N.C.O.A. and List Hygiene

This stands for National Change of Address, a constantly updated database generated by the U.S.P.S. and administered by selected service bureaus around the country. Wherever you get your lists, you want to be sure they have been run against the NCOA database for accuracy. Approximately 20% of Americans change their addresses every year, and any list deteriorates at about 50% per year.

You also want to be sure you run your own lists against the NCOA file to save mailing costs and improve deliverability. Other forms of "list hygiene" include:

• CASS (Coding Accuracy Support System) is a software program that will standardize your addresses, especially where lists have been enhanced. Make sure your data processor uses only CASS-certified software.
• Zip+4. To take full advantage of carrier sort discounts, you must append Zip+4 and carrier route data with CASS-certified software.

• PAVE (Presort Accuracy, Validation, and Evaluation).Your list must also be pre-sorted according to the Domestic Mail Manual. PAVE is another software program that will save you much time, aggravation, and money.

• Barcodes. The USPS requires that reply envelopes be bar coded if they are being carried by an envelope that is itself getting barcode discounts. Also try to use delivery point barcodes whenever possible. This is an extension of the Zip+4 barcode in which the last 10 bars represent the first two digits of the street address. Since the barcode is read instead of the printed address, the mail needn't meet OCR standards -- granting you some additional leeway in envelope design.

• ACS is a less costly alternative to ACR (Address Correction Requested). It provides electronic return of an undeliverable piece for 20 cents instead of the 50-cent rate for hard-copy address returns.

Costs and Testing

There are certain thresholds one needs to reach in order to have a valid test. Test one list and all you'll know when you're done is that was a good (or bad) list. Is that enough for further marketing decisions? I don't think so.

A rule of thumb for statistical reliability in list testing is to mail sufficient names to obtain at least 50 responses. If you anticipate a 1% response, that would indicate a 5,000 name mailing. If you anticipate fewer responses, increase the quantity accordingly.

Other "musts" include:

• Key code every list so you can track results
• Test no more than one variable at a time so you'll know where the spike or drop came from.
• Mail all tests at the same time.
• Test, in order of importance; lists, offer, price, creative, format, season.

You can expect that only about 30% of the lists you test will pay out, so the more lists you test, the more insurance you're buying, and the more you're learning about your market.

Also, if you go into the mail expecting that the revenue from your initial test will fund the next level of testing, you're likely to be disappointed. You should have sufficient capital to test mail, adjust lists, offer, or creative based on results; and re-mail and re-mail again before seeing any substantial revenue.

A software developer had put much time and money into a contact management/calendar program and after learning what it would take to get the product on the shelves in the retail channel, decided to go direct. I created a direct mail package, selected lists and generally supervised a mailing that pulled close to 4% response. I was delighted-- until the client told me, after the fact, that he had needed a 10% response to make enough to keep mailing. Ooops.

In fact, if a "break even" analysis shows you need a 10% response in order to succeed, it probably won't happen. Rethink the project.

A List Selection Model

In constructing a test mailing from list data sheets, I use a letter-number system for prioritizing lists. Letters -- A,B,C, etc. -- designate the list's closeness or compatibility with my customer profile. Since an exact fit would likely be available only from a direct competitor, and since they probably won't rent to you, your "A" designation will have to be approximate, but as close as you can get with the list selections you have.

1 | 2 | 3 | 4 | 5 | 6