The purpose of segmentation is to provide marketers with an understanding of customers and prospects that can translate into improved marketing effectiveness. There is no single approach to consumer segmentation that represents a holy grail for marketers.
To understand segmentation today, it is helpful to be familiar with some of the early attempts at segmentation. In ancient times, the Romans realized that groups of conquered subjects could become better fighting units if they were grouped into units of similar ethnic and geographic origins. Those subjects of Roman legions were called cohorts, and they represent one of the first recorded examples of segmentation of humans into relatively homogenous groups. Religion, ethnicity and political outlook are among the variables that have been used throughout history as a basis of segmentation.
In more recent times, marketers exploring avenues for improved effectiveness are likely to use one of three types of variables in identifying similar consumer segments:
• Geographic or household-specific demographic and lifestyle variables.
• Attitudinal variables.
• Behavioral or transactional characteristics.
The simple premise that similar households will exhibit similar consumer behaviors underlies the use of each variable. To illustrate how the different types of data are used in segmentation, let’s review some of the market segmentation systems built from these types of data.
Two early, well-known consumer segmentation systems of interest to marketers are still around today. They are PRIZM, created by Claritas, San Diego, and built from geodemographic variables, and VALS, created by SRI, Menlo Park, CA, and built from attitudinal variables.
Although both attempt to define homogenous groups of consumers, they do so in very different ways. PRIZM is built on the assumption that similar households tend to group together naturally by geography.
Using U.S. Census neighborhood-level data, PRIZM assigns all households in each neighborhood into a neighborhood group that reflects the neighborhood’s mean or median demographic characteristics, such as median family income.
VALS, on the other hand, is based on the premise that consumers with similar attitudes or psychographic characteristics will exhibit similar types of consumer behavior. Consumers are placed into VALS clusters on the basis of their responses to a battery of questions about their attitudes toward risk, status and other attitudinal indicators.
Both PRIZM and VALS are commercially available segmentation systems built from data independent of a marketer’s transactional customer data. Another common approach to segmentation is to build custom segments that incorporate historical consumer purchase information.
These approaches are based on the belief that past behavior is a critical determinant of future consumer behavior. Custom models can be developed by outside vendors or an inhouse statistical team. Often these models will incorporate additional outside data, such as demographics, as a further refinement and to help describe the resultant segments in terms more meaningful to marketers.
For database marketers, the transaction-based custom segments generally do better in tactical marketing applications such as rank-ordering households based on their likelihood to respond to a given offer. However, they tell the marketer little or nothing about who the consumer actually is. That means that they do not allow the marketer to create a versioned or relevant message; they simply rank all prospects in order of likelihood of responding to a particular offer.
The geodemographic-based approaches tend to have more strategic marketing applications because they describe consumers in terms more relevant to the development of broad-based marketing strategy. They are most useful in applications like broad market assessments, such as comparing one retail trade area to another. However, since they group all households in a given neighborhood into the same group, they fail to recognize the diversity in today’s neighborhoods. Consequently, their use in database marketing results in highly inaccurate targeting.
The attitudinal approaches tend to produce neither a clear picture of a consumer, nor a precise means for rank-ordering prospects, but they do present a useful framework for discussing the motivations behind consumer behavior. Unfortunately, acquiring the necessary data to classify consumers attitudinally requires a cumbersome survey process that, in practice, limits their applicability to research.
Is there a panacea? No, but with some effort, a sophisticated marketer can get close. The ideal segmentation system should:
• Use individual or household level (as opposed to geographically defined) demographic and lifestyle data to describe consumer segments as real people.
• Accurately represent key attitudes and psychographics of each consumer segment.
• Be linked to actual customer transaction histories.
Such an approach would incorporate the benefits of each data type while overcoming the inherent limitations of each. Since the information necessary to accomplish this isn’t commercially available, the marketer must get creative.
By carefully combining available consumer data with well-designed market research, the consumer marketer can realize most of the benefits of the ideal segmentation system. To demonstrate how this can be done, let’s discuss a hypothetical national consumer marketer with a transactional customer database. Initially, the marketer knows nothing more about its customers than what each has purchased in the past.
There are four steps in the development of the ideal segmentation system.
• Purchase and append demographic and lifestyle data to the marketer’s house file. Available from companies such as Equifax, Experian or Acxiom, this data will be used in a rigorous cluster analysis to define key customer segments. By using demographics and lifestyle data to define segments, the marketer can create a useful, descriptive profile of each segment. These segments also can be used to select “look-a-likes” for e-mail or direct mail prospecting.
• Because the segments are demographically defined, they can be linked to commercially available syndicated market research information. This information, available from research companies such as Scarborough Research, Simmons Market Research and MRI, will provide a broad and robust picture of consumer behavior by segment. Knowing its media usage patterns, for example, the marketer can develop the optimal media strategy to reach each key customer segment.
• Historical customer transaction data are then analyzed by segment to understand the buying patterns of each key segment. Typically, this includes RFM analyses, as well as analyses of lifetime value, distribution channel preferences and seasonal habits.
• Finally, the four to six key segments that account for most of the marketer’s business are surveyed to understand the attitudinal factors, such as the primary factors motivating purchase for each key customer segment.
Now the marketer is armed with all the information needed to use consumer segmentation to optimize marketing effectiveness. The marketer understands the demographic and lifestyle characteristics of each key customer segment. It has the complete picture of the broader consumer behavior patterns of each segment necessary to develop and implement a strategic, well-targeted marketing campaign across media. It knows the segments that represent the greatest opportunity for future growth by product line, service type and channel of distribution. It can develop statistical models using transaction histories by segment to rank-order customers by likelihood of repeat purchase. Finally, it has determined the key “hot buttons” by segment for message and offer development.
Through careful selection of commercially available consumer data, rigorous data analyses, and the selected use of primary and secondary market research, the marketer can develop an optimal consumer segmentation system. It will have a system supporting the development of a strategic marketing plan that can be easily communicated within its organization and practically applied to the entire spectrum of tactical marketing programs supporting customer acquisition, cross-sell and retention.
• Scott D. Schroeder is chief operating officer at Looking Glass Inc., Denver, a strategic marketing company. He can be reached at [email protected]