Choice Modeling

Choice modeling is a methodology in which we ask respondents to rate only a limited number of “hypothetical” products from which we can model what their choice would be to any product that could be created from a combination of the features.

The underlying conceptual approach of choice modeling is that any product is really a combination of its component features and that one can create additional products by varying the level of its features. The output of the choice models are mathematical importance scores typically called “utility scores.”

Why Use a Conjoint Model?

We have found that the most accurate and projectable way to collect information on what is important to customers is to derive rather than ask customers to tell us directly what is important to them.

To derive this information we would implement a choice study where respondents are asked to choose between several products. In this way we never ask a respondent to tell us how important price, brand or a particular feature is when they purchase. The resulting conjoint “utility scores” tell us relative importance. Simulation software is available to assist in calculations of share of preference or “winner take all” calculation with the client product/service and all competing products/services.

Segmentation
Data tied to the audience members are selected and collected and fed into the Segmentation, Response and Value Models.

Survey Data (to be collected during phone interviews)
Usage Data (from Client database)
Class Profitability Data (from Client)
Demographic Data (Example: Experian Data Bundle applied at a household level)
Historical Marketing Campaign and Results Data (from Client or Agency)

Contact Optimization
Contact Optimization ensures that the client’s marketing budget is focused on the right customer, based upon likelihood to respond and profitability

The following is a brief description of the tasks involved in Contact Optimization development:

Compute the actual dollar value of each existing, active member, using the profitability by class data
Build regression models to identify prospects that are most likely to activate their benefit based upon existing campaign and usage data in the client database.
Set the prospect contact budget for each qualifying prospect household
Construct a scoring algorithm to assign each prospect the optimal contact stream
Set the prospect contact budget for each qualifying prospect household
Construct a scoring algorithm to assign each prospect the optimal contact stream

Data used for Contact Optimization is independent of the Segmentation data. There will be two models created and delivered in SAS or SPSS.