August 2021

How we helped make Debt Collection more efficient

We have made Debt Collection a much faster and efficient process for our clients


Customer Behaviour Segmentation

Project Overview

Businesses and institutions which loan money to customers often experience customers who don’t pay back the money they loaned in the given period of time. This phase is called the “Recovery” phase. When customers reach this phase, the institution then has to find a method to retrieve their money back. This job is either handled by the money-lenders or a third-party company they hired solely for this purpose. This task can be extremely challegning and inefficient when you have over thousands of customers in the recovery phase. Without any knowledge on who's actually willing to pay back rather than ignore you, it becomes a guessing game. Our clients were facing this issue and wanted a way to know which customers to target.


We proposed developing a smart AI system which helps predict and segment customers into different categories. These categories depended on how likely they were willing to pay back, and over what time period are they predicted to pay all that money in. We started by collecting as much historical data we could, including data of customers who never went in the recovery phase. There were over 40 variables of data included such as number of payments payment, risk level score, etc.

With this data, we trained a K-means Clustering algorithm to segment the customers into 4 categories.

Project Results

The customers were succesfully segmented into 4 categories. Our model helped optimise costs, improve collector efficiency, help prioritisation, and regularly monitor all their customers. This lead to a 4% increase in our client's recovery rate.

Ready to start working together with us?

Send us a message!

Book a free consultation today, and find out how Codepace can help you reach your goals.

Contact us