Segmentation – the devil is in the detail

Bryan Urbick highlighted many of the most common challenges of running segmentation studies in his article ‘A life beyond segmentation’ in the October edition of Research. I wouldn’t characterise segmentation studies as “all the rage”, but I have seen the number of segmentation briefs growing in the last few years. I think this is because more and more clients are seeing the value of a well-designed model.

With research budgets contracting, it seems segmentations are as much in demand as they have ever been. Few types of research are better able to help a marketer decide where they want to play (and where not), and how to prioritise marketing spend and initiatives. They shouldn’t be seen as a silver bullet, but I’d suggest that putting inadequate resource into them is a more common pitfall than allocating too much.

First, collaboration is needed between client and agency in building the model, in order to make sure that the business’s knowledge and category expertise is leveraged, and also to start driving ownership within the business. Too often the stakeholders have seen models that have simply not affected the business, despite significant investment, and thus they come to the project with reservations.

Secondly, when the segmentation has been delivered the work has only just begun.  Even assuming we have built a sophisticated model that bears a healthy resemblance to the category in question, good things don’t just start happening on the back of it. Most immediately, there needs to be a structured engagement plan to disseminate the work through the business. It’s not enough to rely on either the brilliance of the work or the passion of the team.

There also needs to be a clear roadmap of how strategic actions are going to be informed. Talking to the client at the start about how the segmentation will progress to activation (such as portfolio planning or the innovation cycle) will start to clarify what the research will and won’t deliver, and what workload lies beyond.

Bryan also highlights the trade-off between complexity and comprehensibility. Any segmentation will only ever be a simplification of reality, but some degree of complexity is usually required if the model is going to have any realism and credibility. Thus a multi-dimensional approach is often a necessity. How you then manage to describe or package this in a simpler, comprehensible way is the more pertinent challenge – some clients are shifting away from multi-dimensional approaches, not because they don’t believe in the validity of the model but purely because the business struggles to digest and use all that detail.

What Bryan referred to in his article as “modal thinking” seems to be what many segmentations capture through need states (or whatever terminology one uses to define needs experienced on a specific consumption occasion). And if consumers were switching typology across different times of the day, it sounds like they weren’t defined by truly fundamental values, attitudes or personality anyway. Typologies should be defined by the kind of criteria that is core to a consumer’s life, which might evolve over life stages, but shouldn’t really shift between breakfast and dinner time.

There is definitely a benefit in looking at people-defined segments alongside occasion-defined segments within the same model – anything less only gives you part of the picture. When you then understand the relationship between people segments and occasion segments, you start to get to a more powerful model and a more accurate reflection of influences on consumer behaviour.

In order to simplify what can be quite a complex picture, you can examine the relative influence of the two types of segment on consumers’ choices. Choices in frequent purchase or ‘repertoire’ categories (such as confectionery) tend to be driven more by need states, whereas choice in categories with narrower repertoires (such as cars or newspapers) tends to be driven more by consumer segments. Understanding this is a good way to prioritise strategic thinking, rather than putting all segments and combinations under the microscope.

So to make sure your segmentation has impact, understand how the work is intended to be used when designing your process and framework and when selling in the approach. Manage stakeholders’ expectations. Don’t dumb down the model – but do simplify how you embed it in order to maximise comprehensibility and actionability. And don’t underestimate the resource and planning needed within the business, especially at the back-end when insight needs to lead to action planning.

Michael Martin has worked in research and consultancy for 18 years, and specialises in segmentation studies and their strategic application.