Playing to Win

Segmentation & Strategy

Three Important Truths

Roger Martin
8 min readJul 25, 2022


Source: Roger L. Martin, 2022

I have received many reader requests for advice on how to think about segmentation when creating strategy. Segmentation is a huge subject with which I could fill several books, so I thought I would focus on the few key things about which I always think when I am working on segmentation and strategy. To that end, I have dedicated my 39th Year II Playing to Win/ Practitioner Insights (PTW/PI) piece to Segmentation & Strategy: Three Important Truths. You can find the previous 91 PTW/PI here.

Before diving into the three truths, I want to give a shout-out to Craig Wynett, my long-time collaborator at P&G and now (sadly) retired. Many years ago, he introduced me to the work of Byron Sharp and the Bass Ehrenberg-Bass Institute, which is some of the very best stuff out there on marketing and it has influenced my views substantially.

1) Segmentation is Not Deterministic; Its Probabilistic

I find that there is a prevailing technocratic mindset that conceptualizes segmentation as largely (if not entirely) deterministic. That is, once you define a given segment properly, the customers defined to be in the segment will buy the offering you designed for them and those outside the segment won’t.

In real life customer behavior isn’t so deterministic; it is probabilistic. For any number of reasons (your offering is out of stock, she just felt like trying something else that day, his friend who was with him talked him into trying a different offering, etc.), despite believing that your offering is indeed an excellent match for their needs/wants/desires, customers will buy something else. Meanwhile, customers for whom your offering was not designed — not even close — will buy it, again for any number of reasons.

You do have a higher probability of earning purchases from the customers for whom you explicitly designed your offering and a lower probability from those for whom you didn’t. But it is not deterministically so. That means it is a bad idea to treat customers in your segment as if you own their purchase decision. And it is an equally bad idea for you to discourage customers outside your defined segment from buying from you. Certainly, you should spend more time and resources on sending positive messages to your target. But they won’t be the only ones that buy.

2) You Don’t Determine Customers’ Segment; They Do

It is useful to do work to determine exactly which customers are in your segment — their sex, their age, their income, their occupation, their mindset, etc. — as you attempt to design your offering. You have to design the offering with someone in mind.

That notwithstanding, customers decide what segment they are in, not you. Big box mass merchandisers (other than Costco, actually) took a while to figure that out. They thought their segment was low-to-middle income families willing to drive to a more distant retailer than their local supermarket to buy goods at a lower price. But Mercedes and BMWs kept showing up in their parking lots. They shouldn’t have been there! They weren’t in the segment. That is half right. They are not in yours, but they are in theirs. And you don’t generate revenues: they do!

Customers decide whether your offering is a sports car, or not; a cool thing to order at a bar, or not; environmentally friendly, or not. While you are segmenting customers, customers are segmenting you. They create categories, put you in one, and consider you accordingly. You may think their segmentation is nuts, but it just doesn’t matter. Again, you don’t generate revenues: they do.

You must keep your eye on the prize. Your segmentation is only as good as the customer’s interpretation thereof. It is really important to understand how many of your actual customers aren’t in your target segment. They provide clues as to how you might want to migrate your offer toward your actual segment versus your theoretical one. Clearly, if they are buying from you even though you don’t think they should be, there is something very appealing — accidently so — about your offer. So, you shouldn’t scrap it. But they can help you refine and strengthen it.

3) Segments aren’t Homogeneous; They are Heterogeneous

Segments aren’t homogeneous. It isn’t as simple as ‘these are our customers’ and ‘those are yours.’ Instead, there are many shades. Your choices make your offering perfect for someone — portrayed as the black dot in both sides of the illustration above. The circle on the left is a view of the segment from above with the two axes connoting segmentation variables, and the right is a view from the side with the vertical axis connoting value in excess of the best competitive alternative — I think of it as an upside-down bowl sitting on a counter top. For the dot customer, its desires match perfectly with your offering. Because of that perfect match, that customer is at the very center of the circle on the left and at the very peak of the curve on the right.

Radiating out from that perfect customer are customers who value your offering ever less — they are farther from the center of the circle on the left and are farther down the upside-down bowl on the left. If a customer is located at the very edge of the circle on the left, it is indifferent between your offer and that of a competitor (or competitors). And they will be at the lip of the bowl on the right — for them, there is no value advantage of your offer over that of competition. Think about it in terms of heat. Customers in the center are intensely hot about your offer and as you radiate from the cetnter, they get ever cooler about your offer until they are completely indifferent. So, your market is defined by the size of the customer population within the circumference of your circle/to the lip of your bowl.

The bigger the circle/bowl, the more valuable your strategy. The circle is expansive and the bowl deep for Google Search, Apple iPhone, Facebook, Verizon Wireless, or Tide. The bowl may be deep, but the circle is much smaller for Tom’s of Maine, Burt’s Bees, Lush, or Disaronno Amaretto. The more successful and valuable offerings have a shallow and lengthy drop-off from the perfect customer. The less valuable — even though highly valuable to the customers within their tiny circle — have a steep drop-off.

The steepness of the drop-off is what makes the difference between successful and unsuccessful entrepreneurs. Many unsuccessful entrepreneurs design an offering that is extremely valuable to the perfect customer — typically themselves — but the drop-off is so steep that their idea collapses, not because it didn’t create value, but because the steepness of the drop-off makes it impossible to make the economics work. Successful entrepreneurs design their offering in a way that appeals to a much broader audience.

The implication is that you need to design for your perfect customer but equally think carefully about the decline curve. If you obsess about the perfect customer and not about the shape of the value decline curve, you will be out of business. This, by the way, provides an explanation of the phenomenon that the late great Clay Christenson described in The Innovator’s Dilemma. If you focus only on the perfect customer for your offering, you will ignore entrants who appeal to customers at the edges of your circle/bowl, and they will shrink the size of your market, threatening to put your economics into a downward spiral.

Research for Segmentation

I frequently get asked about how to do the research necessary to develop a segmentation model. On that front, I believe strongly in getting a nuanced understanding of customers. I hate flattening down their point of view to a single quantitative measure: What is their opinion of x as expressed by numerical ranking between y and z? I know that it is comforting to statisticians, and they are entirely willing to sacrifice business accuracy for statistical acceptability.

I favor depth of understanding over statistical rigor because the cost of rigor is often irrelevance — i.e., you get statistical significance about something that doesn’t really matter. Plus, any attempt to achieve statistical significance will drive you away from creating the future because the data can only tell you about the past. Statisticians tend to believe that you can figure out how to create the future from statistically significant data, but all you can do is determine how to perpetuate the present.

It is only by utilizing subtle interpretations of deeper, non-statistical data that you can get clues as to what might be the dot in the middle of a future circle/bowl. That means getting close to the customers, understanding their world, listening to everything they say, being beware of overly filtering. In addition, don’t obsess entirely about the average customer. It is important to explore the outliers because an outlier today might be mainstream tomorrow.

People also ask me about utilizing personas. I like the persona tool because it helpfully creates a vivid and inspirational image of the dot in the middle of the circle/bowl. However, personas have one downside to keep in mind and avoid. Undue focus on the persona can cause obsession on the dot in the middle when the real task is to make sure the circumference of the circle is as great as possible, and the bowl has a shallow drop-off in value. That means determining what motivates the choices of customers who aren’t near the center of the circle — i.e. the cooler ones as well as the hot ones. Focus on the persona at the dot can make that effort more difficult. So, use personas, but be careful when you do.

Practitioner Insights

Your job in segmentation and strategy is to find and satisfy a perfect customer who exists at the center of a circle with a circumference that is sufficient large to sustain a viable economic system, which requires a shallow rather than steep drop-off in value from that of the perfect customer. Pricing is a very tricky aspect of any business strategy. An acute understanding the value contour of the bowl will help you price to optimize the size of the circle.

Take segmentation seriously and have a positive and proactive view of the customers that you want to attract. But don’t be too precious about it and be very careful not to overuse statistically significant data. Segmentation is probabilistic not deterministic. And never forget that customers are in charge, not you. Leave the door open to the customers who you don’t think will find your offering attractive. Despite what you think, they might — and you don’t want to get in their way.



Roger Martin

Professor Roger Martin is a writer, strategy advisor and in 2017 was named the #1 management thinker in world. He is also former Dean of the Rotman School.