by Jerry W. Thomas
When the term “market segmentation” is used, most of us immediately think of psychographics, lifestyles, values, behaviors, and multivariate cluster analysis routines.Market segmentation is a much broader concept, however, and it pervades the practice of business throughout the world.
What is market segmentation? At its most basic level, the term “market segmentation” refers to subdividing a market along some commonality, similarity, or kinship. That is, the members of a market segment share something in common.
The purpose of segmentation is the concentration of marketing energy and force on the subdivision (or the market segment) to gain a competitive advantage within the segment. It’s analogous to the military principle of “concentration of force” to overwhelm an enemy. Concentration of marketing energy (or force) is the essence of all marketing strategy, and market segmentation is the conceptual tool to help achieve this focus. Before discussing psychographic or lifestyle segmentation (which is what most of us mean when using the term “segmentation”), let’s review other types of market segmentation. Our focus is on consumer markets rather than business markets, but most of the following concepts also apply to B2B.
This is perhaps the most common form of market segmentation, wherein companies segment the market by attacking a restricted geographic area. For example, corporations may choose to market their brands in certain countries, but not in others. A brand could be sold only in one market, one state, or one region of the United States. Many restaurant chains focus on a limited geographic area to achieve concentration of force. Regional differences in consumer preferences exist, and this often provides a basis for geographic specialization. For example, a company might choose to market its redeye gravy only in the southeastern U.S. Likewise, a picante sauce might concentrate its distribution and advertising in the Southwest. A chainsaw company might only market its products in areas with forests. Geographic segmentation can take many forms (urban versus rural, north versus south, seacoasts versus interior, warm areas versus cold, high-humidity areas versus dry areas, high-elevation versus low-levation areas, and so on). These examples also reveal that geographic segmentation is sometimes a surrogate for (or a means to) other types of segmentation.
Different markets can be reached through different channels of distribution. For example, a company might segment the “tick and flea collar” market by selling the product to supermarkets under one brand name, to mass merchandisers under another brand name, to pet stores under another brand name, and to veterinarians under yet another brand name. This type of distributional segmentation is common, especially among small companies that grant each channel a unique brand to gain distribution within that channel. Other examples of distributional segmentation would be an upscale line of clothing sold only in expensive department stores, or a luxury hair shampoo sold only through upscale beauty salons.
While not common, media segmentation is sometimes a possibility. It is based on the fact that different media tend to reach different audiences. If a brand pours all of its budget into one media, it can possibly dominate the segment of the market that listens to that radio station or reads that magazine. Media segmentation is most often practiced by companies that have some control over the media and can somehow discourage competitors from using that media.
Price segmentation is common and widely practiced. Variation in household incomes creates an opportunity for segmenting some markets along a price dimension. If personal incomes range from low to high, the reasoning goes, then a company should offer some cheap products, some medium-priced ones, and some expensive ones. This type of price segmentation is well illustrated by the range of automotive brands marketed by General Motors, historically. Chevrolet, Pontiac, Oldsmobile, Buick, and Cadillac varied in price (and status) along a clearly defined spectrum to appeal to successively higher income groups.
Gender, age, income, housing type, and education level are common demographic variables. Some brands are targeted only to women, others only to men. Music streaming services tend to be targeted to the young, while hearing aids are targeted to the elderly. Education levels often define market segments. For instance, private elementary schools might define their target market as highly educated households containing women of childbearing age. Demographic segmentation almost always plays some role in a segmentation strategy.
Time segmentation is less common, but can be highly effective. Some stores stay open later than others, or stay open on weekends. Some products are sold only at certain times of the year (e.g., Christmas cards, fireworks). Chili is marketed more aggressively in the fall, with the onset of cooler weather. Football is played in the fall, basketball in the winter and spring, and baseball in the spring and summer (or at least this used to be the pattern). The Olympics come along every four years. Department stores sometimes schedule midnight promotional events. The time dimension can be an interesting basis for segmentation.
People tend to behave differently, and think differently, at different times or occasions. For example, dietary habits and preferences vary by occasion: breakfast is different from dinner; eating out on a Friday night is different from grabbing lunch during the week; Thanksgiving dinner is different from most other dinners. These types of differences can be the basis for segmenting a market. If the goal is to develop new product development templates for a restaurant chain, then occasion-based segmentation might be a good solution. If the goal, however, is to develop the strategic positioning and advertising messages for a new smartphone or a new car, then occasion-based segmentation would not be applicable. In these cases, the goal is one optimal solution, and the occasions do not matter.
Markets can be also segmented by hobbies, by political affiliation, by religion, by special interest groups, by sports team loyalties, by university attended, and by hundreds of other variables. You are only limited by your marketing imagination.
Psychographic or Lifestyle Segmentation
Psychographic (or lifestyle) segmentation is based upon multivariate analyses of consumer attitudes, values, behaviors, emotions, perceptions, beliefs, needs, benefits, wishes, and interests. Psychographic segmentation is a legitimate way to segment a market, if we can identify the proper segmentation variables (or lifestyle statements, words, pictures, etc.).
Qualitative research techniques (focus groups, depth interviews, ethnography) become invaluable at this stage. Qualitative research provides the insight, the conceptual knowledge, and the consumer’s exact language necessary to design the segmentation questionnaire. Typically, verbatim comments from consumers are used to build batteries of psychographic or lifestyle statements (these two terms are used interchangeably). A large representative sample of consumers (generally, 1,000 or more) are then asked about the degree to which they agree or disagree with each statement.
For example, if you were designing a market segmentation questionnaire for an airline, you might conduct a series of depth interviews to help design the questionnaire. You probably would include a behavioral section (frequency of flying, how purchase tickets, who travel with, cities flown to, where sit, airlines flown, money spent on airline tickets, etc.). You would include a major section on attitudes toward air travel (motivations for air travel, fears related to air travel, positive emotions of flying, attitudes about airline employees, checking luggage, buying tickets, and so forth). You would also want to include a section on perceptions of the different airlines; that is, their “brand images.” You could go further and add a section on media consumption or personal values as well. It is at this point that you realize the questionnaire is too long, and you have to make some hard decisions about what questions or statements to include.
The method of data collection is very important, because the questionnaire is so long (often 45 to 60 minutes in length). The telephone is not recommended for segmentation studies because of questionnaire length. Moreover, the various rating scales and attitudinal statements are difficult to communicate by phone, and the resulting phone data tends to be “insensitive” and rife with “noise.”
In-person interviews, online surveys (or even mail surveys) are much better. Rating scales and attitudinal statements can be seen and fully comprehended by respondents. Seeing is much better than hearing, and it produces more accurate answers. Online surveys are especially valuable for segmentation studies, since respondents can take the survey at a time of their own choosing when they can give it their full, undivided attention. A mail survey offers some of the same advantages, but without the questionnaire controls, checks, and safeguards built into an online survey.
Most segmentation analyses are based upon various types of “cluster analysis,” which is a set of well-defined statistical procedures that group people according to the proximity of their ratings. Unfortunately, cluster analysis (regardless of its many types and forms) has inherent limitations and seldom yields coherent market segments. Cluster analysis routines tend to ignore the pattern of respondent ratings and rely primarily upon the proximity of respondent ratings. Too often this leads to clusters, or market segments, that don’t seem to make much sense when cross-tabulated against the original segmentation variables. Another limitation of clustering approaches is that all statements are treated as equal, whereas, in truth, some statements might be much more important than others in explaining consumer behavior in a particular product category.
A better way to achieve a good psychographic segmentation is to first identify the statements that are more important (i.e., the statements that tend to explain or cause specific consumer behaviors). Correlation analysis and regression can be used for this purpose. Factor analysis is also a powerful technique to identify the statements and groups of statements that account for much of the variance in the attitudinal data set. Directly, and indirectly, these techniques can help you identify the most important statements (i.e., attitudes, perceptions, values). Then these statements become the inputs to the final segmentation analysis. Many different methods can be used to “cluster” or group the statements at this point.
The final step is to attach a segment code to each market segment identified and then cross-tab all of the questionnaire variables by the segments. You must then study the segments, and the attitudes/statements that make up each segment, to make sure they make sense and hang together. If the segmentation results don’t make sense, then you have to go back, change some of your assumptions or methods, rerun the analysis, and repeat the cross-tab exercise to apply the “common sense” validity check.
Segmentation studies tend to be large and complicated, so it’s easy for errors and mistakes to be made. Some of the most common mistakes:
Segmenting a segment. For example, someone might want to segment the market for widgets among 18- to 24-year-olds who live in Vermont and buy brand XYZ. As is evident, the client is asking that a tiny sliver of the market be segmented. True, this tiny sliver can be segmented, but rarely are the resulting segments of any value, because they are just too small. General rule: segment the whole market, including all age groups. The market should be broadly defined for a segmentation analysis to be most effective. In other words, don’t preordain the results by sampling restrictions.
Overlooking the “universals.” Many attitudinal statements in the questionnaire will not show up in the final segments, because they tend to be the same across all segments. Statements that everyone agrees with or everyone disagrees with (we call them “universals”) cannot explain much in the multivariate analyses. Variables have to move up and down for the multivariate analysis to work. The highest-rated variables, and the lowest-rated, are likely to fall out of the multivariate equations. However, you should always look at these universal statements. Any one of them might be the basis for a positioning or a strategy that would appeal to everyone. If you find something unique that appeals to everyone, the heck with segmentation. Go for the whole hog.
Creating too many segments. There is a practical limit to the size of segments that companies can effectively target. If you create more than four or five market segments, you run the risk that the resulting segments will be too small to target, at least by mass media. This is not always true, but it is a good rule of thumb.
Targeting all segments. So you have carefully subdivided your target market into five mutually exclusive psychographic segments, and your boss tells you to develop a marketing plan to attack each segment. If all of your marketing is direct mail, and you can identify the addresses that belong to each segment, then you can attack all segments (assuming your product is relevant to all segments). But if you use broadcast media in marketing your product, it is very difficult to target multiple segments because of media “spillover.” What you say to one segment will be muddled and confused by the different messages targeted to other segments.
Confusing the results. Segmentation studies are large and complicated, with enormous amounts of data. It is easy to get lost in this treasure trove of answers and come up with confusing and baffling results.
Overlooking the basics. The dazzle and glitter of the advanced, rocket-science multivariate analyses attract everyone’s attention. No one ever opens up the cross-tabs and looks at the answers to the hundreds of questions asked. Often, hidden in plain view in the cross-tabs are tremendous findings that could form the basis for new or improved marketing strategies, advertising campaigns, or new products. Rarely does anyone analyze this basic data, however.
Targeting people instead of dollars. A market segment might represent a large percentage of the population, but a small part of the market. Always look at the dollar potential of market segments, not just the number of people in the segments.
Nonmutually Exclusive Segments
Virtually all segmentation work, historically, has been based upon the assumption of mutually exclusive market segments. The mutually exclusive model, however, does not always apply to psychographic or lifestyle segmentation (since most of us hold many overlapping and/or conflicting beliefs and attitudes). Therefore, it is wise to develop two distinctly different segmentation solutions: one based upon mutually exclusive segments and one based upon overlapping segments. Both of these segmentation “solutions” should be cross-tabulated by the original questionnaire variables to identify which type of solution yields the most meaningful (and actionable) market segments.
The concept of market segmentation is sound. It’s a way to apply greater marketing energy or force to a subset of the market. A great deal of money is wasted on psychographic segmentations that never lead to any marketing actions.
If you segment the market by psychographics, there are several essential uses of the segmentation: first, target your brand to the largest segment with relevant brand fit (or even target two closely related segments) by media advertising and message. That is, the advertising message is the way to reach the psychographic segment (rarely can a psychographic segment be defined by demographics or geography). Second, segmentation can provide the guiderails for brand positioning. That is, positioning assumes, or takes place in relation to, a target market segment; you are positioning your brand in relation to a market segment. Third, the segmentation can define opportunities for new products targeted to each psychographic segment. That is, the market segments can be a template for new product development. For example, if you find that 15% of the U.S. population belongs to a “safety first” segment when it comes to buying cars, then you can design and build the safest car in the world to target this segment. So psychographic segmentation’s greatest value lies in positioning, targeting via advertising message, and defining new product opportunities.
Direct Marketing Segmentation
In categories where direct marketing (targeted direct mail, for example) is the norm, the number of usable market segments can be large, as many as 10 to 15 segments (in contrast, products supported by broad-reach media advertising can only target a limited number of segments, rarely more than 2 or 3). Direct marketing (especially direct mail) can target 10 to 15 different market segments with different positionings and messages. So, segmentation applied to direct marketing categories follows a different set of rules. The challenge is linking up household characteristics and variables that reside in secondary population databases (let’s use Experian data as an example) with the survey-based segmentation data. Typically, the segmentation survey records would have Experian-household variables appended. Then, once the segmentation is complete, the appended Experian household data can be used to build predictive models to identify market segment membership. These models (called “typing tools”) can then be applied to U.S. household databases to identify the market segment each household falls into. Direct mail advertising can then be precisely targeted to each market segment.
The “typing tool” is a predictive model to help determine the market segment that a household or individual belongs to. In the instance of a mass market attacked by broad-reach media advertising, the typing tool would be based on a set of 4 to 10 short questions that could be used in surveys or CRM interactions to categorize consumers into applicable market segments. In direct marketing categories, the typing tool would be based on variables in the Experian (to continue the example) database, such as household income, value of home, presence of swimming pool, size of yard, number of cars owned, and so forth, to predict market segment memberships for each household in the database.
Artificial Intelligence and Machine Learning
Most segmentations are based on clustering techniques, factor analyses, and choice modeling experiments. Much experimental and exploratory work is in progress to apply artificial intelligence and machine learning techniques to improve segmentation outcomes. Early results are promising, but much work lies ahead before these newer methods can be fully trusted.
Segmentation is one of the most powerful concepts in the marketing toolbox. It’s a chance to apply maximum pressure by concentrating marketing and advertising activities on a segment of the market to change human behavior; for example, persuade people to accept a new product, buy brand A over brand B, accept new taxes to protect the environment, or elect a new member of Congress. Segmentation permits intelligent focusing and concentration of marketing effort to maximize returns on marketing investments. Go forth and segment.
About the Author
Jerry W. Thomas (firstname.lastname@example.org) is President/CEO of Dallas-Fort Worth based Decision Analyst. He may be reached at 1-800-262-5974 or 1-817-640-6166.
Copyright © 2019 by Decision Analyst, Inc.
This article may not be copied, published, or used in any way without written permission of Decision Analyst.