The Credit Recovery Center aimed to improve the company's operating result and, at the same time, credit recovery performance for large clients. Conjecto applied analytical models to the operation files to prioritize efforts and increase the recovery rate.
Context
The volume and heterogeneity of the operation files (in text/CSV format) made it difficult to prioritize collection actions and to measure team performance. The company needed an objective mechanism to focus efforts where the potential return was highest.
Data sources
- Text files (CSV): records of credit operations, payment history and collection status.
Applied solution
Conjecto implemented an analytical pipeline composed of:
- Data preparation via Trifacta, transforming the CSV files into standardized relational tables ready for analysis.
- Inference models (machine learning) that estimate the recovery probability of each operation, making it possible to prioritize accounts with the highest return potential.
- Visual analysis models in Tableau and Power BI to monitor operational and team performance.
Results
- Improved operating result, with teams focused on operations with the highest potential.
- Increased credit recovery performance for large clients.
- Continuous monitoring of indicators through management dashboards.
The project showed how applying predictive models to operational data, even in simple formats like CSV, generates direct gains in efficiency and financial results.