Marketers use data and modeling in unexpected ways to improve mailing results and profitability. Models are designed to predict whether a name will be profitable to mail.
For example, a properly structured model can make a high-dollar offer with low response rates profitable to certain segments of a prospect pool or your existing customer base. Models also can help determine optimal price, offer treatment and billing terms. It all comes down to having a good idea and the right database resources so you can “let the model do the work.”
What is database marketing? Database marketing is direct marketing that uses data and analysis to improve business results. It is the practice of exploiting data patterns that rank order customers and/or prospects to create categories and predict consumer behavior.
For example, a regression model succeeds when it leverages data to rank order names and when it correctly predicts the outcomes a direct marketer needs. The model maximizes marketing results for the outcomes provided in the data sample, such as: gross response; net paid response; high-dollar spending; approved credit; good pay-up; and renewal and retention
Models are “single-minded” in that they maximize a single type of outcome. Should more than one behavior be required to achieve profitability – for example, high response and likely credit approval – then more than one model often is used to rank order names. Three database marketing applications follow:
· Renewals. A single data characteristic may be associated with renewal profitability, and multivariate regression may refine selections. In publishing, for example, segmentation can be achieved by grouping subscriber names by original source.
· Acquisition. Predicting acquisition results when little or no consumer transactions are available (mainly in prospecting and acquisition analytics) requires careful analysis and perhaps purchased data. In these cases, self-reported lifestyles and interests, compiled demographics and area-level aggregated and census data will predict paid direct mail response. Such data are then combined and weighted in regression models to rank order a prospect universe.
· Exchange. When mailers find it profitable, they can exchange lists and order from list cooperatives. In doing so, they supply company-critical transaction data, which are very powerful to all the partners in a data exchange.
Database marketing in action. The table on this page shows many of the applications for which marketers have used data and modeling to improve business results.
Randomly mailed “nths” of customers exhibiting these behaviors are supplied to modelers. The modelers, in turn, append the corresponding data from their marketing database. Modeling and scoring the prospect universe with appended data will increase the likelihood of these behaviors occurring and, in turn, lead to more profitable mailings.
Growing revenue without increasing costs. Consider the following case studies:
· A credit card invitation-to-apply marketer attracted too many applicants with lower-end demographics. Approval rates were unacceptable for many well-responding segments. The mailer developed a database model to improve the likelihood of approval. This was combined with a response model to isolate the most profitable, high-responding segments to mail.
· A magazine publisher examined characteristics of respondents to high- and low-priced subscription offers. The publisher observed that prospects with high-income indicators also were best able to tolerate a higher price. The mailer used a new price-sensitivity model to raise price, greatly improve profitability and expand mail volume without hurting circulation levels.
· A catalog mailer had a voracious appetite for prospect mail volume. The mailer built models on recent new customers to score two large prospect databases. Freshly scored names are now output every quarter for unlimited use in this company’s prospecting.
· A fundraiser uses more than 30 rental lists in its monthly mailings. By passing merged names through a model-scoring process and lowering net arrangements with list providers, the mailer can screen the weakest 20 percent of the prospects individually from the mailing – instead of 20 percent of its lists. This process adds two days to its mail schedule, but it has expanded the number of mailable lists and overall mail volume, improved response rates 10 percent and reduced acquisition costs 20 percent.
· A book continuity mailer relies on prospect mailings to maintain membership levels. Response to the upfront offer was generally acceptable, but weak retention rates made many rental lists unprofitable. A two-step matrix was developed for each list provided from custom response and retention models to lower acquisition costs and increase list profitability.
· A travel services company determined that the expense of mailing its brochures to all past travelers was eating away at profits. Response models for house mailings promoting travel to different regions of the world helped the marketing team sift through 10 years of prospects and transaction data. The new models help the company focus each mailing on the house file names most likely to travel to each of their destinations.
· A retailer wanted to expand support of its local outlets through direct mail. Saturation mailings to certain neighborhoods proved effective, but mailing costs were exceeding ad budgets. A geographically neutral response model was created to weed out poor prospects and improve response rates. The model rank orders prospects without favoring the demographic characteristics of one region over another.
Marketers might be inclined to “let the model do the work.” But success in database marketing requires good planning and a careful analysis of past mailing results, robust and predictive data as well as modeling and scoring resources that will ensure successful implementation.
