Building Customer Segments That Change How You Sell

Why segments often end up in a drawer

Many businesses have a classification of their customers: industry, trades, public sector, private. It sits in accounting or in the CRM, and occasionally a revenue report is sorted by it. In everyday sales, it changes nothing. Every customer gets the same offer, the same service, and the same prices, and the salesperson decides by gut feeling whom to give how much time.

The central question, then: how does a small business build customer segments that actually change something in sales? To answer it, we first distinguish segments from related tools, then introduce two mental models, the nested approach and the price and cost-to-serve grid, weigh how fine segments should be, and finally go through building them in six steps.

A segment that changes nothing about how customers are handled is a column in a spreadsheet, not a decision.

What a customer segment is and what it is not

The goal of segmentation is to find groups of customers whose needs are similar within the group and different between groups. Rangan, Moriarty, and Swartz (1992) cite Wind (1978) for this definition. A customer segment is such a group that the business deliberately handles differently from the rest: with a different approach, a different offer, different service, or a different price.

This website has three related tools that are often confused. They answer different questions and complement each other.

The ABC analysis of customers looks at value, not at need. The ideal customer profile from the guide Defining target customers describes the kind of company you work with best and is, in this sense, the most important segment, but not the only one. The persona comes afterward and describes the people in the companies; see the article Personas in B2B.

The first mental model: the nested approach by Shapiro and Bonoma

Benson Shapiro and Thomas Bonoma (1984) proposed an approach to segmenting business customers that is ordered by effort. The characteristics are nested like layers. On the outside are the characteristics that are easiest to collect; on the inside are those that require knowing the customer well. You start on the outside and only go as far inward as it pays off. The authors themselves warn against applying the approach like a recipe; it has to be adapted to your own situation.

For small businesses, the model's value lies in the sequence. Company data are quickly collected but often do not distinguish by need. Two machine builders of the same size can buy in completely different ways. The third and fourth layers, purchasing approach and situation, are usually the most useful because they directly determine how sales should work.

The second mental model: the price and cost-to-serve grid

Rangan, Moriarty, and Swartz (1992) developed an approach for mature markets in which products have become comparable. There, size, industry, or product benefits alone are rarely a sufficient basis. What matters more is how customers trade off price against service. They place customers in a grid with two axes: the price a customer actually pays and the cost of serving that customer.

On the diagonal from the basic to the comprehensive offering lie the customers for whom price and cost match. Those who pay little get little service; those who want a lot of service pay more. The authors call this line the equity axis. On the other diagonal lie the customers for whom price and cost diverge because one side has more bargaining power. The authors speak of the power axis. Their key idea: a supplier can influence where customers move in the grid through price and service.

The grid connects two articles on this website. The cost to serve can be estimated with activity-based process costing, and the answer for high-effort customers is often a tiered offering, as described in the guide Good, better, best. For customers with a high price and low effort, the point from the guide Why customers leave applies: business customers usually leave quietly.

The sales test: is a segment any good?

Not every group that shows up in the data is a usable segment. I test every segment with four questions. I call this the sales test. If none of them is answered with yes, the segment is a statistic.

• Does the segment change whom we approach and through which channel?
• Does it change what we offer, such as scope, tier, or additional service?
• Does it change how we serve, such as visit frequency, contact person, response time?
• Does it change what we charge, such as price, payment terms, minimum value?

Two minimum conditions come on top. The segment must be large enough for the different handling to pay off, and sales must be able to assign a customer to a segment in everyday work without starting an investigation. A segment only accounting can recognize changes nothing in the conversation.

Two views: few broad or many fine segments

The first view prefers fine segments. The more precise the groups, the better offer and approach fit. Rangan and colleagues speak explicitly of microsegmentation within larger groups and show that only this finer division makes differences in buying behavior visible.

The second view warns about the effort. Dibb and Simkin (2001) analyzed case studies of companies that wanted to introduce new segmentation schemes. The barriers lay in three areas: in the infrastructure at the start, in the segmentation process itself, and in implementation. Dibb (2005) distinguishes hard barriers such as missing data, money, and people from soft ones such as corporate culture, coordination between functions, and leadership style. The hard ones can be fixed early; the soft ones only with difficulty.

For a small business, I consider three to five segments the sensible upper limit. Every additional segment requires its own handling that someone has to sustain in daily work. More segments only make sense once the first ones are actually handled differently. The analysis can be fine; the implementation must stay simple.

Building customer segments in six steps

1. List the customers of the last two years

   With revenue, contribution margin, and, where known, industry, size, and type of order. The list from the ABC analysis is a good starting point.

2. Estimate the cost to serve

   For each larger customer, roughly estimate how many hours of sales, quotes, queries, and special requests the customer causes per year.

3. Group by purchasing approach and situation

   Add characteristics from the third and fourth layers: how does the customer buy, how urgently, how often, with which criteria?

4. Place them in the price and cost-to-serve grid

   Place customers in the four fields by price paid and cost to serve. Mark conspicuous groups.

5. Define three to five segments

   Combine groups that need similar things. Test each segment with the sales test.

6. Define the handling per segment

   For each segment, write approach, offer, service, and pricing rule on one page and create a field for it in the CRM.

The guide Calculating customer lifetime value describes how to determine contribution margin and customer lifetime value per customer. If you set different prices for a segment, apply the justification test from the article Price differentiation: every price difference needs a difference in service, quantity, timing, or effort that the customer can understand.

What changes in sales

Only the handling turns a classification into a segment. For the worked example, it could look like this.

Conclusion and a first step

How does a small business build customer segments that change something in sales? It starts with easy-to-collect characteristics, moves on to purchasing approach and situation, sorts customers by price paid and cost to serve, and combines them into three to five groups. Each segment must pass the sales test: it changes whom the business approaches, what it offers, how it serves, or what it charges. Research shows that segmentation rarely fails in the analysis but in data, time, and habits during implementation.

1. This week: Place your twenty largest customers in the price and cost-to-serve grid, based only on your own assessment.
2. Next week: Check the assessment against figures for contribution margin and service hours.
3. In a month: Define three segments, write their handling on one page, and create a field for them in the CRM.

You can tell a good segment by the fact that a new employee knows how to deal with the customer after one look at it.