Market Segmentation Methods
Decision Analyst’s Advanced Analytics Group searches for and identifies patterns in the data. Rigorous analytic techniques (including factor analysis, discriminant analysis, k-means and hierarchical clustering, latent class segmentation, and Factor Segmentation™) are used to organize consumers into groups with similar attitudes, needs, and desires. The size and market potential of each customer segment is determined, along with the positioning and appeals that should be employed to reach each segment.
Market Segmentation Methods
Factor Segmentation™ begins with factor analysis (hence, the name). The model segments the respondents on a mutually exclusive basis (i.e., each respondent is assigned to one segment only) and may be followed by segmenting on a nonmutually exclusive basis to examine the overlap among segments. Factor Segmentation™ yields coherent clusters of respondents with very similar attitudes and perceptions, and is an important technique in developing targeting, positioning, and marketing strategies.
K-Means Cluster Analysis
K-means cluster analysis attempts to identify relatively similar groups of respondents based on selected characteristics, using an algorithm that can handle large numbers of respondents. This procedure attempts to identify similar groups of respondents based on selected characteristics.
TwoStep Cluster Analysis
This procedure is relatively new. It uses hierarchical cluster analysis and is designed to handle very large data sets. The algorithm employed by this procedure has several desirable features that differentiate it from traditional k-means clustering techniques: the handling of categorical and continuous variables, and automatic selection of the number of clusters. By comparing the values of a model-choice criterion across different clustering solutions, the procedure can automatically determine the optimal number of clusters.
Latent Class Cluster Analysis
Latent class cluster analysis produces an objective segmentation solution that optimizes the number of clusters and the fit of the segmentation model to the data. This model can predict patterns in multiple dependent variables (such as attitudes, needs, and behaviors) as a function of segment membership. It easily incorporates data from different types of questions and different types of scales (e.g., yes/no answers, multiple choice questions, various rating scales, and even volumetric data) without the need for rescaling or normalizing the data. Latent class cluster analysis can introduce secondary variables (brand usage, demographics, etc.) as covariates that correspond with needs, attitudes, and behaviors. Respondents are assigned to the cluster to which they have the highest probability of belonging.
Latent Class Choice Modeling
Survey respondents select their most and least preferred sets of product benefits and rate the influence of these benefits on the purchase decision. Latent class choice modeling classifies customers into segments based on their preferred product benefits. This type of segmentation is ideal for customizing product offerings or bundles to match segment preferences, enabling the firm to maximize business performance.
Once a market segmentation model has been produced, a scoring model (a set of equations) may be developed to allow additional respondents to be classified using the same segmentation scheme. Discriminant analysis is usually used to develop the model, although other forms of regression may also be employed. This analysis identifies the questions that are the most important in determining segment membership.
Database Customer Segmentation
Most companies have multiple databases containing information about their customers’ attitudes, preferences, and buying behaviors, but rarely are these databases fully linked and integrated. Linking the information in these various databases enriches the value of each database. Our Advanced Analytics Group has developed unique ways of linking segmentation solutions based on attitudinal and/or preference data with transactional databases. They “bend” the segmentation solution in a way that optimally finds and creates “hooks” into each database, while maintaining the basic structure of the segments. We can apply these “bending” techniques to identify high-potential customer segments that merit special marketing attention.
Insight Into Target Segments
Segmentations are only useful if they can be applied. Once a few target segments have been identified, further analytic work can answer these strategic questions for each segment:
- What factors (drivers) impact the outcome of interest, such as purchase intent, intent to prescribe, or intent to use? Key Driver Analysis
- Which key drivers are important? Strategic Attribute Mapping
- Which important drivers are opportunities or risks for the brand? Opportunity/Risk Analysis
- What is the market potential? Volumetric Analysis
Key Driver Analysis
This analysis measures and orders attributes based on their influence on the outcome of interest, such as likelihood to purchase or likelihood to subscribe. The attributes included in the analysis frequently arise from these areas:
- Opinions of the brand (image)
- Cost relative to other brands
- Experience relative to expectations
- Value for the money
- Perceived availability (location)
In the example chart below, we can see that among the members of the On-The-Go segment, "Offers good deals" and "Service is friendly" were considered the most important drivers of likelihood to purchase.
Strategic Attribute Mapping
These graphical displays use importance and performance to identify strategically important attributes that are:
- High in importance and high in performance: These are attributes that are important and the brand delivers.
- High in importance and low or average in performance: These are attributes that are important yet the brand fails to deliver.
In the following example map, "Offers good deals" and "Check out lines are short" are considered strengths. "Store is clean" and "Service is friendly" are opportunities for improvement.
Opportunity and Risk Analysis
This analysis quantifies the degree of opportunity or risk for each attribute. That way, effort in maintaining and improving product or service attributes can be optimally allocated.
- Risk drivers are those attributes that must be monitored to maintain purchase likelihood levels.
- Opportunity drivers are those attributes that represent prospects for improvement to increase purchase likelihood.
In the example chart below, among members of the On-The-Go Segment, failing to maintain "Offers good deals" would result in an 8% decline in likelihood to purchase. Increasing perceptions of "Store is clean" would result in a 12% increase in likelihood to purchase.
When asked of a representative set of competitive brands or product types, the results of purchase behavior measures can be translated into a very accurate estimate of aggregate category volume and share. This type of consumer-driven approach to sales analysis has many applications and is particularly useful for new categories or those categories for which there is little reliable sales data available.
Potential volume can be calculated using purchase frequency, purchase volume, and products purchased along with their associated prices. Sales volume is extrapolated using competitive set incidence information and calibrated using known sales data (if available). The projection can then be cross-tabulated by consumer segments and reported in dollar volume, units, or market size in terms of number of consumers.
The example volume analysis below shows dollar volume for the total convenience store category (minus gas sales), along with the potential volume contributed by each market segment.
Per Capita Expenditure
Analytical Consulting Services
Decision Analyst is a global marketing research and analytical consulting firm with over 40 years of experience in state-of-the-art modeling, simulation, and optimization. A team of Ph.D.’s heads up Decision Analyst’s market segmentation work. They also publish many white papers on advanced analytical methods and speak frequently at marketing research industry conferences. They program models in SAS, Sawtooth, and the R-Language.
If you would like more information or would like to discuss a possible project, please contact contact Jerry W. Thomas, President/CEO (firstname.lastname@example.org), or Elizabeth Horn, Ph.D., Senior Vice President of Advanced Analytics (email@example.com), or call 1-800-ANALYSIS (262-5974) or 1-817-640-6166.