What is the “right” segmentation?

I get this question quite often, as everyone seems to have an opinion on what type of segmentation strategy is the be-all-end-all solution. Well, based on my experience, there’s no right answer and, as with many things in life, the real answer is that it depends.

As far as analytically based marketing strategies, I have a true love-hate relationship with segmentation. I’ve worked on countless segmentation projects, written dozens of papers and delivered numerous presentations at conferences on the topic. But the more experience I get with it, the more I realize there’s no perfect answer.

Generally, there are few “camps” of segmentation lovers. The first type is usually the agency or market research lovers who think an attitudinal, research-based segmentation is the only way to go. The common approach is usually to conduct quantitative research (e.g., a survey where respondents answer questions about a brand or category), and then perform a clustering exercise on 1500 records to build segments. In this scenario, the methodology is almost invariably to collect data through a web-based survey, run factor analysis on all my questions and throw my factor scores into a distance-based clustering algorithm – and voila, out pop my perfect segments. Ironically, although it may seem like I’m opposed to this type of segmentation, I’m actually a huge fan of research. However, trying to operationalize a segmentation strategy in situations where I only have coverage on a handful of customers invariably fails. And when I try to “predict” the segmentation, it falls harder. In my opinion, if you want an attitudinal segmentation, then there’s nothing wrong with doing that; but know that what you’re getting is a study on segments, not an operational segmentation to implement.

The next type of segmentation lover relies strictly on life stage segments – which is nothing more than demographic classifications. These take on multiple flavors but usually include two key elements – age and affluence. Again, if I want to describe my customers in a simple and understandable way, this is a good approach. But don’t expect to find massive improvements in financial metrics based on differentiated treatments based on age and income alone.

The third type is usually performed by traditional database marketers who believe that a behavioral segmentation is the way to go. This approach usually means taking customer behaviors (which basically means purchase data) and segmenting based on what people buy and how much they spend. This is actually the school of segmentation that I learned, and its value is in truly differentiating on what customers are doing. However, from a descriptive and emotional perspective, it’s very hard to glean WHY people do things by looking at observed behaviors. Sure, I know why you bought that product but I don’t know what was important to you in making that purchase and I don’t know how you plan on using it.

Ultimately, all of these segmentation solutions have value and they are all, unfortunately, flawed. So going back to the original question, what’s the right segmentation? Well, I’ll tell you what I’d do.

Let’s assume for a second that I’m the chief customer officer at company X (and yes, you are seeing that title more often these days). My segmentation strategy would take on multiple flavors. At the most foundational levels, I would want to create two basic classifications of my customers. I think of these as my base camp segmentation tools. The first is an understanding of the customer lifecycle and the second is defining customer value.

Lifecycle is a simple yet powerful core customer tool that really helps me understand the health of my customer base. Utilizing the number of new customers, active customers and lost customers – and the ways in which they are transitioning to other lifecycle buckets – I can basically describe whether my customer “funnel” is healthy or sick and understand the opportunities and risks I have.

The second base camp tool for me is customer value. Ultimately, this should be defined as long term value, or sometimes called lifetime value (LTV), but it’s basically taking a single version of the truth as to how my organization will value each customer. People like to think value has all sorts of applications, as it cures many things. But I’d use it simply to do two basic things. The first is investment and the second is measurement. Customer value should be used to understand which customers I should invest more in and which I should not waste my resources on. Its other purpose should be for measurement and understanding how my programs and initiatives are making a positive (or negative) impact to customer value. By stating these two uses of customer value, I’m basically saying there are a whole lot of things that it should NOT be used for, such as differentiated treatments, contact strategy, creative strategy and so on.

To complete my segmentation toolkit, once I have my base camp figured out, I would want to apply a behavioral based, attitudinally slanted segmentation strategy to help me think through positioning, treatment strategy and product relevance. The trick is how to balance two data types (research and behaviors) that are hard to link into one solution. I’ll save my approach to that for another post, but I will conclude with this: there’s no way I’d even worry about having a higher-level segmentation strategy without first building my base camp. Start with the fundamentals and build from there. So if you don’t have a good understanding of how healthy your customer base is, and if you don’t have a common view of customer value, maybe you know what you should be focusing on first.

Join the Discussion