Advanced Analytics Blog

How Many Segments Are Optimal?

When deciding upon a segmentation solution, companies often wonder how many groups they should have. Perhaps 3, 5, or 8, or maybe even 10, 15, or 20?

There’s not a magical number, I’m sorry to say. The optimal number of segments depends on several factors, such as how the segmentation will be used, the organization’s experience with segmentation, and statistics.

Activation plays a large role in determining the number of segments. Companies that advertise to segments using mass media, for example, lean toward fewer segments because developing many different creatives is expensive. Companies that use direct mail or email marketing may require many, many, many segments. I know of companies that use 50+ segments arranged as macro and micro segments because it is relatively inexpensive and easy to market to all of them. Service-oriented industries, such as hospitality and foodservice, tend toward many segments as well. Not only do they have to account for the consumer, but they must account for the time of day and the usage occasion.

Experience also impacts the final number of segments. Stakeholders who have previously worked with segmentations often have entrenched opinions about the best number of segments. I have observed over the years that a stakeholder’s belief in the uniqueness of their products and the variety of human attitudes and behaviors within the category influences the desired number of segments. The more stakeholders perceive that their product category is complex, the more segments they think they need to describe category purchasers. And there are organizations that shy away from many segments because they hold the erroneous belief that they must cater to all of them. Not so!

Segmentation helps companies understand the different constituents in their category and identify the groups that are worth pursuing with limited company resources. In other words, all segments are not created equal. It makes business sense to concentrate marketing and new product development efforts on those few segments that are more valuable. This means that a solution can contain many segments, but the organization only needs to pay more attention to a few of them.

Analytical tools consider the “fit of the data to the segmentation model” to identify the number of segments. There are statistical procedures available that will run many different clustering models and literally point the analyst to the best number of segments. These tools can be useful when assembling a short list of potential segmentation solutions. Companies want the solution they adopt to be their choice. So, providing a small set of statistically sound options is ideal. Using statistics as the only criterion to determine the number of segments is not advisable, though. This can result in an overly precise set of segments that are not really meaningfully different.

Good segmentation solutions have face validity (make sense), are easy to explain, contain segments that are reachable, and have clearly defined targets. These optimal solutions can have any number of segments. One caveat is that each segment requires a readable base size—we recommend having at least 100 consumers/decision-makers in each segment. This prevents key business decisions being made based on segments that are too small.

From a short list of good segmentation solutions, select a final solution that contains highly differentiated groups. A solution is well-differentiated when stakeholders can easily think of a customer that typifies each segment. The optimal number of segments should be an outcome of a well-differentiated segmentation solution rather than a driver of the solution chosen.

It is challenging to convince stakeholders who have a preconceived notion about the number of segments that a solution can be optimal with many fewer (or even many more) customer groups. Uncovering expectations regarding the number of segments prior to beginning the process is key. Knowing that a company requires many segments, for example, will inform quantitative survey sample sizes and analysis approaches. Even then, stakeholders and their research partners should keep an open mind about the optimal number of segments and allow the characteristics of a good segmentation to be the determinants of the solution adopted.

Elizabeth Horn

Senior VP, Advanced Analytics

Beth has provided expertise and high-end analytics for Decision Analyst for over 25 years. She is responsible for design, analyses, and insights derived from discrete choice models; MaxDiff analysis; volumetric forecasting; predictive modeling; GIS analysis; and market segmentation. She regularly consults with clients regarding best practices in research methodology. Beth earned a Ph.D. and a Master of Science in Experimental Psychology with emphasis on psychological principles, research methods, and statistics from Texas Christian University in Fort Worth, TX.

• 
•