Articles
August,
2002
Enhancing Your Housefile
Your house file is your business's most valuable asset. Yet most mailers do not maximize the profit potential of their house file, either over or under mailing portions. In the June issue of CatalogAge, I discussed using RFM - recency, frequency, and monetary value - as an effective tool to identify segments to mail. Assuming that you have applied this tool effectively, many segments were identified that were borderline or clearly did not deserve all or even one mailing of your catalog. However, within these segments there remain significant quantities of responsive names that deserve not only one catalog mailing but also every catalog you can send them. To not mail these names is to leave gold on the table.
Enhancing
Names in the Merge
Interaction with Outside Lists
The interaction of your house file with other lists in the merge purge is a highly effective technique to identify responsive names. Yet most mailers miss this opportunity simply letting valuable names they have paid for or exchanged disappear. I recommend that you break every segment of your house file into two segments - those names that remain unique to you and those that duplicate a name on an outside list. This will allow you to fully evaluate the impact of this interaction for all RFM segmentation. The graph above shows an example of the impact of this interaction on response for a typical home décor catalog. Buyer files less than 24 months old that create a multi-buyer with an outside list performed 1.94 times better than those that were unique; files more than 24 months old that create a multi-buyer with an outside list performed 1.93 times better. These older buyers not only have affinity with the example catalog but have also purchased from a competitor in the recent past. They have again come alive! If there were segments of older names that were unprofitable to mail, a portion has been isolated that can be mailed with potential profits. Mail them and evaluate the results.
DMA Preference File
Hopefully, you are using the Direct Marketing Association's Preference File as a suppression file in the merge purge against outside lists that you have rented. The Preference File is a list of individuals who have registered with the DMA their desire to not receive unsolicited mail. It is critical that you apply this suppression to avoid wasting your catalogs by mailing those who do not want to receive them.
Most mailers do not apply the Preference File as a suppression against your own buyer or requestor file - you have the absolute right to mail your own house files unless you have received a request from a specific individual to be removed from your list. However, we recommend that you isolate and test mail those house files that are on the Preference File.
The results may surprise you. You may find that you should suppress the mailing of certain house segments on the file or, conversely, continue to mail them. Table below illustrates the response for our typical home decor catalog mailing for house files that hit and did not hit the Preference File. House file buyers who were registered had a response rate that was 2.01 times higher than those who were not registered. Likewise, registered requestors had a response rate 1.76 times higher.
RESULTS
OF PANDER TESTS - CATALOG A
NCOA
One of the first steps in the pre-merge process is to apply the National Change of Address file to your house files as a portion of the list hygiene processing. Graph below shows the response rate of the house file for our home décor catalog that had a change of address versus those that did not. Buyers who had a change of address had a response rate 1.82 times higher than those who did not. Requestors with a change had a response rate 1.11 times higher. Our experience with different types of catalogs shows these results vary significantly (e.g. for apparel catalogs there is minimal difference).
Interestingly, we have also experimented successfully with certain clients mailing the old address. The logic of this success can follow from the simple demographics of the individual who moved into the customer's former residence to more tangible factors. For example, if you sell greenhouse, barn, pond, or home theatre accessories, there is a high probability that the new resident will have an immediate need for your product after they move in. You might send "dear new resident" a catalog with a well-written letter introducing your company, informing them that the previous owner found your catalog very useful in addressing features of the property.
RESULTS OF ADDRESS CHANGE BY NCOA
Post Merge Techniques
While the above techniques will identify significant numbers of names to mail that would otherwise go unmailed, there are still significant opportunities to both suppress or select names for mailing based on information from cooperative databases. This is true whether you are using RFM for mail decisions or more sophisticated modeling techniques. Even sophisticated house models used by larger companies can only model variables present within the house database. There is a point where your buyers' and requestors' behavior with other catalog companies becomes the most important variable in determining future behavior.
Abacus
Abacus, a division of DoubleClick Inc., is the premier catalog buyer cooperative database with purchase data on over 90 million households and 1,800 member companies. Abacus offers several products that can identify names on your house file to both suppress and / or mail.
Traditional Optimization. Assuming that you have put every name you own into the merge purge and applied the techniques discussed above, you should now send Abacus "batches" of names coming out of the merge that have not been identified for mailing based on RFM or the techniques discussed above. How many "batches" is partly dependent on the size of your house file since each "batch" must be minimally 70,000 to 100,000 names (smaller batches are possible but are less exact in reliable ranking; they can be used to identify non-matches to the database). Ideally, separate "batches" should be sent for your buyers and for your requestors. Abacus will identify the dregs in these "batches" (i.e. the bottom 10% - 35%) to suppress and not mail or the cream (i.e. the top 10% - 35%) to select or mail.
Abacus ranks each "batch" into twenty segments of equal quantity with a twenty-first segment being those that did not match their database. The average mail order activity for each segment will be indicated by RFM. Initially, it is good to track results for the mailing or suppression of different ranges of segments to determine how many segments can be included for suppression or selection. Abacus prices this product at 4.5 cents per name suppressed or selected (which ever is less) with a $1,000 minimum. Optimization of non-house files (prospects) can be included in meeting the minimum, thereby making this product suitable for fairly small house files.
Dynamic Optimization. Our tests have generally shown that Dynamic Optimization provides better results than traditional optimization. It initially requires a larger quantity of names, though, so it is less suitable for small house files. In traditional optimization, your file is compared to the activity of all Abacus members in one process or "round". In Dynamic Optimization, your names are compared to subsets of the members in succeeding "rounds", each subset having different characteristics. For example, in round one, your names may be ranked into the twenty segments that have a purchase in the last twelve months of members in your product category. Round two may have another set of criteria such as another specific product category purchase or a certain number of twelve-month purchases. The third round may then rank the remaining names not selected in the first two rounds by applying traditional optimization. Dynamic Optimization is priced at 4.5 cents per name but has a minimum fee of $2000 meaning that 44,444 should be selected or suppressed from all batches to justify the expense.
House File Reactivation Models. Better than Optimization are Abacus house file reactivation models. Unfortunately, this technique is only suitable for a minimal reactivation file size of 100,000 + names. As with optimization, we recommend that you initially test mailing several ranges of segments with different keycodes to track differential response. Only by such testing can you determine if the process works for your file and how low segment ranking will yield acceptable results. While the pricing of this model is again 4.5 cents per name selected of suppressed, whichever is less, the minimum charge of $1000 must be applied to the single model and target file. Therefore, for example, if you are not selecting or suppressing at least 22,222 from your buyer file you are not fully using what you are paying for and the process becomes less cost effective.
For those catalogs that have non-performing files large enough to apply reactivation models, we have found it useful to alternate the optimization process with modeling. We believe that either product is highly valuable and each may find different incremental names to mail. We are also aware of mailers who use the ranking of either optimization or modeling to tailor specific offers or discounts based on the need to lift response ( i.e., those segments that are less responsive receive a deeper discount offer to entice them to buy).
House File Scoring. Abacus also provides scoring of house files that can be used by larger catalog companies that either employ sophisticated in-house modeling techniques or cross-matrixes with RFM. Costs range from $5000 to $20,000 based on the number of records scored. The scores can be used for a period of three months.
Other Cooperative Databases
The closest competitor to Abacus is Experian's Z-24. With the transaction data of 600 member companies, it has a distinct disadvantage in competing with Abacus. We have found it useful in creating successful prospecting models, but not reactivation models. For those companies doing their merge purge at Experian, another optimization product "DT-ABC" can be applied during the merge itself based on modeling both Z-24 data and demographic data from the Experian database. Our tests have shown DT-ABC to be effective.
I-Behavior and Prefer Network are two relatively new databases with, as yet, fewer members and transactions. While they offer reactivation products, we have yet to test them with a client. However, until they reach a critical mass, it seems unlikely that their product can provide superior results. As they grow and mature, they may offer a viable product, since they model data at the specific product transaction level as opposed to company level.
Conclusion
While segmenting by RFM that we discussed in the June issue of CatalogAge is critical in making mailing decisions regarding entire segments of names, the approaches discussed above allow you either suppress potentially unresponsive names from segments you intend to mail or identify names to be mailed from segments that were, as a whole, undeserving of mailing. We have found these techniques so effective that we recommend placing one's entire house file, no matter how old or large, into most merges to identify the interaction with other lists in the merge. Following the merge, there is no penalty for sending Abacus large files for processing since Abacus charges for what is selected or suppressed by batch (whichever is less). Given the escalating in-mail cost of catalogs, such processing is almost always cost-effective.
I must add that we have found catalogs where some of these techniques have not worked. In a few of these cases, the catalog served such a niche that their customer's activity with other catalogs was perhaps irrelevant. As with most catalog marketing programs, do not do anything without testing. It is relatively easy to hold out test panels to determine if these techniques are valid for your catalog.
In the October issue of CatalogAge, I will discuss how the merge-purge can also be used to identify and isolate responsive and unresponsive prospect names. However, remember that it is almost always more profitable to maximize the mailing opportunities of your existing house files than to mail more prospect lists.
