RFM Turned Upside Down

For example, if the customers mailed have average sales of $3.30 and those in the control group who aren’t mailed have average sales of $1.80, the mail/do-not-mail decision for the group should be based on a response rate of $1.50.

It’s also critical to segment buyers in new ways. While many catalogers have started to segment by method of ordering — Internet vs. phone — that, too, can be misleading. Take it a step further: Divide those who’ve placed their orders online into customers originating on the Internet (resulting from nonbrand paid or organic searches) and those who’ve placed their first order after receiving catalogs. This necessitates adding matchback results and Web tracking information to customer databases. These two groups may have very different behavior in response to subsequent catalog mailings.