Segmentation provides a useful way of understanding demographic, attitudinal and/or behavoural differences within populations or consumer groups: people with similar characteristics may be described as 'clustering together' on the basis of those similarities.
The first question that needs to be asked is whether a particular universe segments at all, in a way that is meaningful to marketers or policy decision-makers. For instance, emerging markets, or groups of people who are already quite niche and tightly knit in nature, may simply not segment very well - it can be like splitting hairs. Other practical questions include whether segments that are found can be replicated consistently and repeatedly, do they make intuitive sense, and are they targetable?
But back to segmentations that work. Most groups which are large and diverse enough to be composed of people with differing life stages, attitudes and product needs can be effectively segmented via thoughtfully designed research. There's no one size fits all solution to segmentation - at MevCorp we approach segmentation design on a case-by-case basis. While different segmentation approaches - demographic/lifestage, geodemographic, psychographic, behavioural and motivational/need states - have gone in and out of fashion over the years, the truth is that they can all be useful depending on the task. Sometimes a hybrid approach (e.g. psychographic and behavioural) is called for.
While a quantitative segmentation gives you a solid grasp on the number, size and profiles of segments, a qualitative exploration phase, preceding the quantitative work, is highly recommended. The exploration phase helps discover how a market or group of people naturally segments, so we're not trying to superimpose any existing assumptions on the exercise. Then we measure and analyse those naturally occuring segmentation characteristics.
The statistical techniques we use to arrive at the segments are, firstly, factor analysis, and secondly, cluster analysis.
Return to questions