Update on Multichannel Matchbacks

by Stephen Lett

Match-backs are not perfect. There are issues with regard to date ranges, match-back logic used, timing of when the match-back is done, multiple catalog titles and more. The process of matching back is not all that complicated. It is simply a matter of matching the mail files retained by your service bureau with your order file.  At Lett Direct, we have looked into doing match-backs at the Stock Keeping Unit (SKU) level. This seemed to be a logical solution for catalog companies who circulate more than one title to the same people with overlapping mail dates. This results in an extra step in order to identify which book the customer purchased from. The problem with this is that there is SKU duplication in both books therefore this “extra step” to identify the matches at the SKU level does not always help. Another issue with match-backs has to do with allocating the Internet portion fairly. For example, if someone purchases a product from the Internet, they become an Internet buyer. However, they will receive a catalog in the mail. They may make a second purchase from the Internet. When we do a match-back, this person will always be considered a catalog buyer after their initial purchase because they were mailed a catalog with a housefile key code identifier. While the catalog remains the largest driver of traffic to the Web, this does make it difficult from the Internet Marketing Manager to verify that they are growing their business and remaining on budget since, in this case, the credit is going to the catalog. This factor will impact the repurchase stats for the Internet. Date ranges are important to accurate match-back reporting.

How many months should you match-back, for example. This varies by how often you mail and how much business is driven by the Web. The problem always occurs in the last catalog mailed during holiday. The match-back will favor this mailing. For example, let’s assume a cataloger make five different holiday mailings. A large portion of the sales from these mailings will occur the last three or four weeks prior to the Christmas holiday. In this example, the last mailing will receive a disproportional amount of the sales from the match-back. I do not favor making any assumption for the allocation of this factor because it means introducing yet another unknown. Any allocation at this point is really not necessary because what really matters is the performance of the housefile buyers by segment for all five mailings combined. Any analysis from names that are mailed more than once should be evaluated in total across all mailings and not necessarily by individual drop or mailing.

Please note how high the RPC (Revenue per Catalog) mailed was holiday 5. Again, this is due to the fact that the last mailing in the season is favored in the match-back logic. When we average the RPC for all five drops in our example, the overall performance by segment can be evaluated. Source code tracking is as important as ever. The fact that match-backs are done on the back-end of a mailing does not mean we should not stress the importance of capturing source codes on the front-end. Continue to capture those source codes on the front-end and use the match-back to help allocate orders through the Web and miscellaneous non-traceable results. Recognize match-back methodology is not perfect. Certain assumptions must be made. Tracing orders and sales to a specific source or key code is critical to arriving at proper conclusions and making the best decision regarding future circulation plans.