In a previous post I suggested that modelers could improve their results by splitting
their datasets according to some critically important variable, such as Tenure (the length of time a customer has been on the file) and then build separate models for each major segment.
The argument being that it is intuitive that the usual set of modeling suspects (Recency, Frequency, Monetary Value, Products Purchased, Source and the whole set of Demographic Variables) will display different relationships with Response or Sales, depending upon the Tenure Segment, and that just adding Tenure as a variable, without taking interactions into account, isn’t sufficient to capture the full effect of this variable.
As if this isn’t complicated enough, I came across an article that questioned fundamental direct marketing beliefs, including the belief that there is a strong positive relationship between customer lifetime and profitability in a non-contractual relationship. In other words, they think that direct marketers think that customers that kind of hang around a long time, buying every once in a while, are profitable and every effort should be made to enhance the relationship between buyer and seller.
Of course, direct marketers who have looked closely at the data know that the costs of servicing infrequent buyers may indeed exceed the margins they yield; and the authors discovered for themselves that the simple relationship between lifetime months on file and lifetime profits is relatively weak (r = about .2 for the two groups studied).
What I did find interesting and potentially actionable was that they could divide a
significant number of customers into four meaningful groups:(Some 9000 households were studied over a three-year period. The households were correctly split into two cohort groups, January and February starters.)
Segment 1. Those that had relatively Long Active Lives and High Lifetime Revenue
Segment 2. Those that had relatively Long Active Lives and Low Lifetime Revenue
Segment 3. Those that had relatively Short Active Lives and High Lifetime Revenue.
Segment 4. Those that had relatively Short Active Lives and Low Lifetime Revenue.
The Graph below indicates that customers in Segments 1 and 3 kind of look alike, behave in a similar fashion, over their first 12 months and then begin to separate over time. No doubt that this is true, the operable question is can this disparity be predicted, and predicted early enough in customer’s life so that corrective action taken be taken.
The argument is that simple RFM analyses will miss this phenomena, and that database marketers, as a consequence of their not understanding that their database consists of these segments, will overspend on the Short Life-High Revenue segment, before traditional RFM analysis will depress mailings to this segment.
So, the key question for marketers is, if this effect is widespread — if there really are customers that come in for a short while, buy a lot and then leave — can they be detected? Will modeling Tenure Segments, as suggest above, and in last month’s article capture this effect. Probably not, at least not by itself. What might work is a Principal Component Analysis of the available purchase behavior data over the last six months.
This approach might discern either a trend in dollars spent, or a trend in the particular products purchased that would indicate that the customer was displaying a pattern associated with customers that buy heavily for a short while and then switch to someone else – for reasons we can only speculate about.