34 Beware of 50/50 splits. Publications sometimes offer an "imperfect" A/B split or a "cluster" split. Instead of having the ideal mix of every other copy, the issue is batched in groups of 50 to 500. Media that offer a 50/50 split guarantee only that each ad will be distributed to 50 percent of the circulation. Be sure there's no geographic bias to the distribution.
35 If you want to test one region vs. another, some media will accommodate geographic splits at no charge, so long as the territorial splits tie in with their normal distribution boundaries.
36 Many media aren't set up to do splits. Others will do splits only on full pages. Split charges can vary from zero to $3,000, so it pays to bargain. Conduct A/B splits in your most-profitable media. TV Guide offers a split-run nationally or within each of eight regions. Delta Sky will do a four-way (A/B/C/D) split, which provides an easy way to test travel and business offers.
37 Be sure every ad is coded separately so you can track results.
38 In measuring the results of a split run, look for a clear winner. If results are separated by only a small margin, you gain nothing. Unless the test outpulls a control by 20 percent or more, it's not a clear winner over a control ad that's had prior exposure in the publication.
39 Track ad results as long as possible but no less than three months. Some ads pull orders years after the issue date. This is particularly true of magazines that tend to hang around barbershops or doctors' offices.
40 Don't be discouraged if your initial results are not blockbuster. I work with an advertiser whose test ad ran in three magazines. Two of the three failed, and the third was only marginal. With what we learned, we still built a long-term successful ad campaign that helped double the client's subscriber base.