Promotion History Databases Keep you on Track
Ever wonder where you would be without your promotion history database, the record of all the direct mail tests and campaign rollouts you've conducted over time? Consider yourself lucky not to be in the shoes of the marketer in the following story.
A few years ago, a publishing conglomerate acquired a health-oriented magazine. The problem: It didn't acquire the promotion history database of the magazine. So while the prior management team had just put into play a new mail effort that had beaten an extremely successful longtime control, the new owner's circulation department knew little of the background on how the mailing program had gotten from old control to new control.
To establish the only benchmark it could, the new circulation team mailed the two controls head to head, and the prior control came out on top. Maybe in back-testing, the same result would have transpired. But the bigger tragedy here is the inability of the magazine's new owner to leverage its direct mail program based on the product's history of wins and losses.
This story leads to a broader question: How do you make your decisions about what to mail and to whom? If you're not tracking the right data, collecting it properly and conducting meaningful analysis, your campaign results could be a lot less than is possible.
What to Track
One size does not fit all when it comes to deciding which details about a mail program are important to follow.
Certainly, you have to take into account your ability to record, maintain and read the data. Also, as Josh Moritz, managing partner of the Westport, CT, direct marketing consultancy Customer Growth Partnership, points out, if you "track too many subcategories, you won't get a big enough sample to get a statistically good read."
Beyond this, the particular elements tracked will depend on each marketer's needs, says Cynthia Baughan Wheaton, a principal of The Wheaton Group, a direct marketing consultancy in Chapel Hill, NC.
Elements most marketers tend to track on mailings to their housefile, says Wheaton, are the offer, list segment, drop or in-home date, products offered, creative approach, shipping and handling, and the in-the-mail cost per thousand (or per piece mailed).
If the mailing is to an outside list, Wheaton recommends adding the list name, segment description, broker and net list rental cost per thousand (or per piece mailed) to the roster of elements tracked on housefile mailings. "The net list rental cost," she explains, "is the total list rental cost divided by the actual circulation mailed."
Because personnel turnover can be frequent, Wheaton finds that tracking lists by broker gives you a reliable source of knowledge about the lists you've used. Brokers keep detailed records on the performance of lists they ecommend, so their help is crucial in the development of mailing projects that extend beyond the basic RFM (recency, frequency, monetary value) scope.
But you might want to go a little further, especially if you use types of mail other than catalogs. In this instance, says Wheaton, you might find it valuable to track the copywriters and designers used, particularly formats/sizes mailed and the status of return postage.
For example: Reader's Digest magazine notes which packages are created by which copywriter/designer team, for both in-house staff and freelancers, says Lou Sassano, the publication's marketing director. This allows Sassano to build a history of which teams do well with each type of mail program or format used by Reader's Digest.
Sassano reports that the publisher uses a digital asset management system called media:bank, which stores creative from all of the company's direct mail campaigns in electronic format.
"We still keep paper samples," Sassano adds, for staff who want to see how the insertion of elements looks, etc.
The big advantage of a digital system for mail samples, Sassano notes, is speednot only the quick access to campaign creative it provides, but the ease with which art directors can retrieve campaigns to update elements for testing or rollout.
While you may not be able to install a digital asset management solution, Wheaton advises marketers to develop some written record that describes each direct mail effort mailed, including "variations between the control and tests."
A final area to keep in mind when collecting information about your direct mail campaigns is media diversity. To get a full understanding of the environment into which your direct mail campaign was received, you need to track the use of other media in play at the time, explains Chris Dickey, director of customer insight management of Digital Evergreen, an interactive communications firm in Kansas City, MO.
"What other media channel was being utilized at the time the campaign dropped? All things being equal, two exact campaigns can have phenomenally different results, if one was supported by radio, spot TV, online advertising, etc.especially if they were well integrated," Dickey says.
While you won't be able to track general media campaigns back to a customer ID code, he says, you can tie e-mail and telemarketing efforts to individual customers, and thus read at least these results from a media integration perspective.
Share the Codes
The key, literally, to tracking campaign data is keycodes. These codes can be any combination of letters and numbers, and tend to be particular to the individual marketer. (For more on keycode systems, see "Cracking the Code," at right.)
Regardless of how many keycodes are usedcustomer, campaign, etc.these tracking codes are useless without proper implementation.
"The bane of existence of any direct marketer is that within an organization, people who don't work in direct marketing pick up codes for all sorts of stuff because they kind of know they have to, but not how to," says Moritz.
He advises marketers to centralize the coding function with the employee responsible for the mailing matrix, which should outline all packages and tests by offer, list, creative and format.
The mailing matrix provides a guide for the data service bureau to assign the correct code to each list segment; this information is shared with the vendors in charge of fulfillment, such as a forms processor or a call center, so that codes can be matched up and results reports generated easily for analysis by marketing, says Moritz.
The IT staff also needs to understand the coding system, because it likely will be the department responsible for linking the response results back to the promotion history database. In many instances, the IT department also generates the reports marketers use for analyzing campaign performance.
In the retail world, Dickey points out, IT sometimes needs to download the coding matrix into the point-of-sale system. Multichannel marketers can use this customer information to tailor marketing efforts to customers at checkout. For example, the company's best customers could be acknowledged with a special discount coupon on future purchases.
Dickey adds that a consistent customer ID code across all corporate databases provides the most reliable tracking. A consistent customer ID code also allows you to take analysis beyond response. "For example, we might want to look at average order as a further metric to understand not only what drove response, but what drove higher average orders," Dickey says.
With a consistent, organized coding system, marketers easily can link the promotion history database and
response history database together. Moritz and Wheaton find that these systems don't need to be integrated
into one databasewhich is a time-consuming and costly undertakingto access your results in a timely manner.
On the other hand, an integrated database has its advantages. At Reader's Digest, the promotion and response data for the magazine's gift subscription and renewal programs was integrated into one database last year, says Sassano. With access to all data in a consistent format, Sassano easily can compare campaign results over the years, based on a variety of variables. For example, he says, he can look at the first effort in a current renewal series and compare its
performance to all other efforts in the first slot of every renewal series. By analyzing performance this way, he
can "get a feel for [which mail piece] performs best in which slots."
The Analysis Stage
The fruits of your labor in setting up a historical database are the reports that contain campaign results. While the other components of response tracking involve some complexity, it's the analysis stage that holds a higher degree of challenge.
Wheaton notes: "Analysis is often complex, and each situation can have unique issues. It is difficult, if not impossible, to do it correctly in a systematized manner. The one thing that is worse than having no information for decision making is to have wrong information. And, that is very easy to do with complex queries that go beyond simple counts."
An example of a simple count that is necessary to plan direct mail campaigns is response per day/week to establish a response curve. With a reliable response curve, says Moritz, you can determine in-home dates for your campaigns, which allows you to plan mail dates for future campaigns and layer other media for extra lift.
"For analytical purposes," Moritz says, "give your promotion about three weeks in-home to mature. Usually, the last responses from a Standard Rate mailing come in about 13 weeks after the initial drop."
Wheaton adds that response curves also help you project what your total dollars and response will be for a given campaign before the orders stop coming in.
Sassano shares that Reader's Digest breaks up reporting by test, looking individually at format, copy, offer and more. Response is indexed against the control for each test variable. Campaigns are indexed by performance on gross response, pay-up, cash with order, subscription term taken (e.g., one year versus two years), and a few other variables.
To improve campaign performance, Sassano says his marketing staff tracks these rates over time, looking for the right combination of mailing variables to improve cost effectiveness.
Obviously, the analysis that matters depends on the needs of each marketer.
"[For catalogers or retailers] this could be as basic as a square-inch analysis, where you look at performance by size of an ad," says Dickey. "You can also run an analysis on average order by product featured or product bundles to determine what optimized higher salesand hopefully higher margins."
Your Future Depends on Your Past
The goal in creating a robust promotion history database is to supply yourself with an accurate record of what you've mailed so you can best determine which direction to take for future mail campaigns.
While your analysis may not change your contact strategy, says Dickey, it might help you target your creative to sub-cells within your mail plan for higher response and average orders.
"A promotion history file lets you respond to people who say, 'we tested that in 1977 and it didn't work,'" adds Wheaton. It allows you to review the circumstances surrounding a previous campaign and assess its results. You may find you have good reason to believe the test will work now based on a different environment.
"Enough people do not document what has happened or the conclusions they've drawn from their analysis," Wheaton says.
Marketers should have a paper trail from the results to the marketing decisions made based on their conclusions, she explains. Then, when a campaign doesn't perform to expectation, you can review what might have gone wrong. It takes courage to check yourself this way, she adds.
In the end, the promotion history database is just the tool to build what you need to perform analysis. It's what you do with that database that makes the difference.
Cracking the Code
Most coding systems are alphanumeric. Josh Moritz, managing partner of Customer Growth Partnership, provides the following example of a simple coding system:
* The AAA stands for package A with offer A, mailing to list A.
* The 012003 stands for the drop date (mail date), or Jan. 20, 2003.
To track different offers, lists and packages, assign them different letters. For example, ABA would represent mailing package A with offer B to list A.
Another example, with slightly more tracking ability, is the following coding system from Cynthia Baughan Wheaton, a principal of The Wheaton Group:
* No. 1 slot: Alpha identifying major catalog/division/program.
Examples: V = Valentine's Day, B = Baby Catalog.
* No. 2 slot: Last digit of year or other seasonal/yearly identification.
* No. 3 slot: Drop number, which ties into a specific drop or in-home date.
* No. 4 to No. 7 slots: List segment number. House list segments may begin with "1," while outside lists begin with "5," and multis (lists which feature prospects that match in-house and outside lists) with "7." It's particularly helpful if the list segment numbers are consistent across mailings, so you can pull the entire response history for each list quickly.
For instance: 12-month, $100+ names from Rental List X are always identified as list 5124.
* No. 8 slot: Alpha description of cover/envelope/offer. This aids the lettershop/printer in quality control. For example, all keycodes on the "Offer B" ZIP string should end with "B."