We are all aware that, over time, marketers have gained access to a wealth of data regarding their customers and prospects. This includes purchase information, demographics, lifestyle interest categories and more.
We, as marketers, should view information as a strategic resource and must continue to uncover and utilize the power of that information. The data can direct us in making targeted offers to those most likely to respond and in building long-term relationships and trust as we deliver relevant messaging. Make sure you are collecting and verifying current information on a regular basis as you interact with customers, and use the services of outside marketing firms to help you fill in the blanks, track moves, etc.
Make sure your entire database is updated regularly. Revisions can include NCOA, PCOA, phone append/verification, email append/verification, demographic or firmographic appends.
Developing the database and filling it with information is only the beginning of database marketing. Segmentation and other research techniques help companies release the full potential of the data. It is this ability to divide the database into subsets containing common sets of characteristics and behaviors that will lead to marketing effectiveness and efficiencies.
The goal of all marketers is to maximize response, customer activation, retention and cross-sell opportunities. There are many techniques we can use. For clients who have not had much exposure in these areas, start with simple profiles or regression models with specific goals in mind.
Start Small, Think Big
Start simply, then add more complex modeling techniques to help achieve specific campaign objectives.
The following is an example of how a client company used each of these methodologies in its acquisition efforts.
- Customer Profiles: A customer profile provides a descriptive view of the customer base by comparing a sample of the client file to national averages using common demographic and lifestyle variables on a variable-by-variable basis. Profiles can be created for each product category the client offers.
A major national insurance marketer has used profile information to achieve better results in its acquisition efforts. The company currently markets six types of insurance products. The descriptive profiles provided information that was used for its list rental acquisition efforts. While dozens of variables were presented in the actual profile report package, the chart (at right) shows three of the demographic variables used and the potential differences by product.
- Regression Models: While the profile assisted in the initial selection of lists for prospecting, regression models further helped target the most likely responders and buyers.
Logistic regression modeling is a technique that predicts future outcomes (response vs. non-response) from previous results using all the available data. The model creates a scorecard that ranks prospective names on their likelihood to take the desired action. Instead of randomly selecting names from a list, selection is based on the highest probability of response.
Our insurance client desired to improve response/conversion results for the company's life insurance products. Using historical data, a regression model was created allowing the client to target the prospects who were most likely to respond. The table (at right) summarizes results with and without the model.
This model resulted in a 91 percent lift in response rates and a 43 percent decrease in cost per lead. The overall results provided a sufficient return on investment, allowing our client to use modeled selection as part of its future targeting efforts.
It is also important to note that this model was created for use on the life insurance product via direct mail efforts. One can't assume similar results would occur across any marketing effort via any channel. In order to realize optimal results, different models would have to be considered when using different channels for acquisition and/or if promoting different products.
Divide and succeed
Develop separate models for each product and marketing campaign type.
Once you have created models, there are two common strategies that can be implemented into your campaigns:
- Marketing Dollar Reduction: This involves omitting the bottom segments (or those least likely to respond), thus reducing your campaign quantity and overall cost
- Marketing Dollar Reallocation: Here, you are soliciting the top segments (or those most likely to respond) more frequently and soliciting the bottom segments less frequently, or not at all.
The technique you use would be based on your objectives, as well as the anticipated ROI.
Regardless of the objective or the techniques used, a model is only as good as the predictive data and the data preparation techniques used. So it is important for each marketer to collect internal data and add enhancement data to the database. This information must be maintained, updated, and cleansed as it will be the resource used to assist in creating more profitable campaigns.