A Segmentation Cookbook – Part 2

DFA is often asked about our approach to segmentation. We outlined our method in the first part of our segmentation cookbook, and on our web site. However, today we go into more detail about the structure of the segmentation models, and definitions which support them. We should add that we are not necessarily advocating these as complete, or best practice, but we find they do work for us as we analyse households, their preferences and behaviours.

But first, here is the best analogy I can draw. Our overall segmentation is like a rubik’s cube:

DFA-CubeWe can look at households from a number of dimensions, all of which will add up to the ~8.5 million households in Australia. We can turn elements around and slice and dice to focus on the specific dimension we happen to be looking at. We can even map our segmentation to other approaches which are out there. To be useful, our segments must:

  1. provide good separation and definition across the segments, without overlaps;
  2. cover the entire based of the population, without significant gaps;
  3. be repeatable and actionable;
  4. align to external data sources such as the Australian Bureau of Statistics.

Over the past few years our approach has proven robust. So here is a schematic of our framework:

DFA-SegmentDiagWe will often report on outputs from our surveys across multiple dimensions. For example, in our housing reports, the Property Imperative, we focus on the Property Status dimensions. When we examine household preferences we use our master segmentation.

We use this master segmentation to structure our surveys and models to ensure we have statistically reliable samples. For each segment, we know how many households exist, their key characteristics and which postcodes they align to. Here is a brief description of each element in our master segmentation. Our full descriptions are only available to our clients.

DFA-SegmentDiscTo complete the picture we have separate segmentation around digital preferences, as used in our reports on Digital Transformation, The Quite Revolution, and we use a different segmentation framework for our SME studies and Mortgage Stress reports. We are currently researching an appropriate segmentation approach for Superannuation, so watch out for this report later in the year!

A final caveat, I would draw your attention to the terms and conditions on our web site:

Digital Finance Analytics  (DFA) provides this site as is, with no warranty as to accuracy, completeness or ongoing availability. Any material on this site remains the copyright of DFA, and if images or content are mirrored or copied, they must be fully attributed to DFA. Material on the site is generally as accurate as possible but may not always be up to date.”

Author: Martin North

Martin North is the Principal of Digital Finance Analytics