Can segmentation ever deliver the goods?

According to a 2004 Economist Intelligence Unit survey, 59% of senior executives in large companies said they had conducted a major segmentation exercise some time over the previous two years, but only 14% felt they had derived real value from their investment.

Segmentation on the basis of demographics (age, class, sex) and psychographics (attitudes, personalities, values) has been a major theme in marketing for decades now. VALS (Values, Attitudes and Life Styles) was launched in 1978. Yet somehow, despite all the work and all the insights, it has never delivered its hoped-for benefits.

David Yankelovich – one of segmentation's earliest pioneers – recently suggested some reasons behind this disappointment. The heart of it: a failure to achieve 'joined-up' thinking and actions. The attempt to divide markets up into targetable, manageable bits ended up working backwards, dividing consumer attributes into unrealistically separate categories (attitudes versus behaviours versus demographics) while dividing companies into warring function silos.

Classic marketing textbooks such as Philip Kotler's Principles of Marketing list many alternative approaches to segmentation – geographic, demographic, psychographic and behavioural, for example. But Principles presents each of them as alternatives: as an either/or without explaining how or why companies should opt for one versus another. Nowhere does it explain how these different approaches might be brought together to create a single, unified approach.

Likewise, in companies, segmentation tended to exacerbate rather than overcome functional misunderstandings. VALS' categorisation of consumers into one of eight groups was an instant hit with advertising agencies, because it lent itself to emotional communication. So agencies used VALS and its successors to craft powerful creative work. However, while psychographic segmentation was great for the communications side of things, it had little practical to say to marketers wrestling with diverging consumption and marketplace behaviours. So the resulting advertising 'simply did not invoke the drivers of commercial activity,' notes Yankelovich.

Furthermore, approaches to segmentation focused on customer behaviours, and product preferences had little to offer to marketing communications and tended to create brand architecture nightmares ('Do we need a new brand, a sub-brand or a brand extension?').

Segmentations based on good old-fashioned demographics found a stronghold as a media buying trading currency (differential pricing for media delivering more ABC1 males aged 16–24 than C2DE females aged 45+ for example), but provided precious little insight into underlying consumer motivations.

Price point segmentation was instantly attractive to finance directors, but left marketers struggling to justify the different prices they were expected to charge. Every silo in the marketing department was therefore attracted to its own approach to segmentation, and none of these approaches really 'spoke' to each other.

AN APPROACH TO A MORE ROBUST FORM OF SEGMENTATION

The ideal, clearly, is one that makes a positive connection between attitudes, behaviours, purchasing preferences and economic value, while also helping different marketing silos work better as teams. The approach should serve, say, both new product development and advertising and media buying.

A recent project (originally not intended to tackle the challenge of segmentation at all) may throw some light on how this might be achieved.

ECR EUROPE RESEARCH

ECR Europe is a packaged goods industry group. It brings major retailers such as Tesco, Sainsbury's, Ahold, Carrefour and Metro together with leading brand manufacturers such as Nestlé, Procter & Gamble (P&G) and Unilever to find better ways of working together in areas of common interest, such as improving supply chain efficiencies and developing better merchandising in stores.

Eight years ago, it set up a working party to learn from 'best in class' companies such as Dell, Hennes & Mauritz, Nike and Yahoo! From outside the packaged goods industry. The working party (including representatives from leading companies such as Tesco, Albert Heijn/Ahold, P&G and Kellogg's) then went one stage further to ask: 'What underlying consumer trends have these successful companies responded to?'

COMPANY RESEARCH IS INEVITABLY TOO NARROW

When they looked at their own internal research they realised just how narrow most of it was. Most market research is commissioned by a particular company to address a particular problem, and never strays far from this remit. Its real aim is to solve the company's immediate problem, not to discover the consumer's agenda. It's extremely useful if you want to know whether your widget should be pink or blue, but not particularly helpful if you want to understand deep, longterm consumer trends.

So the working group initiated a massive research project: simply ask 2000 consumers what matters to them in their lives as consumers – in open-ended discussion – and see what comes out.

The research, by Roland Berger Strategy Consultants, produced a massive databank of attitude statements which were boiled down to a bank of 80 or so statements that summed up the key points from the open-ended interviews. These 80 statements in turn, tended to cluster together to reveal 19 core 'values' which most people either identify with or positively reject in their lives.

Elaborate as this research project was, it still fell prey to a problem identified by Yankelovich. Advanced statistical techniques displaying great 'technical virtuosity' are all very well, but the outcome is often unusable. Managers don't understand them or are suspicious of them. Practical 'Monday morning' conclusions are elusive.

MAINSTREAM VALUES VS FRINGE VALUES

So the group took another look at the data and noticed two things. First, some values are widely embraced by a majority of consumers, others by minorities: mainstream versus fringe.

Second, if you compare all the answers to all statements across the total sample, some values tend to be 'close' to each other (i.e. they tend to be embraced in clusters by the same people) and others are 'distant' (if a person embraces value A, he is statistically unlikely to also embrace value B). The researchers realised it was possible to turn these relative statistical 'distances' into a visual 'map' (See Figure 1).

This map places commonly held values such as 'quality' and 'service' in the middle, with less commonly held values (such as 'Thrill & Fun' and 'Customised' on the edges. It also places values that tend to be held together close to each other, such as 'Fair' and 'Nature', while values that tend not to be held together are placed far away from each other (e.g. Thrill & Fun, and Total Cost).

While the number crunching necessary to do this is awesome (multidimensional scaling analysis) the resulting visualised picture is very easy to 'see'. Figure 2 shows, for example, the values profile of two different individuals. Light grey areas show which values that individual embraces more positively than the average in the sample – with each contour line indicating a degree of statistical significance. Dark grey areas show which values that individual rejects more strongly than the average, again with contour line indicating a degree of statistical significance.

VISUALISING VALUES

Figure 2 shows how this profile is used to visualise a mountain of data so that anyone can grasp its core message, in an instant. The figure shows the values profiles of two very different individuals who we have called Hans and Mike. The light grey areas show the values that the individual positively embraces (where 'positive' is compared to the other individuals in the sample).

The more 'contour lines' it shows, the more strongly committed the individual is, with each contour line representing a degree of statistical significance. The dark grey areas highlight values that the individual positively rejects. A white area shows that when it comes to these values the individual is perfectly representative of the sample – he or she doesn't stand out as caring more or less than the people he is being compared to.

In this particular case, both Hans and Mike are positively attracted to the value '24/7 Protech' – which means they are keen on the latest technology, interested in scientific innovation, want quick and efficient access to information, and favour 'cold' transactions.

But there the similarities end. For Hans, issues such as 'Quality' and 'Service' are important, as is concern for the environment. He's positively turned off by the adrenaline-seeking, risk-loving value of Thrill & Fun and related values such as Carefree, Vitality and Passion. Mike, on the other hand, lives for these values while also keeping a sharp eye out for best value. He's also positively irritated both by altruistic concerns for fairness and the environment and for boring, traditional priorities such as quality and service.

HOW VALUES ARE REFLECTED IN BRANDS AND MARKETS

In this way, a picture of each individual's driving values can be identified in a statistically robust way. But that is only the beginning. This information can be used to throw new light on brands and markets.

Once again, our number crunchers went to work. The first thing they did was to statistically aggregate data from many different individuals, to compare the values of users and non-users of specific brands.

Figure 3 shows some examples from Germany. BMW buyers are positively 'pro' values such as 24/7 Protech, Personal efficiency, Service, Quality Proven, New & Cool and Passion while positively rejecting values such as Nature, Fair, Pure, Smart shopping and Total cost. Users of Aldi, the hard discount grocery chain, embrace almost the exact opposite values.

By mapping the values of actual users of brands, it is possible to build a picture of each brand in the market, profiled not by the functional attributes of the product or service concerned, or what its marketers like to think their brand stands for, but by the values of the people who actually buy it. This is called the brand's 'actual values proposition' because, like it or not, this is the brand's actual performance in the marketplace. This is the profile of the people the brand is currently successfully attracting.

These values profiles identify target markets more clearly. By grouping individuals with similar profiles together, patterns emerge. Some groups coalesce around some clusters of values; other groups coalesce around other clusters. These clusters are called 'archetypes'. And drawing on questions about age and income reveals a picture of the nature and economic value of each archetype.

EIGHT MAIN CONSUMER ARCHETYPES IN GERMANY

Thus, Figure 4 shows the eight main consumer archetypes in Germany with archetype names that have been invented to describe their main attributes. For simplicity, the values names are not shown, but they have the same positions are previous figures. The Maximalists archetype accounts for just under 14% of the population. Their average age is 35 (a relatively young group). They tend to be male (56% vs 44% female), and more likely to be single. Of those that are married, they tend to fall into the 'double income no kids' category; they are the most affluent archetype.

Minimalists on the other hand account for about 9% of the population, have average income and an average age of 42. They are dutiful types, rejecting status symbols and living by the motto 'less is more'. Humanists meanwhile account for 16% of the population, are highly ethical and sceptical of the benefits of new technologies, with an average age of 50. They have below average income.

HOW ARE VALUES-BASED SEGMENTATIONS USED?

The results are more than intriguing: they are extremely useful. The data can now be used to guide every aspect of brand decision-making from identifying desirable changes to brand propositions (to appeal more or less to certain values groups), to identify innovation opportunities, to reorganise brand portfolios, to judge advertising creativity and even plan media buying.

What's more, the fact that the whole process is data-based helps create a common language that unites all parties, including accountants and marketers. And the fact that the results are so visual makes it easy to discuss and communicate, whether it is at the level of strategy formulation or the finest detail of execution.

THE RULES FOR ROBUST SEGMENTATION

From the experience gained so far, it looks like some general rules for robust segmentation processes may be emerging.

1. First, the segmentation needs to embrace all the dimensions of consumer reality – emotional, functional, economic, etc. – in a way that can be 'joined up'.
2. Second, it needs to be data based.
3. Third, these data must have a firm centre of gravity: the 'ultimate building blocks' of markets – individuals, not products or brands or aggregated groups of individuals. Hard, irrefutable data are essential to win credibility among all factions, and for the many practical applications that are needed.
4. Fourth, the data must be modular so that they can be aggregated, sliced and diced at will, and 'attacked' from any angle for any purpose.
5. Fifth, the data also need to be 'taggable' so that, say, demographic, income, propensity to spend or media viewing data can be appended as another data field to each individual's statement bank answers. One of the benefits of the profiling approach is its ability to turn 'soft' emotional values into hard, taggable data via the scoring of answers to the statement bank.
6. Because it is taggable it should also be 'drill-downable' so that different users can interrogate the same data banks for different purposes; so that the finance director can ask 'What is the relative economic attractiveness of these different segments?' and get a useful answer, while the media buyer can ask 'What are each of these segments' distinctive media habits?' and get an equally useful answer.
7. Even though the actual statistical analysis at the heart of the process may involve a good deal of rocket science, it must be 'visualisable' so that anybody can immediately and intuitively 'see' what we are talking about, without their head spinning from jargon, getting drowned in data or wrestling with arcane concepts they don't understand.
8. It provides a common language. When the finance director and the media buyer talk together, they can point to the same picture and be confident they are talking about the same thing. Only if the data are visualisable in this way can they act as a common language that brings the various silos together.

CONCLUSION: TOWARDS MORE 'JOINED-UP THINKING'

Over many years many different silos have grown up in marketing: the advertising agency and its creativity and emotional empathy; the database marketer characterised by precision and analysis; finance with its preoccupations; product development, and so on. Perhaps one of the reasons we find segmentation so hard is because we have segmented ourselves so much. When we 'focus on the consumer' we see different aspects of the same reality depending on where our starting point is. Segmentation thus has a nasty habit of rebounding inwards. To make a positive contribution, segmentation needs to overcome, rather than exacerbate, this tendency.

ACCOR HOTEL GROUP

By identifying which values archetypes it wanted each brand to attract, Accor, the hotel group, used values analysis to clarify and redefine its various hotel brands. For example, it repositioned the Dorint-Novotel brand to woo 'performers', who value personal efficiency highly, changing its product offering accordingly: introducing automated check-outs, car hire facility, 24-hour food, wireless computing availability, and so on.

In contrast, it repositioned its Dorint-Sofitel brand to appeal to values such as Passion, Classic and Customised. It addressed these values by introducing more experienced concierge staff, fine art on the walls, fine wines in its restaurants, real fires and a live piano in the reception area, a library, etc. By understanding clearly which values it wanted its brands to appeal to, it knew what changes to make to each brand's proposition.

ROCHE DIAGNOSTICS

Roche Diagnostics used values analysis to sort out its brand architectures and global brand strategies. Roche produces a wide range of medical diagnostic equipment for use in a variety of medical conditions. It had developed different brands for each line of equipment, but was surprised to discover that the values held by these brands' users were very similar.

While some rival brands attracted technology innovators or low-cost alternatives, Roche brands attracted users committed to values such as Proven and Quality. Roche used this understanding to create one, single umbrella brand for all its equipment in these areas (such as the Affymetrix AmpliChip pictured), with a much clearer focus on product attributes that address these values: reliability, clear and understood processes, and so on.

This article is based on Moment of Truth: Redefining the CEO's Brand Management Agenda, by Andreas Bauer, Bjoern Bloching, Kai Howaldt and Alan Mitchell, published by Palgrave Macmillan. www.palgrave.com

This article featured in Market Leader, Spring 2007.

FIGURE 1

FIGURE 2

FIGURE 3

FIGURE 4