The butterfly effect: estimating "faux-new" customers

Purpose - This paper sets out to explore how the principles developed by Sir Ronald Aylmer Fisher, FRS (1890-1962), a gifted British evolutionary biologist, geneticist and statistician, can be applied in today's retail environment to estimate new customers becoming visible through your loyalty or database marketing program - customers whom we like to call "faux-new" customers. Here, we have used the word "faux" to mean "fake" or "false" - customers who look new in the next month (because we did not observe them in the first month), but are indeed customers (because they made a purchase before the first month, a month for which we do not have data). That is, there are "faux-new" customers and "actual-new" customers whom we will observe in the second month.Design methodology approach - The paper uses data and statistics from numerous loyalty-marketing programs to support its conclusions. It investigates a technique discussed by R.A. Fisher in counting species and applies the technique toward counting customers. It showcases the Poisson distribution assumption in modeling customer frequency of purchases.Findings - The study found the technique to be robust against a large real-world data set: the technique was predictive despite some dubious assumptions.Practical implications - When statisticians and marketers attempt to estimate new customer growth rates, they must be careful. It is tempting to estimate the monthly growth in your customer base by observing the number of customers in July, and then observe the number of "new" customers in August. In this case, however, the results would be overestimated. Many customers who transacted in August but not in July are not actually "new" customers - they are "faux-new" customers. They could very well have transacted in April (assuming that one does not have the luxury of observing months before July). To estimate the actual number of new customers, it is necessary first to need to strip out the number of customers that were estimated "zero-purchase," or "faux-new," customers in July.Originality value - This paper explores innovative statistical techniques for marketers and business analysts who estimate customer growth based on multiple periods of customer transactions.