Skip to main content

4 Easy Steps to Customer Segmentation for SaaS Pricing

Pricing is fundamentally a discussion about value. And value differs between customer segments. If you don’t understand the opportunity for value creation across different segments, you’re going to have a difficult time with your SaaS pricing. In my previous post about customer segmentation*, I discussed the importance of segmentation for SaaS pricing and the variables available in a segmentation analysis. Now it’s time to explore the methods available for customer segmentation.  

At the end of a segmentation process, you need to know two things:

  1. Who your customers are
  2. What your customers want

The need for these two outcomes presents an immediate problem. Where do you start? Do you start with who customers are? Or what customers want? Let’s dive into the segmentation methods first, and we’ll come back to this point. 

The Two Customer Segmentation Methods 

SaaS Pricing is a multi-step process involving two different methods of customer segmentation:

  1. A priori segmentation (aka basic segmentation)
  2. Post hoc segmentation (aka value-based segmentation)

A Priori Customer Segmentation

A priori is a Latin phrase meaning “from what is before.” This segmentation method uses pre-existing classification variables. For example:

  • Company Revenue/Size
  • Industry Vertical
  • Geography
  • Technology Adoption Attitudes
  • Decision-maker or End-user title
  • Any of the General or Product-specific Observable variables I described in my previous post

The primary advantage of a priori segmentation is that it’s a faster and cheaper method since it leverages existing research or classification schemes.  Also, it may be reasonable to use a variable like industry vertical (e.g., retail, hospitality, manufacturing) to segment customers since you might expect customers in the same vertical market to deal with similar challenges. 

This method has drawbacks because you do not start with customers’ needs when defining the boundaries of segment profiles. For example, if we used demographics as our profiling variable, we may target 72-year-old white males living in a castle in England. In this case, you would target Prince Charles and Ozzy Osbourne, two individuals with clearly divergent interests.  

Also, it can be challenging to adapt pre-existing segmentation schemes to the new business context in which we will use them.  For example, segmentation of existing customers by how much they are spending with you today. If you are trying to price and package a new product, that existing scheme may no longer have relevant or valid assumptions for the new offering. 

Post Hoc Customer Segmentation

Meaning “after this” in Latin, post hoc segmentation derives segments based upon customer value drivers. I refer to this as value-based segmentation because we try to understand what customers value and why they buy a product. 

In a value-based segmentation, we perform primary research to understand the customers’ behaviors, preferences, contexts, motivations, obstacles, and desired outcomes. This research helps us understand the customer’s view of the world, what drives them to purchase, and how our product creates value.

Pumping Iron

Since we priced beer in the last post, let’s get a bit healthier in this one and use the example of choosing a gym and work off some calories. There are many different options for consumers to choose from, but we’ll simplify it a bit and outline three possible choices: 

  1. A standard gym (e.g., YMCA, 24-hour fitness)
  2. A high-intensity cross-functional group fitness gym (e.g., CrossFit)
  3. Hiring a personal trainer for a gym or home workout

For even this relatively simple consumer choice, there are still many dimensions a customer may consider to make their choice. For example:

  • Will I have the personal accountability I need to show up?
  • Is there a location within a convenient distance from my home or work? 
  • Do they offer an array of group class options? (e.g., strength, yoga, aerobics)
  • Is there nutritional support available?
  • Do I already have a standard workout routine, or will I need somebody to plan it for me?
  • Will coaching be available to show me proper form, so I don’t get injured? 
  • Is there a strong community with other gym-goers? 
  • What additional facilities are available? (e.g., sauna, pool, basketball courts)

If you talked to 20 prospective gym members, likely, you would not get a consistent ordering of importance to those value drivers. Each person would value one item over another because of their contexts, goals, constraints, and frustrations with existing solutions. This difference in preference among value drivers would cause them to choose a solution that was the most suitable for them. 

Because we ask questions directly related to the outcomes customers are trying to achieve and their context, value-based segmentation is flexible to the business context in which we will use the segmentation scheme. However, because primary research is required, it takes more time and cost than an a priori approach.  Also, it requires sophisticated statistical analysis, which will require assistance from personnel with statistics or data science backgrounds.

Executing a Customer Segmentation for SaaS Pricing 

I recommend using both a priori and post hoc methods. Most managers conduct an a priori segmentation to determine “Who Customers Are” and stop there. It is an excellent first step, but not the final one. You also need to understand “What Customers Want.” 

The segmentation process I use follows this high-level structure:

  1. Align the team on the goals and process
  2. Hypothesize segments using an internal a priori approach
  3. Conduct market research and analyze post hoc segments
  4. Create personas and validate internally

Align on Goals & Process

Start the Customer Segmentation Conversation Early

As I’ve written before, the sooner you get started on this process, the better. Segmentation is the first step in a long chain of marketing activities – targeting, positioning, product development, pricing, and messaging. Building a product for everyone and telling Marketing to “go position it for a target customer” is a bad strategy. 

Get Executive Buy-in on the Importance of Segmentation 

Your leadership may not naturally think in segments or understand the importance of segmentation. Some may even actively insist that “our product is for everyone.” You’ll need to convince executives why segmentation is crucial to the marketing process and why you should prioritize it. Do this at the outset. It doesn’t matter what methods, variables, or fancy statistics you use. If you fail to get buy-in on the concept at the beginning, you’ve wasted your time. 

Help them understand the benefit of using a single segmentation scheme across the company (at least for the same product or product portfolio). I’ve heard from clients, “At my company, Marketing has a set of personas and Product uses another, and Customer Success uses another.” I don’t understand how anyone thinks this is a good idea. If the Product Team is building for one customer and Marketing is acquiring a different customer, nobody will be successful. Having fragmented ideas of who the company is serving will lead to subpar financial performance and unhappy customers.

Hypothesize

Create A Priori Segments 

Start your customer segmentation process with information “in the building.” Talk to each customer-facing team within the org to understand their perspectives on the customers you serve. Ask your teams these starting questions:

  • Sales: What are the characteristics of excellent prospects? What are the key selling points that win those deals? 
  • Customer Success: What are the key drivers of customer churn? What telltale signs do you look for that help you predict whether or not a customer will be successful?
  • Marketing and Product: Who are we currently marketing to/building for, and why? 

This exercise socializes the idea of segmentation within the company. It also surfaces knowledge from different functional areas about why specific customers are successful with your product, and some are not. 

Next, build a customer segmentation strawman. (Contact me for a free customer segment template to use.) In a working meeting, create three or four segments with the executive team, given the prior team inputs. Ultimately, this step helps unearth unarticulated assumptions, creates a forum for a constructive conversation about risk and disagreement, and identifies areas for further research.  Have the team look at this a priori segmentation strawman together and ask, “What are we least sure about?” 

Many companies usually stop here, assuming that this level of segmentation is sufficient, but that is rarely the case. 

Research & Analyze

Customer and Prospect Interviews

Now that you understand which areas need further research, talk to your market. Prospective, current, and churned customer interviews are all of the value here. Don’t bias your research process by only talking to “friendlies.” You already know these customers well and are unlikely to get new information. I advise against focus groups because they tend to produce biased, unusable data.

Your goal is to identify customers’ Jobs-to-Be-Done (JTBD) and validate risk areas identified in your a priori exercise. What are the outcomes, motivations, contexts, relevant competitive alternatives, barriers, and obstacles your market faces around this area?  Remember that a customer’s JTBD exists with or without your product, so you shouldn’t focus only on the people aware of your company or offer. 

How many interviews should I conduct? 

A good rule of thumb is ten interviews for each a priori segment identified in the previous step. Frequently, I plan for ten and allow for changes to the plan based on how much “surprise” is happening with each conversation. If you’re not learning anything new after several interviews, it’s probably safe to say you can stop and use your time more wisely on something else. As I’ve written before, pricing research is all about mitigating risk. You can reduce risk with even a single interview and then look at the time, cost, and information trade-off as you proceed.

Customer and Prospect Survey

Now it’s time to collect data that can help you validate your hypotheses around customers’ JTBDs discovered in the focused interviews at a much larger scale. We need to decide what metric we will use to segment our market once we complete data collection. I prefer the Opportunity Score as defined by Tony Ulwick in What Customers Want. (Note: I’ll create a post on this book soon, so I won’t belabor the formula here.)

The Opportunity Score has several advantages:

  • It acts as an explicit separator between customers based on need and the opportunity for value creation.
  • Later, when you’re working on which segment(s) to target, you will have a clear picture of customer needs and how you can position your product as the answer to solving those needs.
  • If you do this before building the product, you will be more likely to focus your innovation efforts in the right direction. Again, early segmentation helps many downstream activities, including product development, pricing, positioning, and messaging.
A Couple of Pro Tips for Survey Design

Some folks will recommend using Likert scale variables when asking importance and satisfaction ratings on JTBDs. In my experience, this leads to poor data as survey respondents tend to rate everything as important. You want to force people to make trade-offs among the options. Therefore, I recommend a survey method with a forced ranking like a MaxDiff. 

Also, there is a trade-off between the number of response variables versus sample size that you need to manage when running cluster analysis to group your segments. For clustering to work well, you should ensure that you have 10-100x responses as there are response variables. For example, if you have identified 25 customer jobs in your interviews, you would want to collect 250-2500 survey responses to get reliable segments. 

Factor Analysis

If there are many response variables, you may consider grouping response variables via factor analysis. Be careful, however, as factor analysis can lead you to discard potentially meaningful sample variation, losing insight on relevant segmentation dimensions too early in the process. Appropriate use of factor analysis is a blend of art and science, and management should work closely with their analyst counterparts at this stage. 

Create Customer Segments via Cluster Analysis

Like factor analysis, determining the appropriate number of clusters given the data requires domain expertise and business judgment. There are trade-offs for choosing that number. On the extremes:

  • An individual customer could be a single segment, or 
  • the entire customer base could be a segment

Both extremes have apparent drawbacks. Creating too many segments will produce a significant overhead for all areas of the business. Aim for 3-4 segments in the targeting step, but your initial segmentation scheme may create more. Leverage the “Criteria for Effective Segments” discussed in the prior post and input from your data scientist to figure out what makes the most sense for you, given your data set. 

Tackling the full depth of cluster analysis is beyond the scope of this post. However, I like Market Research: The Process, Data, and Methods Using Stata for a clear step-by-step guide for performing this type of analysis.

Profile & Validate

Now Create Those Personas!

Personas are fictional user archetypes created to represent groups of customers who share common goals. As mentioned in the previous post, jumping into persona development without proper research will leave you with poor personas. But now, you have all the research to create personas that will be useful across the business. 

Create your personas by appending Observable profiling variables related to your segmentation data based upon customers’ JTBDs. For example, Industry Vertical, Company Size, User Role, or Geography can be helpful here. The goal is to translate the quantitative and statistical realm information into a common language that the rest of the business can understand. 

Before finalizing personas, distribute your created personas to the sales, product, and marketing teams to test your findings. For examples:

  • Can a salesperson categorize prospects based upon what you’ve produced? 
  • Can you work with Sales to create a simple set of qualifying questions to ask prospects if Observable criteria are insufficient?

At this point, you haven’t yet selected your target segment(s). You are purely validating that what you’ve come up with is realistic and valuable.

Much later in the pricing process, you will want to do a segmentation based upon cost-to-serve and work with the team to decide who you will target, but I will defer those for future posts!

*Note: I use customer segmentation and market segmentation interchangeably in this context. I’m not explicitly referring to your current customers when discussing customer segmentation, but instead all potential customers.