Improved receivables processing for debt recovery

Context & Objectives

A credit management company collecting debt needed to develop more analytical agility and work more efficiently. The client, active in debt recovery, was facing a growing number of cases to handle and the conciliators hired by the client to collect debts were overwhelmed in certain regions. The challenge rested in selecting files with a high probability of being collected. From their experience, they had noticed that some debt factors were influencing the recovery rate from the debtors. But, despite a clean database, they were not able to bring all these elements together.

First, data had to be gathered to be able to predict the segments and identify debtor profiles. Then, the client had to ensure a regular follow-up of the debts by an optimal geographical coverage of these conciliators.

Thus, the client set sights to process the receivables entrusted to it to improve efficiency, profitability, and client performance. They called on Agilytic to rapidly deliver a solution to help them focus on cases with a higher chance of recovery while recommending the best route to the conciliators.

Approach

We worked with two core modules to handle the client’s different requests and ease the process of receivables recovery:

1. Predictive Module

We gathered data from the debt files with information like the amount of the debt, the activity of the creditor, and the regional recovery rate. Next, we built up a historical dataset on more than 150k previous claims and enriched it with external data to be able to identify typical debtor profiles more easily.

After this data collection and analytical data building phase, we segmented the receivables according to their receptivity to the collection channel used and the type of receivable.

We then developed a scoring model to assess the probability to fully recover a debt, identifying the debtors for whom a legal procedure has a high probability of leading to recovery. During this process, we closely collaborated with the client to validate the result. Together we reviewed the working assumptions to ensure the final quality of the model and the client adoption. We tested the model on recently closed cases to assess its reliability over time. On top of the fully implementable algorithm, the client has received detailed explanations on the influencing factors.

We eventually applied this to all the receivables. Thus, we could categorize the debtors according to their receptivity to the collection channel to use and their probability of being collected.

2. Geo-Marketing Module

Optimizing the geographical breakdown of the conciliation zones was required to ensure a straightforward follow-up of the files in progress. To begin, we subdivided each postal code into nine clusters.

We used the data processed in the Predictive Module to optimize and quicken the conciliators' coverage areas. This required the support of the client to precisely define the available resources (in terms of conciliators) and the constraints that would be related to the relocation of some of these resources.

Next, we developed a geocoding routine for addresses to determine latitude/longitude (via Google API) and assign each address to a cluster (a sub-category of a postal code).

Processing this data, we also set up a reporting system to measure the individual performance of the conciliators according to certain criteria related to the claims to monitor the quality of the geographical coverage.

Results

We developed a portable version of the Geo-Marketing application and tested the solution in the client's environment. The automated application checks every week for the presence of a file, processes it, and sends it back to the client.

The client can now focus on the top 10% of the cases that have an average recovery rate of 90%. With the vast number of debts handled each year, knowing where to invest the right time and effort is crucial for profitability. Moreover, the quality of the cases assigned to the conciliators has a big impact on their motivation and productivity.

And the client obtained all these results in less than eight weeks of work.