At its core, any cluster analysis approach attempts to divide your dataset into several discrete sub-groups of individuals — with members of a group thinking & behaving similarly to each other and different from the members of other groups.

While a segmentation tries to find the exact right lens to divvy up the addressable market, there are other more narrowly focused approaches that can surface unique insights. These insights can provide researchers, marketers, and product managers a better lens into how users and customers think by identifying underlying patterns that may not be immediately obvious.

Below are a few examples of how you can apply cluster analysis to real-world business problems to get a better handle on the undercurrents flowing beneath the surface that may otherwise go unnoticed:

Example #1: Cluster Analysis in feature prioritization

The challenge: The vast majority of proposed features for a new consumer technology device have relatively similar value to your target audience, even when cutting the results by pre-existing audience segments.

The solution: By running correlations between a wide variety of demographic and attitudinal predictors and individual feature preference data, we can identify a subset of key audience differences to explore. By clustering based on these variables, we may discover two key groups who both responded very positively to different subsets of features, whose differences are not accounted for in the existing segmentation.

The hidden insights: While the existing segmentation model may be the best solution for analyzing the entire market, it may not sufficiently explain the differences (e.g. work needs vs. home needs) that are driving variance in feature preferences for this new device.

Example #2: Cluster Analysis in customer profiling

The challenge: Pre-existing commercial segments that were expected to have divergent preferences for three different pricing models instead show similar preferences. This seems counter-intuitive based on what is already understood about these organizations based on their customers, size, and revenue.

The solution: By performing cluster analysis (LCA) on the various infrastructure configurations each organization leverages, we can identify distinct sub-groups which exist in varying degrees across the segments and provide a clear delineation in pricing model preference and price sensitivity.

The hidden insights: By deploying cluster analysis we can identify a different set of driving variables which are better able to explain customer preference. This different cluster set could help to add depth and further enrich our understanding of the variances within each segment.

These are just a couple examples of how cluster analysis techniques (like LCA) can enhance your understanding and hone your strategies. The common theme that we’ve found is that when results don’t make sense, you should re-examine your assumptions. One of the best ways to do this is to identify different groupings to evaluate key results. Clustering works across both consumer and commercial audiences, and a wide variety of business goals.