How to use Customer Segmentation To Learn

Kristen Kehrer
Mar 10, 2018 · 4 min read

It’s easy to get people to buy into the idea of “Don’t test to win, test to learn.” However, when it comes to segmentation, it’s sometimes more difficult to get people over to the camp of “Don’t segment to align with your previously held ideas, segment to learn.” This may be because this saying is certainly not as short, sweet, and catchy; but I’m a strong believer in segment to learn.

I’ve been approached before with: “We did a segmentation of the market; can you tie this back to our customer base?”. This means that they did a segmentation based on survey data of the population, these people could be using your product, but they’re not necessarily using your product. This survey probably asked about the functionality they need, what they’re trying to do, and who they are.

Although sometimes we might do things with data that mystify and dazzle, we are not wizards. I cannot take your segmentation of the market and find those same segments in our customer base. Our customers are not necessarily representative of the market, and I don’t have your market survey data for all customers. But what I have is even more precious, actual behaviors that your customer has taken.

These actions include things like:

· I can see how often someone has visited our website

· Are they buying certain products and not others?

· What are they doing on our site? Reading informative articles? Visiting a lot but not converting?

· How often they’re purchasing.. Are these less expensive or more expensive products?

· If they’re calling customer service.. Did they just need help finding a product? Or were they unhappy and looking for a refund?

· How long they’ve been with us

· How were they acquired (channel)?

· Are they being upsold or cross-sold products?

This type of information is hopefully in your database, somewhere. If not, you may be able to find a way to get at it.

We could also append data from a data vendor if we have the budget. There are companies like Epsilon, Full Contact or Axiom (to name a few). If you have the budget to do this, and send them customers name, address, and some other information, they can add columns for things like:

· Income

· Race

· Education

· Employment

· Spend behavior

· Lifestyle and interests

· And more!

This would give you lots of great data to play with. The other option might be appending Census data at the zip code level. All this data could be analyzed and potentially be meaningful in creating a segmentation.

There is another instance of this problem manifesting itself more directly. When we do find ourselves in the position to do a segmentation of our customer base, I sometimes hear: “We’d like to do a segmentation, I’m thinking segments like….” Here people are explicitly using a segmentation to reinforce their previously held beliefs.

Think of how you’d be short changing yourself. Think of all the variables that you could come up with! Be innovative! Create variables that the business has never looked at before. Try to identify a way to determine:

· Who are your seasonal customers?

· Who in your base responds to your marketing campaigns? And how?

· Time between different forms of engagement with purchasing your product or visiting the website.

You can create a segmentation around acquisition or retention, where you know what you’re trying to optimize. And these types of segmentation certainly fall into “segment to learn”, my favorite is when we use an unsupervised algorithm.

There was a great article on by john Sukup on DataScience.com (link), that explains some drawback of k-means and offers some different solutions. Check it out! I also used his code to make the clusters visual at the top of my article.

And once you have that output that makes sense, start learning! Learning is a manual process that typically involves doing a lot of crosstabs (at least in my experience it has been), but you can take that back to the business with a big smile and recommendations on how to target the customers in the segments. Show them what you’ve learned and let them know it’s ACTIONABLE. These are the segments that you can easily add to your database and use to build campaigns.

Segmentation is an enjoyable experience, where you learn a ton about who your customers are. This will allow you to help determine what type of content might be most appropriate, and nurture these customers appropriately. I’m amped up just thinking about it. Hope you are too.

Visit my website! Here

Kristen Kehrer

Analytics Leader, Data Lover, www.datamovesme.com

Welcome to a place where words matter. On Medium, smart voices and original ideas take center stage - with no ads in sight. Watch
Follow all the topics you care about, and we’ll deliver the best stories for you to your homepage and inbox. Explore
Get unlimited access to the best stories on Medium — and support writers while you’re at it. Just $5/month. Upgrade