how easy it is to prove whatever point you want by choosing which numbers to present and how to present them.
In the case of this posting, I am referring to its fairly rash generalization regarding loyalty and reward programs. There are probably as many different types of loyalty and rewards programs as there are published studies about them. Loyalty programs could be something large and complex, or as simple as a frequent coffee-buyer card from your local shop. To state that 61% of marketers believe that the consumers who take part in these programs are their best and most profitable customers demonstrates such an oversimplification as to make this statistic practically meaningless. How did the survey respondents choose to define loyalty program or best customer and which ones were included, or excluded? There are no consistent definitions of these concepts and I have rarely met a marketer who has actually pursued a data-driven assessment of their own program to find this statistic to be true. It depends on so many factors including the type of products or services being offered, the competitive context, the types of rewards being offered and the types of consumer behaviours required to earn these rewards. Depending on how a program is set up, its heaviest users could actually be the least profitable customers.
Too often marketers are willing to turn over any quantitative assessment of marketing initiatives to the data geeks or finance and take the answer at face value, without questioning the results (unless of course they are positive). There are usually many ways to skin the proverbial cat, including such things as definitions of test and control cells, definitions of success and what costs are included in profitability calculations. And depending on how these various factors are defined you could come up with very different results. Since these calculations are used to support decisions about potentially significant major marketing investments, you need to be completely confident in how these calculations were done and what was, or wasn’t, included. I strongly encourage marketers to get more involved in the analysis and understand the definitions being used, how the results are calculated and what other factors could influence the outcomes instead of simply going along with an answer because it was calculated to 6 decimal places.
By Paul Tyndall, Senior Manager, Predictive Modelling & Segmentation at RBC. Paul is also a member of CMA’s Marketing Technology and Database Intelligence Council.
