Clients seek greater targeting capability with their media buys. But what do direct marketers look for when creating more accurate predictive modeling that generates more revenue, more profitably than past efforts?
Identifying best customers and attempting to locate prospects like them represents the fundamental hypothesis of direct response customer profiling. Out of this information, direct marketers create predictive response scores for the most targetable of all media -- direct mail and email.
When developing predictive modeling, analysts overlay external data by matching customer information to external files available from sundry partners such as MRI, Simmons, InfoUSA and so forth.
Based on this enriched customer information and the skillful application of multivariate analysis, analysts create database segmentation schemes for customer databases. Marketers then create specific offers that the analysis predicts the various customer and prospect segments will find most attractive.
Most customer models use 5 or 10 prospect segments rated by propensity to respond.
The nest step involves testing the segmentation schemes for validity.
Ongoing regression analysis compares the mailed lists with the response data in order to refine the customer modeling.
How successful is this approach? It depends.
There are well over 50 reputable modeling consultancies and service companies that specialize in this single activity. And the market continues to expand.
The criteria for using this approach as opposed to the proven and less costly methods of renting response names that work well for a given client's category as well as Recency, Frequency, Monotary (RFM) customer database segementation is driven by volume and the need the expand circulation.
Response files available for trade or rental are shrinking. And many clients require deep service area prospect penetration not available from rented names. Hence the need to find ways to make compiled lists work because they offer nearly 100% market penetration. This means customer profiling using
enhanced data available from such firms as Experian, Acxiom, PRIZM and Abacus to name a few.
How do direct marketers select the best customer modeling vendors? Here are some suggested criteria.
- Use suppliers that have done successful work within the client's industry. They tend to know the quirks that can make big differences in a client's response rate.
- Select vendors that have the greatest number of partner relationships offering rich enhancement to uncover profitable segments other suppliers might miss.
- Make sure that the partner data represent actual response data and not just secondary research. Primary research reflects what people say in an interview as opposed to what they actually do when they pull out their checkbooks.
- Verify that the independent functions (i.e. the data set used to evaluate the outcome in any regression analysis) are available sets of criteria when selecting your direct mail or email segments for acquisition.
Be aware such programs rely heavily upon the client's ability to identify and deliver the customer names, addresses and individual purchase history. Otherwise, the entire project may uncover faulty response predictors in the absence of actual customer specific sales data.
Those clents with true relational databases stand the best chance of creating the profit making opportunities targeting special offers to both their own customers as well as in the acquisition of new, profitable customers.