MARKETING MATERIAL

Market segmentation is widely talked about in the literature and by practitioners.  Segmentation appears to be generally regarded as an important marketing tool (Mercer, 1996) - dividing consumers of a product into more or less homogeneous subgroups which might need to be evaluated or treated differently (e.g. for media selection):

            But the status of "segmentation" is not in fact very clear - what it means, where it is expected to apply, and even whether it exists.  The leading marketing management text has devoted a chapter to it for more than three decades (Kotler, 1967).  In contrast, the leading consumer behavior text did not mention segmentation in the index (Engel, Blackwall, Kollat 1978).  In explanation, Wensley (1995) recently sought to distinguish between weak, semi-strong, and strong forms of segmentation.  On the other hand, since Wendall Smith's original paper (1956) there have been no significant reports of clear-cut brand segmentation results.  Instead, many discussions of segmentation in the literature (recently, for instance, Jeididi et al, 1997, Krieger & Green, 1996, and Ramaswamy et al, 1996) still continue to put forward statistical techniques (new or old) by which to search for segmentation, without illustrations of any well-established results. 

BRAND OR PRODUCT SEGMENTATION?

            Consumers all differ from each other in their consumption of products and brands (they are "infinitely segmented").  But are there any groups of consumers which are more homogeneous in their relevant behavior within such groups than between?  This depends on whether we are distinguishing between (i) product categories, (ii) brands, or (iii) subcategories or varieties of a brand.  Thus:

    ·       Product Categories

            Here segmentation is often self-evident.  Thus dog food is only bought by dog-owners (though the segmentation for other categories may not quite so distinct).  Again, heavy, medium and light buyers of the category differ, and this may also relate to the size of family.  But such analyses are perhaps better described as "defining one's market."

    ·       Brands 

            In most cases, these are functionally very similar to each other; any brand which develops a competitive advantage is usually quickly copied.  Our theoretical expectation is, therefore, that different brands are so similar that they generally  appeal to similar kinds of people.  Our empirical finding here is that this is in fact so.

·           Subcategories and Varieties

            A product category is usually made up of functionally distinct sub-categories (e.g. moist and dry dog foods; large and small motor cars, or expensive and cheaper ones;  pack-sizes; pre-sweetened breakfast cereals; unleaded gasoline; etc).  Leading brands in the overall category are also typically represented in these subcategories (e.g. pack-sizes). Thus, there often is segmentation, between different subcategories, but not always so; or not strikingly so (most soup flavors or different car colors do not segment, except for company limousines and funeral cars being black).

            Over the years, segmentation has mostly been linked with media planning, where the bulk of advertising is for brands rather than categories or subcategories (nor yet for varieties.)

            In this paper we focus on brand segmentation.  This is partly because it is where the general expectations and the facts seem to most obviously conflict. It extends an earlier study (Hammond, Ehrenberg and Goodhardt, 1993 &1995) of segmentation which showed that there was virtually no segmentation between competitive brands across a range of packaged goods product categories. In order to generalise the previous results, in this paper we take the analysis further to wider ranges of:

                        (i)         Products and services, including for example cars and financial services.

                        (ii)        Other segmentation dimensions, including life-style.

The conclusions are the same as were found previously, namely that brands which are competitive and therefore seen as similar or substitutable appeal to much the same kinds of people.

METHODOLOGY

            Following Hammond et al (1993), we have examined whether users of competitive brands differed on a range of potential segmentation variables. 

            The data were provided by BMRB International's Target Group Index surveys (TGI) in the UK for 1993 and 1995, which cover a large number of brands (in excess of 1,000) and are based on especially large samples (covering in excess of 45,000 adults)  The TGI uses an extensive questionnaire which is personally placed, but returned by mail.  The basic measure analysed here was that of penetration, where the brands were recorded as either "most often" purchased, or, for infrequently-bought products, "most recently" purchased (rather than the market shares based on consumer panel data that were available in the previous study).  We have examined "Housewives" for groceries, and "Adults" more generally elsewhere. For life-style variables the analyses were by whether respondents "agreed" with the statement for that brand.

            The form of the analysis was the same as for Hammond et al. For each available segmentation variable, the total number of consumers buying each of the leading brands "used most often" were converted into brand profiles. Table 1 illustrates this for age, applied to toilet soaps.

[Table 1]

            The age-profile of each brand's "users" was then compared with the "category

users" profile by age.  The deviations are shown in Table 2, with their Mean Absolute Deviation (MAD), for each brand, shown as a summary measure in the right-hand column of the table. 

[Table 2]

The average MAD was 3.2, i.e. just over 3, as in the Hammond study.  The larger deviations again mostly occurred for smaller brands, with smaller sample bases. Hence the equivalent weighted average MAD was smaller, at 2.3.

            Similar analyses were also carried out for a selection of the 100 lifestyle/attitudinal statements recorded for each brand in the TGI data base.

RESULTS

SOCIO-DEMOGRAPHIC

                 Nine of the seventeen different variables recorded by TGI  were examined, but three were chosen as being typical, for reporting in detail here. These were: Age (by 5- or 10-year age groups), Social Group (A to E) and Education ("Still studying" or age of completion).  The summary data for the 340 separate leading brands, across 25 separate markets (as in Table 3), are reported in this way. 

            As stated earlier, the average MAD overall was 3.2. This was in line with the average reported by Hammond et al across a range of very different markets.

[Table 3]

            Two categories have relatively high MADs, at 5 or 6:  Chocolate (and sweets) (with some of the items geared for children) and life assurance (with some companies appealing more to the younger and upmarket here).  Even so, the results on the different categories are relatively uniform, with MADs of about 3, which may be considered small in the context of what might be expected if significant segmentation was occurring. Within the individual market, there is occasionally a somewhat higher MAD.  This mainly occurs for functionally different items, sub-markets in effect, as also in the earlier Hammond study.  An example is Johnson's Baby Soap in Table 2.  Expensive car makes (BMW's etc) is another. 

Even if the other six socio-demographic ‘segmentation variables’, which were examined, are taken into consideration much the same picture emerges. Thus, for two large markets, an extended set of variables is shown in Table 4 below:

[Table 4]

As can be seen from this example, which is typical of the results across the other 23 categories, even newspaper profiles - an important part of the media's marketing armoury - do not deviate markedly from the category average in terms of "used most often). 

            The Hammond finding that the deviations from the category were somewhat greater for smaller brands (i.e. smaller samples) is also observed in most of the markets here (as in Table 2 for toilet soaps, already shown); the average MAD's weighted by market-share being generally slightly smaller in all tables.

LIFESTYLE ATTITUDES

            TGI also record answers to 100 lifestyle attitudinal questions for each brand. Again, there were no marked deviations in the MADs, between the brands, across each of the (25) markets. Taking the market for washing powders, in Table 5  for example, the average MAD for the brands investigated was again just under 3.

[Table 5]

Nor did particular variables stand out; for example, for the typical selection of the attitudes across the brands shown in table 6.

[Table 6]

CONCLUSIONS

            The lack of brand segmentation found by Hammond et al has been fully supported by this substantial extension to 25 other varied product categories (including cars and financial services) and seven further potential socio-demographic segmentation-variables.

            As already stressed, this lack of segmentation arises with competitive brands.  In contrast, product categories as a whole, and who functionally differentiated subcategories like pack-sizes or price bands, can appeal to different segments of the population.  This contrast should explain the apparent conflict between our negative segmentation results here and the emphasis commonly given to the idea of segmentation.  Segmentation can apply to markets or submarkets as a whole, but not generally to competitive brands within a market.

            Other research work (Mercer, 1996) shows that "Segmentation" is sought by almost all marketing organizations. As reported by their managers, it is "formally" used by almost half of all the organizations surveyed (and by almost two thirds of the larger ones, with turnovers in excess of £200m) and informally used by another 50%.  It is supposedly seen as being one of the most useful frameworks, at least theoretically: a rating of 3.6 - on a scale of 1 ("Least useful") to 5 ("Most useful"). It is seen as being especially useful in terms of providing a customer focus.

            Any market as a whole is undoubtedly very heterogenous and usually has "segmented" subpatterns - groupings of relatively homogeneous consumers.  Thus, there are heavy and light buyers (of the category and/or of the brands), and usage patterns and attitudes can differ systematically. But in the area of brand competition - our focus here - segmentation may simply be yet another case of the Emperors' New Clothes.  We are for example not aware of any case of unambiguous and substantial brand segmentation that has ever been reported, let alone replicated.  Papers on segmentation and textbook discussions are invariably linked to possible analysis techniques for segmentation, not to any results.  Usually the author looks for a better technique, implying therefore that nothing much has by way of segmentation been found so far with the old techniques.

            Segmentation is generally talked about as being highly desirable, e.g. to allow efficient targeting.  But that is only so if the market is segmented, and strongly so.  In such a case it is, indeed, best to know the facts behind this (e.g. that motorists with older cars do not buy unleaded gasoline).  But the almost compulsive search for (brand) segmentation as the way to profitable marketing is misconceived:  Identifying as "your market" the 30% of consumer with warts (or whatever) means foregoing the remaining 70%.  That is nonsense unless the 70% "don't use dog food at all".

REFERENCES

Engel J, D. Kollat & R. Blackwall & (1978), Consumer Behaviour, Dryden Press

Hammond K, A S C Ehrenberg, & G J Goodhardt, (1993), “Brand Segmentation: A Systematic Study”, Marketing Education Group (MEG) Conference, July

-------(1995), “Market Segmentation for Competitive Brands”, Centre for Marketing Working Paper No 95-201, February

Jedidi, Kamel, S Harsharanjeet & Wayne S DeSarbo (1997), "Finite-Mixture Structural Equation Models For Response-Based Segmentation And Unobserved Heterogeneity", Marketing Science, Vol. 16 No. 1

Kotler, P (1967), Marketing Management (3^rd ed.), Prentice Hall

Krieger, Abba M, & Paul E Green (1996), "Modifying Cluster-Based Segments To Enhance Agreement With An Exogenous Response Variable", Journal of Marketing Research, Vol. 33 No. 3

Mercer, D (1996), "Management's Commitment to Marketing Theory Compared with Actual Practice”, Marketing Education Group (MEG) Conference, July

Ramaswamy, Venkatram, Rabikat Chatterjee & Steven H Cohen (1996), "Joint Segmentation On Distinct Interdependent Bases With Categorical Data", Journal of Marketing Research, Vol. 33 No. 3

Smith, Wendell R (1956), "Product Differentiation and Segmentation as Alternative Marketing Strategies", Journal of Marketing, Vol. 21 (July)

Wensley R (1995), "A Critical Review of Research in Marketing", British Journal of Management, 6 (Special Edition).

Wensley, R (1996), “Forms of Segmentation: Definitions and Empirical Evidence”, Marketing Education Group (MEG) Conference, July

Table 1 - Toilet Soap: Brand Usership Profiles - By Age

Table 2 - Brand Deviations by Age

Table 3 - Average MADs Across 25 Categories

Table 4 - Brand Deviations by Extended Variable Range             

Table 5 - Washing Powders by Brand across Attitude

BRAND                                                 MAD

                                                                (attitude

                                                                variables)

Ariel Automatic                                                   2

Ariel Liquid                                                          2

New System Persil

                Automatic                                             4

Original Non-Bio

                Persil Automatic                                 3

Persil Liquid                                                        2

Persil Concentrated                                           2

Table 6 - Washing Powders by (selected) Attitude Statements across Brands

[back]     [home]

hits