Tony Ulwick is the pioneer of jobs-to-be-done theory, the inventor of the Outcome-Driven Innovation® (ODI) process, and the founder of the strategy and innovation consulting firm Strategyn.

He is the author of  Jobs to be Done: Theory to Practice (IDEA BITE PRESS, October 2016), “What Customers Want” (McGraw-Hill) and numerous articles in Harvard Business Review and Sloan Management Review.

When thinking about market segmentation, an important first question to ask is, “What is a market?” After all, how you choose to define a “market” will determine just what it is you will try to segment. For example:

• You could define a market as a group of people — and segment those people based on similarities in their demographics, psychographics, attitudes, behaviors or the generation in which they were born.
• You could define a market as a technology, such as semiconductors — and segment your market around technology applications.
• You could define a market as an industry vertical, such as health care or retail — and segment your market around geographic region or size of the business.

But when it comes to informing product marketing and innovation, what is the best approach? To answer that question, let’s remind ourselves why we want to segment a market in the first place.

Companies employ market segmentation methods as a means to discover groups of customers that have similar needs so that product offerings can be uniquely tailored to — and effectively positioned at — each segment. In short, market segmentation is a method companies use to discover segments of opportunity: groups of customers with different under/over-served needs.

Consequently, the segments resulting from a good market segmentation scheme should possess these 4 characteristics:

1. Homogeneous population: the population of each segment should be homogeneous, with little variation in the needs customers believe are underserved, overserved and appropriately served.
2. Mutually exclusive: the set of needs that customers believe are underserved, overserved and appropriately served should be different from segment to segment.
3. Collectively exhaustive: 100 percent of the customer population should be effectively classified by the segmentation scheme so all customers are accounted for. This ensures all segments are discovered and the most attractive segments can be targeted.
4. Actionable: each segment should possess characteristics that make it attractive for pursuit with a unique market/product strategy, e.g., disruptive, differentiated, dominant, sustaining, etc.

When it comes to putting Jobs-to-be-Done Theory into practice, two alternative approaches to market segmentation have emerged: Milkshake Marketing (a methodology, as I will explain, that is technically flawed) and Outcome-Based Segmentation (a methodology that is proven to work in practice).

When Clay Christensen employs his Milkshake Marketing segmentation methodology, he defines a market as a group of people who buy a product, e.g., milkshake buyers — and segments them into groups of buyers that have a unique product use case. That product use case is defined by what is considered a unique set of circumstances faced by the buyer and the emotional and functional jobs the customer is trying to get done in that set of circumstances.

So what’s wrong with segmenting a market around product use case and then profiling that segment by the jobs that are causing people to buy a specific product in that circumstance? Quite simply, as a methodology it fails to reliably reveal segments of customers that share a unique set of under/over-served needs.

Consequently, while the Milkshake Marketing approach may be useful for uncovering unique marketing opportunities (selling more of an existing product by telling stories about a use case), it is not the right tool for innovation.

Where does Milkshake Marketing go wrong?

If a company wants to use market segmentation as a method to discover groups of customers who have the exact same under/over-served needs, then it follows that the most effective means for segmentation would be to segment around unmet customer needs. Only after the need-based segment is defined should a company identify the unique set of circumstances that are causing customers in each segment to have the same under/over-served needs.

Milkshake Marketing segmentation does the exact opposite — it makes the use case or circumstance the bases for segmentation, and after the segment is defined, it seeks to understand customer jobs/needs in that circumstance. It is not needs-based segmentation.

In addition, when employing the Milkshake Marketing method, only people who are using a specific product (e.g., milkshakes) are considered, leaving those that bought competing products (eggs and muffins) out of the market segmentation analysis. These two factors contribute to the creation of a market segmentation scheme that is technically flawed along several fronts.

The population of each segment is not homogeneous

The segment of milkshake buyers that find themselves on a long, boring commute to work and want to stave off hunger do not all share the same set of under/over-served needs. Some milkshake buyers in that circumstance may value finding something to do with their free hand, while others may not. Some milkshake buyers may be time-constrained and underserved along many fronts when variations in the restaurant’s ability to delivery cause them to be late for work. Other milkshake buyers may have all the time in the world and have few, if any unmet needs. Segmenting the market around a unique product use case does not guarantee the resulting segment is comprised of a homogeneous population, i.e., a segment in which all members agree on which needs are underserved, overserved and appropriately served.

There is no guarantee the segments are mutually exclusive

When segmenting around circumstance/product use case, it is possible, even likley, that the same under/over-served needs will show up in multiple use cases. For example, when it comes to buying milkshakes, it is quite possible that the segment of morning commuters who buy milkshakes and the segment of parents who buy milkshakes in the afternoon share a common set of unmet needs — both may complain about waiting too long in line, not getting what they ordered, or feeling bloated after finishing their shakes.

If the needs that customers believe are underserved, overserved and appropriately served aren’t different from segment to segment, then the segmentation scheme is rendered useless.

The segmentation scheme is not collectively exhaustive

When segmenting a market, a company would typically want to understand how the entire population of interest is segmented, not just a subset. The Milkshake Marketing methodology does not seek to, and is unable to, account for the entire population — it cannot place each individual into a segment.

Discovering the morning commuter and the afternoon parent purchase segments of milkshake use does not account for all milkshake purchases. Nor does it account for people who purchased other products from those same locations in the morning or afternoon. It offers an incomplete view of the market, leaving possible marketing and innovation opportunities hidden from view.

Milkshake Marketing segmentation methodology does not segment the market around unmet needs.

It risks leaving possible opportunities hidden from view and targeting unattractive segments.

The action you’d take to pursue a segment is unclear

When segmenting the market by circumstance/use case, there is no guarantee that the segment population has an actionable set of under/over-served needs. Worse yet, half the segment population may have one set of unmet needs and the other half of the same segment may have a different set of unmet needs. If the segment population is not homogeneous, then a single market and product strategy cannot be devised for that segment. This is one reason why the milkshake client in the famous Milkshake Marketing example was wise to reject the team’s recommendations.

When segmenting a market, each segment should contain a homogeneous population and the segments themselves should be mutually exclusive and collectively exhaustive. While the Milkshake Marketing methodology is not capable of producing a segmentation scheme with these characteristics, Outcome-Based Segmentation methodology was designed with this end in mind.

Outcome-Based Market Segmentation

It’s true that a circumstance can cause a need to be unmet, but the way to figure out the circumstance (of which their could be thousands of possibilities) that is causing needs to be unmet is not to speculate on the thousands of possible circumstances. Rather, it is to segment the market around unmet needs to discover segments of customers that have unique under/over-served needs — and then determine the circumstances that are causing those needs to be unmet.

This thinking has an important impact and poses a major constraint on how a market must be defined in order to discover the ultimate segmentation scheme — to find the ultimate targets for innovation, you must define a market in a way that makes it possible to segment around the customers’ needs, forcing us to define exactly what a “need” is.

For this reason we define a market as the core functional job that a group of people are trying to get done — and segment the market around the metrics they use to measure success when getting that job done. These metrics, a special form of need statement called desired outcomes, are the customer’s needs. This thinking forms the foundation for our outcome-based segmentation methodology.

What makes outcome-based market segmentation a breakthrough approach is the way it defines a “market” (as a functional job people are trying to get done). This enables a company to define a complete set of needs for that market — for all those people trying to get that job done — and to use those needs as the bases for market segmentation.

More specifically, when applying Jobs-to-be-Done Theory to market segmentation, we study the core functional job-to-be-done, making it the unit of analysis, and uncover all the customers desired outcomes along with all the emotional, related and consumption chain jobs (and outcomes) associated with the core functional job. The data set employs a large number of variables — often totaling 150 or so. (See Customer Needs Through a Jobs-To-Be-Done Lens for more details.)

Next we survey up to 1000 people to document the circumstances they were in the last time they executed the job — and to rate the importance and satisfaction of each desired outcome and job statement, given the solution they chose to get the job done, whether it be a product, service or a homegrown work-around. (See Quantify Your Customer’s Unmet Needs.)

Using statistically valid data that is representative of the total customer population, we are able to discover how the market is naturally segmented. In other words, we are able to use factor and cluster analysis to discover and size all the unique under/over-served outcome-based segments that are representative of 100 percent of the market. Analyzing the profiling data that is collected enables us to discover the circumstances that are causing the outcomes to be underserved.

Outcome-based market segmentation makes no assumptions about the optimal theoretical framework. Instead, the empirical data (knowledge of what needs are important and satisfied in a representative sample of the population) drives the segmentation selection. Half the population may think outcomes 10, 15, 20, 30 and 50 are highly underserved. The other half may think that outcomes 7, 11, 19, 32 and 61 are highly underserved. Cluster analysis reveals these segments within the data.

Once a company knows with confidence what segments of opportunity exist and why, i.e., what circumstances explain causation, they are able to devise a product portfolio strategy that addresses each of the segments they choose to target. They also know size of each segment, which is critical when forecasting revenue potential.

The segments discovered using the Outcome-Based Market Segmentation methodology inherently (by design) possess all the characteristics of a good market segmention scheme. Outcome-based segments contain a homogeneous population and are mutually exclusive, collectively exhaustive and highly actionable.

In Competing Against Luck, Clay Christensen et al say:

“In the context of a data-obsessed world, it might be a surprise that some of the world’s greatest innovators have succeeded with little more than their own intuition about a Job to Be Done to guide their efforts.”

True, but unfortunately, not all of us are born great innovators with perfect intuition. Most of us need a repeatable, proven process to ensure success at innovation. While Christensen’s Milkshake Marketing segmentation approach relies on observation and intuition to try and find segments worth targeting, the ODI methodology uses hard data and statistics to discover segments of opportunity that companies can count on to make innovation both predictable and profitable.