Traditionally in the financial services area, age and income always have been found to be the most predictive factors in customer acquisition. Some marketers, however, have found that selection programs using these two factors may miss many important aspects of financial services customers.
A major brokerage firm had been using age and income based list selections to target their direct mail to prospective mutual fund buyers. Analysis showed that even though the “over 30 years old, $100,000-plus income” segment was more likely than the general population to respond, more than 90 percent of their buyers were not coming from this segment. In other words, targeting this group would provide a lift over random selection, but would also exclude the vast majority of likely responders. To understand what was going on, they built a model using financial and demographic data. The model showed them how to select prospects so that they could effectively reach more than 75 percent of likely buyers instead of 10 percent – a huge improvement over their age and income selection strategy. The model illustrated that buyers of investment products today can no longer be effectively targeted just by age and income. With so many investors in the market, net worth, risk preferences, and attitudes are becoming important in defining customer segments.
In another case, a major mutual fund company marketed a complete line of different funds by direct mail and telemarketing. They were using just age and income to select prospects from compiled lists. They wondered if perhaps they were missing the mark by not selecting different lists based upon their prospects’ expected tolerances for risk and their fund preferences. A consultant took a sample of their investor files across several of their different funds, overlaid demographic information, and developed segment profiles. It was clear that they were indeed selling their different funds to markedly different segments; in fact, in several cases, the segments buying conservative funds were mutually exclusive from those investing in aggressive funds. Their marketing plan that had tried to target both types of funds with the same lists could not possibly reach both types of buyers, and would actually do a poor job of reaching either group. Based upon the model results, they revamped their entire strategy to target their funds to specific segments.
Getting your hands on the data
What data, other than age and income, is available to marketers? A look at a compiled database like AmeriLINK shows what can be appended to a customer or prospect file.
- House value
- Equity in home
- Homeowner vs. Renter
- Length of residence
- Dwelling type
- Auto Loan Indicator
- Children Present in home
- Number of adults in the home
- On Line Household
- Refinance propensity
- Finance Loan Indicator
- Credit activity
While this data is potentially available, there is a problem. Most marketers rely on rented lists for monthly acquisition mailings. Such lists can often be rented by age and income, but seldom (if ever) come with the type of data listed above. It is not economically possible to append this data to a list that is rented for a single use. That is why many marketers are asking their service bureaus to create prospect databases for them.
A prospect database differs from regular acquisition rented lists in that the names are usually rented for a full year. The list owner is paid each time one of his names is used in a mailing. To assure honesty, the prospect database is usually maintained at an independent service bureau. Once the prospect database is set up, put in a single format, and de-duplicated, all the records are usually appended with the type of data listed above. With a prospect database, the promotion history (who was mailed and who responded) is stored in the database as well. With demographic data and promotion history, those with a prospect database can do advanced analytics that is usually able to predict with some accuracy which consumers will respond best to each of the company’s products.
Results of using a prospect database
A major insurance company had been getting one qualified lead in twenty cold calls from lists selected by age and income. With the help of a service bureau, they built a prospect database. Using this database, they were able to pre-qualify leads using customer profiles developed from transaction history and demographic enhancement data. By matching their prospect database against the profiles, the company was able to determine the customers most likely to buy a life, auto, or homeowner product. Using the new system, they were able to get one or two qualified leads from ten cold calls — more than double their previous rate. The prospect database plus behind-the-scenes work eliminated $3 to $4 million dollars per year in data costs.
Using a prospect database with analytics
A continuity club faced a difficult problem: direct mail solicitations based solely on age and income were attracting the wrong kind of customer. Direct mail solicitations offered a number of “free” products for shipping and handling costs plus the promise to purchase additional products at “full price”.
The solicitations attracted many customers who had little loyalty and frequently refused to pay even the shipping and handling costs.
- A high percentage of individuals who responded to these solicitations would obtain the “free” products and never pay.
- Those who paid for their initial selections often failed to meet their full commitment, or terminated shortly after meeting their commitment.
- The Web was no solution: online signups yielded payment rates even lower than direct mail.
- This led to a default rate that was unacceptable. There was a high degree of customer defection.
To solve these problems, the continuity club had created a number of in-house response models that inadvertently targeted likely responders who were unlikely to pay and remain loyal. The in-house models that focused on likely payers produced mailings with response rates that were too low to be profitable. The challenge was to find prospects with balanced probabilities of responding and paying.
The continuity club asked KnowledgeBase Marketing to create a two model chain based on appended data from AmeriLINK. One model focused on response and one focused on payment after response. The analysis showed that creative use of all available information provided good predictive power. Some of the most important variables were:
- Occupation concentration.
- Zip Code or Zip + 4
- Percentage defaulted.
- Number of applicants divided by number of club members.
As a result of using the models to direct the acquisition program:
- No-payment rates were reduced by 20 percent when mailing to the top three deciles.
- Overall, the mailings shifted from negative returns to a cumulative profit of 5 cents per piece when mailing the top three deciles.
- In-market tests demonstrated the strength of the approach, and replaced the previous models as the “gold standard”.