Cluster analysis is a way of categorizing a collection of "objects," such as survey respondents, into groups or "clusters." Markets may be composed of distinct segments, consisting of customers who have different needs and desires, and who want products with different capabilities. Cluster analysis, widely used within marketing research for the past 25 years, can be especially helpful in identifying potential market segments. Ensemble Analysis is a newer approach that leverages multiple cluster solutions (an ensemble of potential solutions) to find an even better, "consensus" solution.
Cluster/Ensemble analysis lets you see whether survey respondents "cluster" naturally into identifiable groups. You can use product preferences, desired benefits, usage habits, product requirements, or other variables to explore the underlying structure of your market.
Once identified, clusters can be characterized in terms of demographic or lifestyle variables, product consumption, company size, SIC codes, etc. Cluster analysis can help identify market segments and evaluate their potentials as targets for strategic marketing. It can provide insights into: how to position your products, opportunities for new products, targeting sales efforts, and distribution channels to use.
A problem faced by users of cluster analysis is that every cluster analysis always produces clusters, whether there is any underlying structure in the data or not. Because we humans have the ability to read meaning into even the most random of patterns, the fact that a solution seems reasonable is no guarantee that the results would be reproducible with a different sample of customers, a different set of variables, or at a different time of the day.
Our Convergent Cluster & Ensemble Analysis (CCEA) System addresses this problem. It uses a "k means" method of determining clusters (or for developing a consensus solution, based on the notion of "clustering on clusters") that involves iterating from random but strategically chosen starting points. CCEA automatically replicates each analysis 30 times. Each replication is compared to every other to assess its reproducibility. The most reproducible solution is chosen automatically, and its level of reproducibility is reported.
You can use CCEA with data from any source, so long as you format the data as .csv (comma-separated values) files. Such files are saved easily from common programs such as Excel.