Articles & Papers

Multidimensional Segmentation

William D. NealCJMR (Canadian Journal of Marketing Research)


In segmenting markets most researchers use a single set of basis variables that have ranged from demographics to psychographics to product category related attitudes to product related behaviors to derived importances from conjoint exercises to latent structures, depending on the era and the proclivities of the researcher. Yet, it seems quite restrictive to limit the basis for segmentation to only one type of variable. Buyers may use many criteria for determining their response to the selling proposition in a category. These criteria are multidimensional in nature, possibly encompassing attitudes, needs, values, benefits, means, occasions, and prior experiences, depending on the product/service category and the buyer.

Furthermore, a segmentation scheme based on only one set of basis variables may limit the utility of the segmentation information to the firm. The various users of market segmentation information have quite different needs. For example, managers of the product development function may want to have the market segmented on perceived values and benefits being sought. Conversely, the folks in marketing communications may want to see the market segmented into groups of buyers who have similar needs, desires or psychographic profiles. And, sales managers would like to see the market segmented on sales potential or profitability.

A segmentation scheme based on multiple bases, using separate segmentation schemes for each dimension is often more useful and more flexible for planning marketing strategy and for executing marketing tactics. Thus, one may consider several different segmentations on a sample of buyers using different bases, say, stated needs, benefits, and amount spent in the category.

In the past, such segmentation schemes were confusing and produced far too many segments for marketing managers to address effectively. Yet, as we approach the era of micro-niche marketing and as we leverage the emerging tools of direct marketing, market planners need to seriously consider market segmentation schemes that support far finer targeting efforts.

Over the last couple of years a few research and consulting firms have developed and refined methodologies for executing and reporting such multidimensional segmentation schemes, which generate a large number of very homogeneous segment cells. Often these cells are then dynamically aggregated into master segments for strategic planning. Thus, we may have one static set of strategic segments and many different segmentation schemes for various tactical marketing initiatives, all inter-related.

One may visualize a three-dimensional segmentation scheme as shown in figure 1.

multichart.jpg - 40859 Bytes

In this model, one would perform three separate segmentation schemes, each with a different set of basis variables. For example, the X-axis may be a "needs-benefits" segmentation based on derived importances from a conjoint exercise. The Y-axis may be a segmentation scheme based on buyers' perceived price range or ability to buy. And, the Z-axis may be based on clustering the responses to a battery of questions dealing with customer priorities and desires.

Each surveyed customer is now a member of one segment in each of the three segmentation schemes, and is uniquely assigned to a single cell in the segmentation matrix. Done well, respondents in each cell of the segmentation scheme are very similar on all three dimensions and decidedly different from respondents in other cells on at least one set of basis variables.

It is not unusual for these types of segmentation schemes to have 200 or more cells. Obviously quite large sample sizes are needed for this approach to produce reliable statistics at the cell level. Even with very large sample sizes, many cells may be sparsely populated, but that is very acceptable because it indicates that certain target markets may be too small to be profitable and should be either aggregated with other market segments or dropped from consideration as a specific target market.

Alone, this segmentation approach provides great insight into the structure of the marketplace. However, each cell of the segmentation scheme, along with means and distributions of all descriptor variables can be put into a small database and manipulated to provide a more dynamic understanding of market structure and allow the user to re-form the cells into new segmentation schemes. A well-designed segment manager program allows the user to dynamically aggregate cells into specific market segments based on the varying needs of different internal functional and departmental users while using a common base of homogeneous cells for all of the segmentation schemes within the company. Thus, any specific tactical segmentation scheme can be directly liked to the strategic segments, or to any other tactical scheme.

As an aside, we have learned that using a database of all respondents, tagged to their various segmentation assignments provides better information and, in the long run is easier to aggregate into new segmentation schemes. However, the software needed to handle that process requires a much higher level of sophistication.


A better understanding of the concept and utility of multidimensional segmentation can be demonstrated by an example of the technique.

A regional business services company, selling a basic set of relatively undifferentiated services and products in a highly regulated environment undertook a major multidimensional market segmentation in order to prepare for impending deregulation and open competition.

The major goals of the project were:

  1. Segment the market into groups that would reveal marketing opportunities and allow the firm to develop both a defensive and offensive targeting strategy.
  2. Determine market segments that can/must be retained in the current service territory to assure profitability the defensive strategy.
  3. Determine the market segments that should be targeted outside of the firm's current service territory the offensive strategy.
  4. Given the target markets, determine the product/service bundles that will need to be offered to maximize segment penetration and segment dominance.

Prior to finalizing the design of the research project, several interviews with senior managers and expected users were undertaken to assess their understanding of a segmentation strategy and to make sure it was understood what the firm expected to get from the segmentation research. As part of this process it was found that several different functional departments within the firm needed a segmentation scheme that would address their particular concerns. Market planners, product planners, customer service, and marketing communications were particularly concerned that the segmentation strategy would give them specific direction on how best to address their customers and fulfill their mission.

A three-dimensional segmentation approach was recommended, accepted and executed. The three dimensions were:

  1. Customer values based on the derived importances from an adaptive conjoint exercise. This yielded eight "values" segments.
  2. Energy expenses based on a classification of customers into five expenditure groups, a priori.
  3. Customer energy management priorities based on clustering respondents based on their answers to a battery of questions dealing with stated needs and priorities for dealing with the service. This yielded seven "priorities" segments.

Upon completion of the research, all customers in the firm's customer information file were scored on all three dimensions and by master segment using neural networks. Thus, each customer was assigned to one of 280 unique segmentation cells and to one of seven master segments.

Also, detailed briefings and workshops were prepared for various workgroups and functional departments within the firm. These workshops discussed the results in detail, discussed specific uses of the segmentation schemes, and introduced the users to the two software tools delivered with the project a segment manager program (for interactively forming tactical segmentation schemes) and a market response simulator developed from the conjoint exercise in the survey.

Primarily, the segmentation project was used as follows:

  1. It provided a unified, common view of the marketplace and in the process it debunked several customer myths that had permeated the firm for many years.
  2. It allowed the senior management team to develop a fact-based strategy for executing both defensive and offensive operations.
  3. It helped the firm identify four broad master segments that they would target from a defensive perspective.
  4. It helped identify two master segments that provided the best opportunity for sales expansion outside the firm's service territory.
  5. It clearly demonstrated the product/service bundles that would optimize penetration of targeted segments.
  6. It gave clear direction to new product development and in the process killed several new product development efforts that were underway because they either did not address the needs and values of the targeted segments or because there was be little estimated demand for the service based on the results of the market response simulator.
  7. It helped initiate a re-branding initiative.
  8. It gave specific direction to the development of detailed communications plans for each targeted segment.

After the research and reporting of results were completed, the sponsoring firm initiated a re-design of their customer satisfaction survey to capture key drivers of value and to monitor changes in segment size and needs.

Some cautions and recommendations in using Multidimensional Segmentation

As in all market segmentation efforts, senior management must be involved at the onset and a strategic decision is required to segment a market. A strategy of market segmentation must permeate the entire organization. The firm's marketing organization must be able to execute alternative marketing strategies and vary pricing, promotion, and distribution systems. R&D must be able to execute product variations and manufacturing must be able to produce those variations in order to meet the varying needs and values of customers in different segments. Finance must be able to report costs, margins, and profits by market segment, and marketing research must be able to monitor and measure purchaser response and provide feedback to the organization by market segment.

Use pre-study workshops to discuss the approach being proposed, to explore possible applications of the segmentation information, reach agreement on deliverables and reporting formats, and to solidify management buy-in to the strategy.

Senior managers, if they have used a segmentation strategy in the past, have usually dealt with very simplistic, unidimensional segmentation schemes. The introduction of a multidimensional approach can be disconcerting to many. Emphasize the flexibility of the approach and the utility of the outcomes to the various functional departments within the firm. Again, management briefings and departmental workshops go a long way toward obtaining interdepartmental support.

In the data analysis phase of the segmentation study, be sure to allow sufficient time for detailed exploration of data structures, for exploring alternative segmentation schemes, and for validating the results. Ultimately, the selection of the final segmentation solution is judgmental, but external validation is essential.

Finally, post segmentation workshops are extremely helpful in broadly communicating what was learned in the process, explaining the deliverables and their operational utility to the various departments, and for showing how to use the segmentation results to address business and marketing issues.