Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
386423 | Expert Systems with Applications | 2010 | 6 Pages |
Organizations that use data mining “scoring” models to estimate the expected revenue resulting from sending a customer some marketing contact such as a catalog, solicitation for a donation, or other offers are underestimating the value of the contact and under-investing in their customers. When a customer responds to the contact the organization will receive the revenue due to the order, which is estimated by the scoring model, but it also benefits because the customer has become a better customer and will thus be more responsive to future contacts. The paper proposes a correction that accounts for the incremental change in customer status and illustrates the correction on two data sets from different industries. In both cases we find that the magnitude of the correlation approximately equals that of the short-term return from the contact. The correction is highly correlated with short-term returns and so the correction has little effect on the rank ordering of the customers. Topics for future research are identified.