Segmentation

“The reports of my death are greatly exaggerated”
The death of segmentation has been predicted before. There are articles in 200820112014, just pick a year and you’ll find someone writing about how segmentation is dead. One of the most high profile quotes came from IBM CEO Ginni Rometty, who in 2013 had this to say:

“Data will spell the death of customer segmentation and force the marketer to understand each customer as an individual within eighteen months, or risk being left in the dust”

What’s the deal?
Machine learning and big data have been the biggest driver for the death of segmentation. As we collect more data points on customers, we can use this data to further split our segments.  Instead of having 5-10 groups, all of a sudden we could have thousands of groups based on demographics, attitudes, behaviors, dog count, etc. In order to manage these groups, computer algorithms can dynamically create ads for the thousand segments and then machine learning can optimize.

One prominent example is Netflix, which divides its 100mn customers into 1,300 “taste communities.” Think of each taste community as a different segment, but to make matters more confusing, customers can be in multiple “taste communities.” In order to send/develop content to 100mn people in 1,300 taste communities, you need to have a world class engineering and marketing technology group which relies heavily on automation.

But is this right for every business?
Machine learning segmentation works for companies like Netflix, which offers large amount of products (in the form of shows) with similar cost structure per customer. But for companies which have a wide variety of customer spend, it may not work as well.  

Airline companies don’t need 1,300 segments. I don’t have any knowledge of the revenue distribution of Delta’s loyalty program, but frequent fliers probably make up a significant portion of the airlines revenue. You could argue there are a thousand different segments based on where these members are flying, but if we needed to make strategic business decisions – like how to build an airport lounge, or what perks we should offer in first class, using 1,000 microsegments probably won’t help us.

Where does segmentation fit in?
Segmentation helps companies set a high level strategy and determines who their most profitable customers are. Segmentation may oversimplify a business, but it helps stakeholders think about who their customer actually is and rationalize their business. After you have segments, if you want to microsegment the death out of everyone and target them using sophisticated machine learning, auto optimization and marketing technology, I am fine with that.

Just don’t buy the tombstone for segmentation just yet.