Segments are assessed on the following criteria:
• What buyers desire from the product.
•Purchase channel(s) used.
• Marketing communication channels and product information channels predominantly used by each segment.
• The competitive situation within the segment (e.g., unmet needs, the proportion of buyers who are loyal to you, the proportion of buyers loyal to your competitors, the proportion of buyers having a negative perception of you, etc.).
•The financial characteristics of the segment (e.g., number of buyers, spending level of the average buyer within the product category).
We don't necessarily obtain segments by dividing the sample one time. We may use a two-step division to maximize the differences between segments on multiple characteristics (e.g., spending level in the product category and perception of you).
In recent years, advances have been made in segmentation analytical techniques. One is a family of techniques called latent class. One of the techniques is especially attractive, in that it focuses on an outcome variable. That is, the mathematical function of this technique is to identify respondents who differ from each other with respect to the drivers of an outcome variable such as purchase interest or overall satisfaction. This mirrors the marketing purpose of segmentation: divide a market into distinct groups of buyers who differ with respect to what influences purchase behavior.
Another advance has occurred in an analytical technique called discrete choice. In the past, discrete choice analysis produced utilities only at the aggregate level (e.g., the total sample). A new method of calculating the parameters in the model, Hierarchical Bayes estimation, has been developed that produces respondent-level parameters (utilities). These utilities can be used as variables in the segmentation.
A specialized form of trade-off analysis called MaxDiff has been developed that yields attribute preference scores that can be used as variables in the segmentation.