Testing Hidden Markets (1,296 words)
By Malcom Decker
When asked about what color choices would be available in his Model T, Henry Ford said, "You can have any color you want so long as it's black."
Is there such a thing in the United States these days as a single market for any nationally-distributed product or service?
Yet, many mailers treat their markets—that is, the universe of all the lists they think they can mail profitably—as if every person on those lists shared identical characteristics, wants and needs, by mailing the same package to all of them.
Even Reader's Digest, one of the most sophisticated mailers in history, with a peak circulation of more than 17 million, mailed only one control until sweepstakes came along. After that, it mailed two: The second package went to that segment of its universe that never responded to sweepstakes.
Boardroom Reports has tested and had limited success in developing different controls for different segments of its market, according to Boardroom executive vice president Brian Kurtz. He cites additional costs and complexity of execution but sees the main problem for most mailers as a reliable method of segmenting the market.
Dave Shepard, president of David Shepard Associates, a database marketing organization, says that determining whether or not a market includes two or more segments is a manageable project, but adds that it is not likely that every market has two or more significant segments. He emphasizes "significant."
Meanwhile, The Wall Street Journal continues its long-established practice of mailing several different controls, each aimed at a different segment of its multi-faceted corporate/financial/investment market, and testing each of these controls in each segment regularly.
Are you still offering a black Model T? That is, do you have just one control package?
You may have conducted research, followed by live tests, that tells you that your market is bell-shaped, and the data prove that it doesn't make sense to mail more than one control. OK.
But if you haven't done the research and the testing, you don't know whether your market is clustered around one peak or whether it has two or more significant segments that would each support different creative efforts and yield more than one package aimed at dead center.
The same package for everyone implicitly makes the assumption that the positioning, copy, graphics, format and premium offered in one package are optimum for every person you mail to: man/woman, young/old, high income/low income, urban/suburban/rural, heavy users/light users, amateur/expert, conservative/liberal, etc.
Is there any likelihood that such an assumption is accurate? Or is it more likely that at least one significant segment of your universe would respond more strongly to a package with a different caption or graphic on the carrier envelope? A letter with a different lead? Or different positioning? A brochure of a new design? Or the same package in a different size or format?
Testing to Create Different Controls
I'm not talking about A/B split testing to find the stronger of two packages (which is another aspect of segmentation), but rather of finding a basic differentiation between two major segments of the same market and creating a distinctive control package for each.
There are other examples of mailers who target distinct groups within their markets and the one that offers the clearest data is the Institute of Children's Literature, a home-study school. It initially focused its mailings on schoolteachers, the definable category from which it drew the largest percentage of students for its writing course, and results were marginal. Some years later, A/B split tests of premiums revealed two less distinct but larger groups of writers: aspirants, or "wannabees," and those who seemed to identify themselves as writers, whether or not they were published.
The Index premium actually depressed results among "wannabees" by 19 percent while it boosted results among "writers" by 28 percent for a swing of 45 points at a 95-percent confidence level. Copy was written to reflect these differences, an appropriate premium "How and Where to Sell Your Writing" was developed for the "wannabees" and response improved to 36 percent over control while it lost among "writers" by 10 percent; they apparently expected more or different information than they received.
The duality of the market has been confirmed in many other tests. One of these was a simple "YES" peel-off label on the order card. Both "wannabees" and "writers" responded positively, but 44 percent more "writers" enrolled.
A test of carrier envelopes (one with a cartoon vs. one without) showed that "writers" were slightly negative about the cartoon image—perhaps more serious about writing—while "wannabees" preferred it by 108 percent.
In two different tests of the "Growing Up" concept (a simulated children's book with the addressee's name as the "author"), "writers" responded by enrolling at a rate of 24 percent and 37 percent higher than "wannabees" who apparently were less able to project themselves as authors of children's books.
A similar, but parallel, development took shape at The Children's Writer, the institute's newsletter. Enrolled students, inquirers and outside lists were each mailed packages with significant copy differences when it became clear that they would not all climb into the same black Model T at the same rate.
An old marketing rule of thumb states that 20 percent of customers account for 80 percent of a product's sales and even more of its profits. Is it possible that your "20 percent" and your "80 percent" would each give you more sales and profits if each segment received a different package tailored to its distinct characteristics?
How Rented Lists Can Help Your Research
You may well ask, "How do I determine whether or not there are two or more distinct segments within my market that would respond better if each were tested separately?"
There are a variety of research techniques available to help you answer the question but you may already be aware of clustering of results from your rented mailing lists. If not, put them in rank order according to their average performances over a series of mailings. Look for clusters of similarity between lists—according to how the names got on the list, the product or service offered, price, terms and type of offer used to generate the inquiry or sale, categories of media used, and any other factor that suggests that the people on one list came from "a different place" and are demographically or psychographically different from those on another list. Then select the strongest of all the differences you find, the ones that suggest different positioning, or a significant copy or design variation. Focus your analysis on the bottom half of your lists because your control is presumably doing well among your top lists.
The next step is to develop two test packages to run against the control in a perfect A/B/C split. While the product and price remain constant, the creative positioning, emphasis on benefits, style of graphics, physical format, and premium offer should all vary.
Whichever way you approach market segmentation, the prospects are reasonably good that you'll at least end up with one package that beats your current control and, with good fortune, you'll develop a control package for each segment of your universe that, in total, yields more than your current single control.
As a bonus, successful segmented marketing also holds the promise of expanding the universe of lists to which you can mail profitably. Try it. In the late Dick Benson's parlance, "It's test-worthy."
Malcolm Decker is president of Malcolm Decker Associates. He can be reached at (203) 422-0518, or by e-mail at email@example.com.
Create a Different Control for a Different Audience: "Writers" responded better to this test of the "Growing Up" concept than "wannabees" who apparently were less able to project themselves as authors of children's books.