This site uses cookies. By continuing to browse the site you are agreeing to our use of cookies. Find out more here

Saturday, 30 August 2014

Why so much segmentation is rubbish

A whole issue of The Journal of Marketing Management was recently devoted to a review of market segmentation. The conclusion was very clear: after 70 years of marketing, it is still done badly. Indeed, in a recent Harvard Business Review article it was claimed that in the US, 85% of 30,000 new product launches failed because of poor market segmentation.

“Boy George and the Pope are both ‘A’s in terms of socioeconomic status, but they don’t behave the same”

The reasons are not difficult to explain. Anyone who says that they segment according to socioeconomics, geodemographics, psychographics or the like is guilty of the worst kind of a-priori production orientation. Boy George and the Pope are both ‘A’s in terms of socioeconomic status, but they don’t behave the same. Nor do all women between the ages of 18 and 24, nor does everyone in my street, and so on. These segments are only useful at a very high level of aggregation – they do not explain behaviour.

There is also a widely held belief that new media and changing patterns of behaviour have made traditional market segmentation irrelevant. They haven’t, nor will they in future. These are the excuses of lazy and incompetent marketing departments who fail to use the tried and tested process of needs-based market segmentation to understand these new consumer behaviour patterns.

Once we’ve defined the market in terms of needs rather than products, we need to describe the flow of goods and services from suppliers through to end use. Wherever important decision-making junctions appear, it is here that segmentation has to take place. Then we must understand what people buy, where they buy it, how they buy it and of course why they behave in this particular way. This will usually reveal at least 20 to 30 micro-segments – too many of course – so a simple clustering will reduce this to seven or eight. All that remains is to describe who populates these segments, which is where factors like socioeconomics and demographics come into play.

We now have some proper needs-based segments, so our CRM system will work. We can even deal with customers in a personalised manner.

But looking to product attributes, service quality, size of purchase and so on as a basis for segmentation is just wrong. They explain why much market research in the world is a waste of time and why I have to spend so much of my life travelling around the world correcting mega-expensive consultant-based segmentation assignments that turn out to be as useful as a bird of prey with a squint, producing loads of non-actionable and confusing data.

As for the question of whether new channels, new media and resultant changing behaviours have made traditional segmentation irrelevant, the answer is a resounding no.

A letter to The Times gave a great example of what goes wrong when needs-based segmentation isn’t carried out. A reader in Surrey had received a leaflet telling him he could save over £200 on his car insurance. But the small print revealed that “all price-saving comparisons included in this leaflet are based on a 44-year-old female living in the Darlington area, with comprehensive cover but zero no-claims discount, driving 12,999 miles per year in a 2002 Rover 25 1.4”. “If she would like to get in touch with me,” the reader offered, “I will pass the leaflet on to her.”

This is a classic example of why most communications, especially direct mail, miss the mark completely. Any marketing manager who says that they had a 3% response to a direct mail campaign is saying that they have annoyed 97% of their customers.

I know that most of my learned and experienced colleagues, on both sides of the Atlantic, will share my own frustration that after so many years most companies and even many academics still have not understood that needs-based segmentation lies at the heart of successful strategies. Any market, b2b or b2c, consists of 100% of what is sold and bought. Our task is to get into that market and understand what these different patterns of actual behaviour are. This is the essence of successful market segmentation.

Malcolm McDonald (until recently Professor of Marketing and Deputy Director, Cranfield University School of Management) has written a chapter on market segmentation for The Marketing Century, published by Wiley on 10 March to mark the centenary of the Chartered Institute of Marketing. The book is available from the CIM at a special price of £12.50.

Follow us on
Follow us on Twitter

Readers' comments (6)

  • Many needs or attitudinal based segmentations fail not because the segmentations are poor but because they cannot be overlaid onto media planning tools. If you have an attitudinal segmentation but you don't know the attitudes of those on your customer database, you can't directly define the segment of each person on your database. Developments recently have enabled bridges to be formed between segmentation algortihms and database variables and the degree to which segmentations can be applied to practical marketing decisions should be improved.

    Unsuitable or offensive? Report this comment

  • Very interesting article.I agree with the basic argument,and with Dave's points, and they prompt me to pose a few questions:

    1. Does anyone have experience of linking a "needs based" -or "needs states" -segmentation with an attitudes based approach? (The former takes account of varying purchase occasions and channels, the latter referring to an approach where an individual's underlying values -perhaps a more precise term than attitudes - are assumed to be reasonably consistent over time, and so the "whole individual" can be described pithily by a "cluster" name). How exactly (and simply!) is this done?

    2. In addition, how do we avoid considerable loss of "richness" in reducing the number of micro segments from 20-30 to 7 or 8, as Mr McDonald describes?

    3. Finally, how does the process apply where one does not have such easy access to purchasing data, for example in the public sector where the dependent variable of behaviour change is often not easy to capture?

    Twitter @StephenHooker

    Unsuitable or offensive? Report this comment

  • I cannot even believe garbage like this is published. Pure marketing hype with NO science behnd it. I've run global segmentations informing the world's largest technology companies fo years - the worst part of this article is “all price-saving comparisons included in this leaflet are based on a 44-year-old female living in the Darlington area, with comprehensive cover but zero no-claims discount, driving 12,999 miles per year in a 2002 Rover 25 1.4”. Either this author doesn't understand basic statistical analysis, or is blatantly misusing results to bolster their own image.

    Unsuitable or offensive? Report this comment

  • @Michael DeHart

    "True does not mean true always or true everywhere..."

    Unsuitable or offensive? Report this comment

  • To pick up on one specific point Malcolm states that 85% of new product launches fail. The implication being that this is too high?

    However, how many new product launches do we think can be sucessful? In the figures above 4500 new products are sucessful, is that close to the maximum the consumer and the channels can handle? If there had been 60,000 new product launches would 85% have failed and 9000 been sucessful, or would we still have seen about 4500 winning through?

    Market research is at best only about improving the odds, it does not deliver certainties.

    Unsuitable or offensive? Report this comment

  • It seems a little too convenient to blame it on segmentation alone.

    how many researchers would know and have been bold enough to tell the client the mistakes they have made (wrong price range, wrong product range, bad packaging, low distribution, wrong media budgets, wrong segment to be in) - and just how many clients have listened in at all?

    I have run several tracking studies, and its a thing of beauty when our recommendations have been acted upon and the tracking results portray it beautifully. In the rare instances it doesnt fulfill, there is still an understanding and learning of what didnt work, and the client ends up richer.

    Segmentation is just the beginning - I do admit in a lot of cases its badly done, but its not the lone reason for failure

    Unsuitable or offensive? Report this comment

Have your say

Please add your comment. You can include links, but HTML is not permitted.
Your email address will not be displayed on the site. All comments are moderated.

Mandatory
Mandatory
Mandatory
Mandatory