Kalpana Heliya
Par 100 posts (V.I.P)
Segmentation
The idea of segmenting markets is almost as old as marketing - though first formalised in the 1950s. It is intuitively and empirically clear that some people represent more fruitful targets than others, by virtue of their physical characteristics, economic or social status, education, habits, attitudes, values and mindset. The aim is to find large enough groups that are internally homogeneous and clearly different from each other, on dimensions that matter within a given market (1, 6, 16).
Marketing communication uses segmentation in three main ways:
1. to understand the structure of markets in terms of the benefits different people look for within a product category
2. to provide a market-based basis for media/channel selection, enabling effort to be focused as far as possible against the most attractive targets
3. to understand how different groups approach the brands in a market, in order to tailor the content of communications to the most promising targets.
Segmentation may be a priori - typically based on age, income, and so on - or post hoc, being derived from analysis of survey data (4, 6). The latter is increasingly dominant, as the need for deeper consumer insight has developed (4, 16).
Over time, segmentations have moved from relatively simple, demographically anchored systems, based on combinations of three or four basic criteria (age, sex, income/social grade) to a more elaborate range of possibilities. Further, researchers, particularly, have argued that segmentations should be ‘holistic’, operating across the range of possibilities, if they are to provide effective ways of discriminating between consumer types (4, 12).
Technically, segmentation studies require the collection of data from a population and the application of statistical analysis to split the population into ‘clusters’ of similar-seeming individuals. There are a number of different ways of doing this, of which cluster analysis and factor analysis are the most common. These techniques have both advantages and shortcomings, especially if used in combination, which are exhaustively argued over by statisticians (2, 4, 5, 6, 7).
For the marketer, who has to use the results, the key considerations are more practical than technical (2, 4, 6).
Does the segmentation generate clusters that fit what we think we understand about our market?
Does the clustering process provide me with new or valuable insights into the category and/or my brand?
Are the clusters ‘usable’, in terms of either creative approaches or media planning (preferably both)?
Are the clusters likely to be reasonably stable over time?
Can I apply the clusters internationally, or at least on a broad regional basis?
Can I apply the clusters to my existing customer databank?
This last takes the process into customer relationship management (7, 29). Existing customers are easier to influence than non-customers, and existing customers can be divided into groups in terms of their value to the company/brand. An effective customer database should contain a substantial range of information about existing (and lapsed) customers, which should provide a firm basis for segmentation and tailored communications (22, 23).
It then becomes possible to use the insights from this analysis to provide the basis for targeting efforts to gain new customers (without which any business will, eventually, die, because of the ‘leaky bucket’ phenomenon: a significant proportion even of apparently ‘loyal’ customers inevitably disappears, over quite a short timescale).
Segmentation criteria can be separated into four broadly different categories, in terms of their potential use: ‘media’ criteria, ‘category’ criteria, ‘brand’ criteria and ‘values’ criteria.
MEDIA CRITERIA
Media audience research routinely collects a range of demographic information, including geography, in reasonable detail - though the age breaks may not always fit those that marketers might wish, ideally, to use (18, 19).
Increasingly, too, media research examines people’s relationship with different media, so that it is possible to target viewers of particular TV programmes or channels, or readers of particular publications, on the basis that they are especially suited to the proposed communication for a particular brand (13, 14). An added dimension comes from data on ad avoidance - a substantial proportion of people are, or claim to be, ad avoiders, and this can vary by medium (17).
CATEGORY CRITERIA
Whether people buy the category, and, if so, whether they are heavy or light users (20), is a key variable. Further, there are likely to be a variety of different motivations for category purchase, which may, for example, be related to specific benefits (1, 3, 31), mood, need.state, social situation or time of day (2, 10, 28). Similarly, in more or less technical markets, there are clear divisions between geeks and others, and the well-known groups of adopters (30). Car (or house) purchase may be heavily influenced by lifestage, as are financial services (4, 21, 27).
BRAND CRITERIA
It has been argued that competing brands show no significant segmentation, either by demographic or attitudinal factors (9). Though this is broadly true (but not very helpful), marketers still need to identify market segments for targeting purposes, even if they are then targeting the same segment as their competitors. In practice, of course, they are usually trying to target more precisely, or to communicate a compelling brand difference - even if the same authors cast doubt on their ability to do so (see 5).
Brand criteria have, historically, been based either on physical product perceptions, pricing (25) or on brand imagery and attitudes to the brands in the market - in particular user imagery, since this ties back to consumer segmentation. It is at brand level, too, that benefit segmentation comes into play: here, a classic example is toothpaste (1, 3, 24).
A more incestuous form of brand criteria is provided by ‘shoppergraphics’, in which the target audience for a brand may be defined by users’ propensity to use particular retail outlets, or even other brands in related categories (12, 13).
VALUES CRITERIA
‘Psychographics’ covers a range of attitudinal criteria, of which potentially the most useful are ‘values’. These are the more or less permanent aspects of people’s view of life, which may be mapped fairly simply into, typically, around a dozen different groups, which distinguish inward and outward focus, sociability, openness to new ideas, and so on (8, 10, 26). A looser, and less obviously permanent, version of this is ‘lifestyle’, which is also widely used as a way of describing target segments as a creative stimulus (28).
DEVELOPING A SEGMENTATION
Segmentation studies need to collect a considerable range of data, in order to allow for the sorts of analysis that enable researchers to look at the full breadth of criteria indicated above. A typical project is well described by (15), and the main issues that can arise in planning a project are outlined by (2, 3).
A key factor in planning segmentation research is to be clear about objectives: what is the segmentation project for? How is it going to be used? Who is going to use it - in particular, does it need to apply internationally? Questions like these will influence the detail of the questionnaire, and the nature of the resulting segmentation (10).
A segmentation study typically involves the following stages.
1. A consumer survey, collecting product/ brand usage data, perhaps media usage, and a range of rating scales, which are the basis for the analysis.
2. Factor analysis of the rating scales to develop a small(ish) group of underlying dimensions in the data. This requires some creativity on the part of the analyst to arrive at a ‘best’ solution. Factor scores are attributed to each respondent.
3. Respondents are clustered, using cluster analysis, on the basis of their factor scores.
4. The clusters are profiled, using the rest of the survey data (3).
This procedure is common, but highly suspect technically, and a number of researchers have proposed ways to improve it (2, 3). Apart from technical failings, a common criticism is that clusters are unusable or uninformative. Part of the fault appears to lie in the way scales operate - especially on international studies - and a technique called maximum difference scaling has been proposed to solve this. Similarly, rather than straightforward clustering, a procedure called latent class analysis can improve the clarity of the clusters (2, 3, 5, 6).
Once the clusters have been identified, it is common practice to give them distinctive names (2, 14). This is often criticised, but can be seen as an aid to understanding, and helpful to creative people who have to target communications to specific groups. Systems like Claritas’s PRIZM divide whole populations into groups with names like ‘Affluentials’ (11). The practical reason for doing this is to render the segments both recognisable, throughout the company, and hencemore usable - and used. Nothing is more useless than an expensive segmentation study gathering dust in a cupboard. Segmentation is at the heart of marketing. It seems sure to remain there, even if the technicalities are less than perfect.
:SugarwareZ-296:
The idea of segmenting markets is almost as old as marketing - though first formalised in the 1950s. It is intuitively and empirically clear that some people represent more fruitful targets than others, by virtue of their physical characteristics, economic or social status, education, habits, attitudes, values and mindset. The aim is to find large enough groups that are internally homogeneous and clearly different from each other, on dimensions that matter within a given market (1, 6, 16).
Marketing communication uses segmentation in three main ways:
1. to understand the structure of markets in terms of the benefits different people look for within a product category
2. to provide a market-based basis for media/channel selection, enabling effort to be focused as far as possible against the most attractive targets
3. to understand how different groups approach the brands in a market, in order to tailor the content of communications to the most promising targets.
Segmentation may be a priori - typically based on age, income, and so on - or post hoc, being derived from analysis of survey data (4, 6). The latter is increasingly dominant, as the need for deeper consumer insight has developed (4, 16).
Over time, segmentations have moved from relatively simple, demographically anchored systems, based on combinations of three or four basic criteria (age, sex, income/social grade) to a more elaborate range of possibilities. Further, researchers, particularly, have argued that segmentations should be ‘holistic’, operating across the range of possibilities, if they are to provide effective ways of discriminating between consumer types (4, 12).
Technically, segmentation studies require the collection of data from a population and the application of statistical analysis to split the population into ‘clusters’ of similar-seeming individuals. There are a number of different ways of doing this, of which cluster analysis and factor analysis are the most common. These techniques have both advantages and shortcomings, especially if used in combination, which are exhaustively argued over by statisticians (2, 4, 5, 6, 7).
For the marketer, who has to use the results, the key considerations are more practical than technical (2, 4, 6).
Does the segmentation generate clusters that fit what we think we understand about our market?
Does the clustering process provide me with new or valuable insights into the category and/or my brand?
Are the clusters ‘usable’, in terms of either creative approaches or media planning (preferably both)?
Are the clusters likely to be reasonably stable over time?
Can I apply the clusters internationally, or at least on a broad regional basis?
Can I apply the clusters to my existing customer databank?
This last takes the process into customer relationship management (7, 29). Existing customers are easier to influence than non-customers, and existing customers can be divided into groups in terms of their value to the company/brand. An effective customer database should contain a substantial range of information about existing (and lapsed) customers, which should provide a firm basis for segmentation and tailored communications (22, 23).
It then becomes possible to use the insights from this analysis to provide the basis for targeting efforts to gain new customers (without which any business will, eventually, die, because of the ‘leaky bucket’ phenomenon: a significant proportion even of apparently ‘loyal’ customers inevitably disappears, over quite a short timescale).
Segmentation criteria can be separated into four broadly different categories, in terms of their potential use: ‘media’ criteria, ‘category’ criteria, ‘brand’ criteria and ‘values’ criteria.
MEDIA CRITERIA
Media audience research routinely collects a range of demographic information, including geography, in reasonable detail - though the age breaks may not always fit those that marketers might wish, ideally, to use (18, 19).
Increasingly, too, media research examines people’s relationship with different media, so that it is possible to target viewers of particular TV programmes or channels, or readers of particular publications, on the basis that they are especially suited to the proposed communication for a particular brand (13, 14). An added dimension comes from data on ad avoidance - a substantial proportion of people are, or claim to be, ad avoiders, and this can vary by medium (17).
CATEGORY CRITERIA
Whether people buy the category, and, if so, whether they are heavy or light users (20), is a key variable. Further, there are likely to be a variety of different motivations for category purchase, which may, for example, be related to specific benefits (1, 3, 31), mood, need.state, social situation or time of day (2, 10, 28). Similarly, in more or less technical markets, there are clear divisions between geeks and others, and the well-known groups of adopters (30). Car (or house) purchase may be heavily influenced by lifestage, as are financial services (4, 21, 27).
BRAND CRITERIA
It has been argued that competing brands show no significant segmentation, either by demographic or attitudinal factors (9). Though this is broadly true (but not very helpful), marketers still need to identify market segments for targeting purposes, even if they are then targeting the same segment as their competitors. In practice, of course, they are usually trying to target more precisely, or to communicate a compelling brand difference - even if the same authors cast doubt on their ability to do so (see 5).
Brand criteria have, historically, been based either on physical product perceptions, pricing (25) or on brand imagery and attitudes to the brands in the market - in particular user imagery, since this ties back to consumer segmentation. It is at brand level, too, that benefit segmentation comes into play: here, a classic example is toothpaste (1, 3, 24).
A more incestuous form of brand criteria is provided by ‘shoppergraphics’, in which the target audience for a brand may be defined by users’ propensity to use particular retail outlets, or even other brands in related categories (12, 13).
VALUES CRITERIA
‘Psychographics’ covers a range of attitudinal criteria, of which potentially the most useful are ‘values’. These are the more or less permanent aspects of people’s view of life, which may be mapped fairly simply into, typically, around a dozen different groups, which distinguish inward and outward focus, sociability, openness to new ideas, and so on (8, 10, 26). A looser, and less obviously permanent, version of this is ‘lifestyle’, which is also widely used as a way of describing target segments as a creative stimulus (28).
DEVELOPING A SEGMENTATION
Segmentation studies need to collect a considerable range of data, in order to allow for the sorts of analysis that enable researchers to look at the full breadth of criteria indicated above. A typical project is well described by (15), and the main issues that can arise in planning a project are outlined by (2, 3).
A key factor in planning segmentation research is to be clear about objectives: what is the segmentation project for? How is it going to be used? Who is going to use it - in particular, does it need to apply internationally? Questions like these will influence the detail of the questionnaire, and the nature of the resulting segmentation (10).
A segmentation study typically involves the following stages.
1. A consumer survey, collecting product/ brand usage data, perhaps media usage, and a range of rating scales, which are the basis for the analysis.
2. Factor analysis of the rating scales to develop a small(ish) group of underlying dimensions in the data. This requires some creativity on the part of the analyst to arrive at a ‘best’ solution. Factor scores are attributed to each respondent.
3. Respondents are clustered, using cluster analysis, on the basis of their factor scores.
4. The clusters are profiled, using the rest of the survey data (3).
This procedure is common, but highly suspect technically, and a number of researchers have proposed ways to improve it (2, 3). Apart from technical failings, a common criticism is that clusters are unusable or uninformative. Part of the fault appears to lie in the way scales operate - especially on international studies - and a technique called maximum difference scaling has been proposed to solve this. Similarly, rather than straightforward clustering, a procedure called latent class analysis can improve the clarity of the clusters (2, 3, 5, 6).
Once the clusters have been identified, it is common practice to give them distinctive names (2, 14). This is often criticised, but can be seen as an aid to understanding, and helpful to creative people who have to target communications to specific groups. Systems like Claritas’s PRIZM divide whole populations into groups with names like ‘Affluentials’ (11). The practical reason for doing this is to render the segments both recognisable, throughout the company, and hencemore usable - and used. Nothing is more useless than an expensive segmentation study gathering dust in a cupboard. Segmentation is at the heart of marketing. It seems sure to remain there, even if the technicalities are less than perfect.
:SugarwareZ-296: