I recently reviewed a report generated by a business trying to evaluate how many multichannel customers it had. The report suggested that the majority of customers were multichannel. In the late 1990s, almost all of its customers were acquired via catalog marketing. In 2009, almost all of its customers purchase online. As a result, most of the customers purchasing multiple times had purchased from multiple selling channels.
Marketers face this challenge all of the time, don’t they? At what point is an old order no longer relevant?
I like to run a quick model, one that forecasts spend in, say, 2009, as a function of spend during 2008, 2007, 2006 and so on. In fact, back as far as the data mart allows me to go. I create variables for each year, summing demand during each year. I then regress historical years against 2009 spend.
The coefficients represent weights, telling me how relevant each dollar is from the prior year's. For instance, a model might look like this:
2009 spend = $50.00 + $0.30* (2008 spend) + $0.15* (2007 spend) + $0.08* (2006 spend) + $0.04* (2005 spend).
The model tells me that 2007 demand is worth half as much as 2008 demand ($0.15 / $0.30 = 0.50).
At this point, I apply the coefficients to dollars spent in different channels. If a customer spent $100 in 2005 over the telephone using a catalog and $50 in 2009 via paid search on your Web site, I can weight each purchase:
catalog weight = $100 x 0.04 = $4.00; and
online weight = $50 x 0.30 = $15.00.
Based on the equations, the customer is 79 percent likely to buy online ($15 / $19) and 21 percent likely to buy via catalog ($4 / $19) — somewhat multichannel, but essentially skewed to the Web.