For many franchises, direct mail is a critical tool for acquiring new customers. So, what is the most important ingredient of a successful direct mail program? It is list targeting. A direct mail industry truism is that having the right names on the mailing list is far more important than either the creative copy or the offer.

How do you target your direct mail? If you are like many in the franchise world, you may know what you want but sometimes need to make painful compromises that yield mediocre direct mail results. Here are some of the deadly targeting compromises and two case studies of franchises that found a way to avoid these pitfalls.

Compromise No. 1: Limitations of Selection Criteria
If you're like many marketers, you fall back on your best judgment and use "selection criteria" to pick names for your mailing list. "Selection criteria" is a euphemism for crude rules behind list selection.

Selection criteria rules are probably generally correct but are almost always very simplistic. The accuracy of these manual selections is highly dependent upon both the expertise of the individual concocting the rules and the accuracy of the underlying data.

Some available selection criteria such as data about a person's age tends to be fairly accurate, while other data such as "estimated income" is usually just that -- estimated. For example, if you ask for names of people making over $125,000 a year, as many as half your list may not actually earn that much money.

Even if you can create complex rules and can pick accurate data fields from the list vendors offer, each additional selection criteria often comes with a substantial increase in the cost of the list. But don't give up.

Compromise No. 2: "One Size Fits All" Targeting
This occurs when you have a mailing serving multiple markets, but you have only one targeting methodology

Your direct mail vendor wants to know what your name selection criteria are. The vendor provides a long list of demographic selection criteria: age of head of household, estimated income, length of residence, family composition, ethnic group, religion. You know your market so it should be straightforward, right?

Let's say that your franchise provides a service to upscale consumers and your mailing will cover 12 franchise locations in four different cities. It's time for compromise No. 2.

You decide that an estimated household income of $100,000 should be good selection criteria. But $100,000 in income in Boston, MA, is very different than $100,000 in Billings, MT. You know that these markets have completely different costs of living, and what you would really like is a disposable income selection criterion.

No, that is not offered. You decide to be a bit more selective and select names over $125,000 to better accommodate the Boston and New York markets, knowing that you are being too selective for your other markets. In this compromise, you have tried to make one set of selection criteria fit multiple markets. One size never does fit all, at least not well, and your mailing results will suffer. But they don't have to.

Compromise No. 3: Geographical Selection
Now your list vendor asks for a geographic selection. Clearly, just selecting a state is not targeted enough, so many franchises use ZIP codes. This in itself is a compromise. ZIP codes are fairly large population subdivisions -- about 10,000 people. Using finer subdivisions would produce more accurate targeting, but that data is probably not on your customer file.

You end up with a list of ZIP codes associated with each of your markets. However, just as one size does not fit all demographically, not all ZIP codes are equally productive in yielding new customers. What you would really like is a more sophisticated breakdown for each franchise location. For example, you have less chance of acquiring a customer as that individual's travel time to the franchise increases. Your direct mail list should reflect that fact.

If you've read up on modern marketing technologies, you may know that names selected via sophisticated predictive analytics can yield increased response rates by 30%-50%.

Recently, the field of direct marketing analytics has been rocked by the introduction of a technology that reduces the time needed to build a predictive targeting model by 99%. "Machine-learning" allows computers to do analytic tasks that, until recently, required weeks of professional time. This reduction of the time results in a reduction in labor and, ultimately, a dramatic reduction in the cost of predictive models.

Now models are so plentiful that they are often free with the purchase of a list for use in direct marketing. Compromise No. 1 (Limitations of Selection Criteria) is now a non-issue.

Predictive models can be built for each franchise location. These custom-tailored models pick up subtleties in local demographics, cost-of-living differentials and ethnic mixes. The time of shoehorning "one size to fit all" is dead when superior customization proliferates and is affordable.

Over 100 list brokers, marketing agencies and printers have adopted this technology for list selection since its introduction about two years ago. Many of them serve the franchise direct mail production market.

A cleaning service franchise had been directing its franchisees to select lists that target households based upon income and geography. After hearing of a new, supposedly better targeting methodology, a test was devised to measure the impact of the new approach.

Twenty thousand customer acquisition mail pieces were sent using the traditional selection approach to targeting. At the same time, 20,000 pieces were sent using sophisticated predictive modeling technology.

The modeling technology found that income was in fact an important predictor of response to cleaning service offers. However it also found a complex mixture of home value, length of residence, presence of young boys in the household and presence of pets to be predictive.

When the dust settled and the new customers were counted, the machine learning approach yielded over 56% more new customers and over 60% more revenue, since the predictive model brought in customers who spent more as well.

The superior targeting leveraged the investment in the mailing to produce a 105% increase in the promotion's ROI. Word spread, and now over 90 of this cleaning company's franchisees use custom predictive models to select names for acquisition.

Pizza sales are often driven by direct mail couponing. Getting the coupons to the right households at the right time can make all the difference. One franchisee of a national pizza chain operates 17 locations in and around a mid-sized Midwest city. As in many cities, different neighborhoods of that city have significantly different demographic, ethnic, and economic profiles.

Using just one selection criteria would have guaranteed a mediocre list. Instead the franchisee had a different predictive model built for each of the 17 locations. The models cost nothing and the list cost was on par with what the franchisee had been paying for inferior lists.

The results were impressive. The unique aspect of this campaign was that in little time, and for no additional cost, immediate results were attained by simply thinking outside the traditional direct mail "box" and applying a new technology.

Once again, news travels fastest in the franchise world when it travels by word of mouth. Forty-seven additional locations of that chain immediately converted to custom predictive models to shape their list selection.

Technological changes are reaching down to help small direct mailers such as franchisees target their mailings with the same precision that the county's largest mailers enjoy. Each location can now apply laser-like precision to make direct mail work better and even work where it has never worked before.