Why Segmenting

Everyone who ever started building a brand positioning using segmentation studies or profiles of consumers in different countries must have experienced this moment of despair when you wish you would not have started from that kind of data.

Well, what's the problem with knowing the differences between groups of people? It must be good. It's hard data, isn't it.

First of all we have to consider the statistics a research agency applies to get to consumer segments. In most cases it's about a cluster analysis that tries to explain variances in data. (This is incorrect terminology-wise but a very good metahpor for what they do.) So, even if later on segments get those simple names like e.g. "Traditionalists", we have to understand, that they get their names based on some variables that make them statsitically different from other clusters of people. In most cases it's the variables of minor "strength" that make them different since the universal dimensions rather tend to unite groups of people. But the statistics of cluster analysis favour the over-average-index and devaluate the absolute figures within a segment. Later on, clients look at "Traditionalists", as if this segment was really mainly and only driven by "Traditionlism", whereas it's just the statistical analysis that was driven by it. The thinking in index vs "real" percentage is always a logical problem for a planner and there is no easy way out. The index characterses a group quite well vs others but does not depict which traits and expectations PREDOMINANTLY drive perception and consumption of those people.

Secondly, we have to see that segments imply that positioning a brand is about making an offer to a specific, homogeneous group of people. That's almost the exact opposite of the task we most often get as planners: GROWTH, meaning finding new people who could want our brand for some kind of reason. So, when using segmentation studies as a basis for positioning, in most cases you start to look out for combinations of segments that could be united by a brand proposition despite the differences between them. So basically you start by finding differences to then look out for similarities. Why do that? Why don't start with similarities straight away?

Another big problem about most segmentations is that you will hardly ever be able to recruit exactly those segment members for a concept test or to target them precisely through media. Just because it's absolutely not practicable to use all those dimensions and differences in dimension scores to represent the segments in other quantitative tools. So, bascially you pretend to target specific needs of specific people but will never "meet them in person" again.

Am I too critical about the segmentation approach? Yes I am, probably. Segmentations are a good way to check positioning concepts if they are really really simple. And a very good way to make them simple - and usable - is to NOT cluster people but occasions! Once you accept the fact that people are just bodies carrying around potential behaviours, you instinctively start being interested more in those behaviours than in the carriers of those behaviours. I know the dominant ideology of marketing today is "individualism" and the believe that people make their own decisions and every one is special. But that is just an ideology, not a fact of nature.

There are major advantages of such an occasion-based segmentation approach. First of all: occasions as opposed to people are easily separable, whereas people are hard to separate into completely distinct groups. So a party drinking occasion is definitely different from a drinking-alone-at-the-bar-occasion; whereas a person who drinks often at the bar AND at parties would be a problematic case when grouping people as in meaningful consumption segments.
Secondly, you can always ladder up benefit-wise starting from occasions but you can never do that starting from people. Of course, laddering is not the only way to arrive at a brand's proposition, but it's one of the most powerful... still.


Same thinking applies partially to finding a global positioning starting from country differences. Though countries, of course, are more "real" than statistical segments. But nevertheless, we would automatically start looking for similarities after the differences had made us crazy and dizzy enough. And the problem with that is that similarities are common denominators between segments - say the most general notions you could think of. Well, actually this is not a real problem as such. We just could start there straight away and work out how to make the brand special - though in touch with universal drivers of behaviour, instead of thinking about how to make it "fit" into contries ...or segments ...or other boxes.
(Of course, if your title is "international planner" or sth. alike you should rather emphasise differences between countires to justify your job and travel cost:-)