Rethinking customer segmentation

Search for ‘types of market segmentation' online, and you'll be bombarded with the same four categories repeated over and over in SEO-optimized articles. (If you must know: demographic, psychographic, behavioral, and geographic).This email is not, however, about the coming AI-driven armageddon of original thought. It’s about segmentation and where its future lies.

‍

Get Off the Bench

Segmentation is a powerful tool. The idea that different types of people behave differently with respect to your product or marketing makes intuitive sense. Historically, marketing organizations have used segmentation to develop targeted marketing strategies and optimize marketing spend.

In practice, though, a lot of segmentation work warms the bench in marketing departments. What’s going on?

Off-Base

Segmentation, according to an HBR piece on AI-driven segmentation, means identifying a discrete group of people with similar needs who are likely to react in similar ways to some action taken by you, the brand.

Except, the authors note, most people leave out the last part, the part about action.

Instead, marketers end up with segments like Middle-Aged Long-Term Suburban Apartment-Dwellers who need non-permanent closet storage solutions and secure nearby offsite mini-storage. The ensuing leap of faith: MALTSADs just need to learn about the new Expand-o-Matic Shoe House and they will become customers. Oh, and all MALTSADs will behave the same way.

Both of those things may be true, but neither has been validated by the segmentation work, which can only take you so far. Why? Because the segmentation was likely generated one of two ways:

1. Via internal data about customers, which may be terrific in defining what customers purchase today but less useful in predicting future behavior toward a new product.
2. Using survey data, which tells you nothing about how customers will respond to the Expand-o-Matic when they encounter it in real life. Also, survey data is often biased.

Are there alternatives to either of these segmentation sources? Anything new on the horizon?

Surveying the Field

For years, data science has been crushing it by analyzing the boatloads of data produced by digital platforms and products. Empirical results occasionally defy conventional wisdom. An analysis of digital music customers, for example, showed that heavy user segments are far less price sensitive than light users. That’s wildly different from typical behavior in packaged goods, where people who purchase products in high volumes expect a discount via couponing or some other device.

Data science has also produced dynamic segmentation, which relies on real-time data to segment and re-segment users based on context, behavior, and other factors. Companies like Klaviyo, Hubspot, Segment, and others provide dynamic segmentation tools online that can optimize inbound traffic by predicting behavior.

But how do you segment when you have no data about customer behavior to predict from? Maybe your product does not yet exist and you want to learn about potential segments. Or maybe your new product is struggling and conventional segmentation tools are not yielding results. Or maybe you are looking for incremental audiences for existing products.

Put Me In, Coach!

We’ve been developing “marketing-based segmentation” at Spark No. 9. Tapping audience data on advertising platforms can identify prospective audience segments not just by demographics, but also by factors like affinities with brands, engagement with certain media, values, interests, behaviors and more.

Using multiple ad platforms to triangulate segment size may yield a better result than extrapolating from surveys. But the real benefit here is understanding the marketing and sales responsiveness of each segment—something survey-based segmentation can’t do.

Targeting prospective new segments with ads for your product gives you real-time behavioral insights about your potential market. Compare the behavior of new segments with that of existing customer segments and suddenly you can quantify the value of each segment.

Other benefits:

1. SPEEDY. Leveraging ad platforms to segment your market is fast. It takes days or weeks, not months or years like conventional survey-based segmentation.
2. DYNAMIC. One of the big problems with segmentation is that people do not stay within segments–they move around as needs and behaviors change. Ad platform algorithms are constantly updating where individuals belong based on most recent behavior.
3. READY TO PLAY BALL. When you are ready to launch your segment-based initiative, your audiences are already in place, road-tested, and ready to be targeted on ad platforms. There’s no translation from survey data to targeting platforms.

So that’s us. What about you? What are you seeing that’s new in the world of market segmentation?