1. Reads
  2. Lite

How to Effectively Use Machine Learning and AI for Customer Segmentation: An Illustrative Example

Data Scientist, DirecTech Labs
Jun 14, 2018 12:43 PM 3 min read

Editor’s Comment: What degree of insights are possible through Artificial Intelligence (AI)? How to make these inputs actionable from a business perspective? These questions are often lost in the theoretical narrative of AI, so it’s worthwhile to examine a real-life example to reflect upon its power.


Let us assume you run a gym and take most business actions based on experience and often only a hunch. Whenever you encounter one of the below questions, your best option may be to guess or seek an external opinion:


  • Who are your most important customers? Who should you focus on the most?
  • Which products are popular with existing customers? What are those products’ principal purpose? To increase strength or to stay in shape?
  • Which customers should be first offered your new products and services such as spa & sauna, personal trainers etc.?


To answer such questions, you will attempt to do some form of segmentation, based on 1 or 2 parameters like demography, region etc.


e.g. Millennials are more likely to buy a certain product or service or customers belonging to a region are likely to invest in a certain offering.


The human brain can easily segment and visualize customers when there are around 1 or 2 parameters involved. There is no way the human brain can crunch more than 3 parameters simultaneously.


AI has the capability to assess hundreds and thousands of parameters together, providing unbelievable results. Availability of abundant data and computation power has made this possible in recent years.


These very questions were troubling C&C, a gym operator, which hired a Data Scientist. Fortunately, C&C as storing its customer data in a methodical format as seen below:

How to Use Artificial Intelligence for Effective Customer Segmentation: An Illustrative Example


Using Artificial Intelligence for Customer Segmentation


Interestingly, using unsupervised AI learning methodology the Data Scientist was able to segment 1,000 C&C customers in the following segments:

How to Use Artificial Intelligence for Effective Customer Segmentation: An Illustrative Example
*Quadragenarian means between ages of 40 to 50

We can hence derive:

  • There are c. 50% active people, comprising Quadragenarian and Young Active Males, using most services of the Gym. This is the segment generating the highest revenues. Their friends and family are either current or potential customers.
  • There are 40% young, inactive females who use the Gym’s services to increase their strength. C&C should seek to improve and customize its services for this segment and convert them into active, high revenue customers.
  • 10% customers are not really adding any value to the business and are unlikely to use the Gym’s new services.


AI can bring rich and actionable insights such as the above in the domain of customer segmentation. Businesses should realize the importance of in-depth data and the power of Machine Learning to take their progression to the next level and maintain an edge against competitors. In my 10 years of professional experience, this is the single most important point of learning I have derived.


(We are now on your favourite messaging app – WhatsApp. We strongly recommend you SUBSCRIBE to start receiving your Fresh, Homegrown and Handpicked News Feed.)