SOME FOLKS can’t leave well enough alone. When Nancy Miller, marketing analyst for Parkersburg, WV-based Woodcraft Supply, first started using a neural-network-based segmenting software, she reduced her mailing quantity and still increased revenue. But she saw room for improvement by giving the model more data-and realized a 27% boost in total sales.
Miller had been using Advanced Software Application’s ModelMax since 1994, a segmentation program that allowed her to mail 13% fewer catalogs while realizing a 45% hike in net profits.
Until she started planning for her May 1997 mailing, Miller had ranked her data according to performance models using customer transaction data from one campaign, taking into consideration more than a dozen variables including month of purchase, total lifetime value of the recipient and SCF code, as well as several custom-generated seasonal variables.
But even that did not eliminate all of the guesswork: Occasionally the modeling would require that she make value judgments when rounding out the field to meet a mailing quota. Was a customer who made a purchase in the last two years, but spent less than an individual who bought more items within the last three years, more or less valuable?
As a test, Miller modeled and scored three years’ worth of purchase data. When she re-ranked the file, her projected response rate from the top segment was 114% higher than the previously seen results. But modeling is one thing, actual results are another: Ten months after the mailing, the response rate was 170% higher than it had ever been.
Part of the reason the response rates increased so dramatically was stricter criteria for the top segments. The top level of names segmented by ModelMax was 41% smaller than the group that had been mailed to earlier.
But the sales per copy to these most valuable customers skyrocketed by 152%, and the total sales amount, despite the substantial drop in quantity, rose 49% above the year-earlier benchmark.
Overall, the May ’97 mailing resulted in a 27% lift in sales and a 2.38% growth in sales per copy. Not only was Woodcraft Supply reaching stronger, targeted candidates, but Miller had mailed to almost 25% more house names than she had before-a move that would normally dilute response rates.
Miller won’t, however, blindly rule out mailing catalogs to the bottom segments. If she needs a few extra names to round out a mailing, she will trawl the lower depths for newer purchasers, even if they haven’t spent enough to qualify as choice candidates.
“If you don’t mail to them again, you don’t give them a second chance to buy. If you never mail to your bottom rows, you run the risk of ‘wearing out’ your best people,” she says.
Network
