Boosting Debt Repayment With Enriched Data

Use better datasets to understand your receivables and improve your repayment prediction engines
Receivables management firms must understand how likely a borrower is to repay their outstanding loans to create better recovery strategies that yield results. This is complicated by regulations that prohibit the use of specific data in order to avoid discrimination based on specific criteria. For receivables management analytics, this means that finding the right data to build reliable models and predictions is a complex minefield.
The problem

Our customer has built models based on internal historic data on previous interactions but is looking to expand its predictive capabilities while remaining compliant with anti-discrimination and privacy regulations. Compliance needs mean that the company is limited in the external data sources they can pull from. Moreover, the inability to use any personally identifiable information (PII) means that they must be able to extract useful and focused insights from contextual data.

The solution

Our customer enriched their internal datasets with location-based data that provides a broader look at potential repayment capabilities based on external conditions. Additionally, they used the Explorium platform to engineer new features based on their internal data which were able to provide a quick uplift in their models’ accuracy. Some of the sources our customer used to enrich its data include: 

  • Financial indicators based on location that can indicate socioeconomic status and repayment ability 
  • Contextual geospatial data related to financial stability by region
  • Data related to external events that could impact financial stability
The results:

More accurate repayment forecasting and more relevant features

Using the Explorium Enrichment Catalog, our customer was able to build more robust datasets that improved their models. Additionally, the Explorium platform was able to validate and add new features to their models that resulted in a noticeable uplift of 4%. More importantly, the company was able to build models that were accurate and gave reliable results even without the presence of personally identifiable information. When it comes to determining the likelihood of repayment, our customer was able to build features that more precisely provided answers and improved their rate of collections, as well as their overall revenues while reducing defaults.

Enrich debt collection models with better data to improve repayment forecasting and provide better insight.

Other resources you might be interested in