Customer Segmentation
Get to know your clients better than they know themselves.
Not all clients are created equal.
Customer segmentation can help you understand your customers and meet their unique needs and expectations. Every customer is different and every customer journey is different so a single approach isn’t going to work for all.
This is where customer segmentation becomes one of the most valuable processes.
What is customer segmentation?
Customer segmentation is the process by which you segment your customers into segments based on common characteristics. These characteristics can be based on demographics (age, sex, location, etc) or more sophisticated methods such as behavioral analysis, so you can communicate and market these customers more effectively.
The segments of customers can be built in two ways: by defining them manually, in what is called supervised learning, and by mining them from data, typically using an unsupervised learning technique, like clustering.
Having your segments defined, your company can now begin building a list of marketing personas. This is typically used to define the brand messaging, and positioning and improve how a company sells.
Customer segmentation is not the same as market segmentation
While they share a similar name, customer segmentation and market segmentation are not the same things. Whereas market segmentation relates to the whole market, customer segmentation is specific to your part of the market.
For example, your bank might be focused on retail investors, while another might be targeting small and medium-sized business owners. These are two different market segments, with different needs, and both banks will provide different services.
However, your clients might still be segmented more granularly. You might have for example two large groups: one composed of investors with investments as their only source of income, such as day traders or swing traders, and another composed of people who depend on their salary but still want to enjoy the benefits of investing in the stock market. These two groups would be considered two different customer segments.
Types of customer segmentation
Any characteristic belonging to a customer can be used to cluster them into different segments, however, choosing those characteristics should be given careful consideration. Some characteristics are more relevant than others, and some might even be unethical to use.
Customer segmentation can be created in two main ways: manually and automatically, through unsupervised learning.
Manual customer segmentation
This type of segmentation can focus on two main types of characteristics: demographics and behavior. Demographic can include factors such as age, geography, urbanization, income, relationship status, family, job type, and even sex.
Any of these characteristics, or combination of characteristics might have a correlation to a certain set of needs and expectations.
In banking, behavioral characteristics can include things such as:
Source of income;
Size of the accounts;
Amount invested;
Client debts to the bank;
Number of banking products;
Number and size of transactions (deposits, withdrawals, and transfers);
The number of accounts: current account, savings, investment accounts, etc.
Etc;
As mentioned before, there is no single approach to use here. The characteristics to use vary from company to company and from time to time, as some of these segments might even disappear while new ones form.
Unsupervised customer segmentation
Another more sophisticated and modern technique is to cluster users automatically using artificial intelligence. If built and applied correctly, AI algorithms are capable of identifying patterns and customer segments in a business.
While these segments can be used directly, many times they are used only to get a first draft of what the customer segments look like before a human or a team of humans generates segments based on the AI’s output.
Why should you segment customers?
Customer segmentation is popular because it helps companies better understand who their clients are and what they are looking for. This can help not only identify what solutions to develop and promote, but also the best way to communicate them to each person. By doing this, customer acquisition is facilitated, and the same happens with customer retention.
In summary, effective customer segmentation will help you acquire more clients and increase customer lifetime value, which means they will stay longer and spend more.
How to segment customers?
To segment your customers, you should start by looking at subsets within your customer population. Ask yourself: What ties certain customers together, and how are certain characteristics correlated?
It might help to define:
Demographics: who they are
Behavior: what they do
Needs: what they want.
Some products will have a strong tie to demographics. For example, student loans are more common in the younger demographic, while having a checking account is equally common in different demographics.
Corporate clients analysis
A corporate client will have different characteristics and products than a retail one.
This means we should look at these clients through different lenses, and build more specific features for them.
Some of these features can be related to their business, such as the industry they are in, financials such as revenue, and operational, such as offices and employees.
These features will help you distinguish clients with different attrition and growth rates.
Attrition and Growth levels
One key important feature you’ll want to explore on your corporate clients is how many salaries they are paying today and how much that compared to the same month in the previous year, or other timelines.
This will give you an idea of the company’s attrition levels, i.e., how often are employees leaving the company over those specific time periods.
You can also explore the evolution of their revenue over time and, depending on the accounting information you have for that client, how much share of wallet you have for that customer.
Our Experience
We’ve built clustering algorithms to segment clients based on their different levels of engagement with the bank and banking products, obtaining 4 different segments based on:
Age
Assets
Logins
Card usage
Number of banking products
Number of insurances
Loan simulations
etc.
When crossing these segments with clients who purchased personal loans, we realized that over 50% of our customers had little to no interest in purchasing. These clients were older than average, had very little engagement with the digital channels, and had few banking products or services.
On the other hand, we had one segment with around 13% of clients which totaled almost 60% of the clients with personal loans. These clients were younger than average, highly digital, and had several products from the bank.
This analysis helped the bank build its customer journey and focus its efforts on the digitalization of the other segments, increasing the sales of personal loans as a consequence, and advertising more loans to the clients that belonged to the more digitalized segment.
Want to learn more?
Miguel Cabrita
Senior Data Scientist, Co-founder
Miguel has helped various companies in banking and finance implement lead scoring and AI solutions.
Having a strong technological background and understanding of business processes and the banking industry helps him detect specific needs and offer the necessary AI solutions for each of them.
