Success story
Optimize
Non-Performing
Loan Management
Improve debt recovery by combining explainable AI and optimization to assign every case to the right collection strategy.
NPL allocation
across collectors
recovery strategies
for every debtor
monthly assignment cycles
optimization with
configurable constraints
Banks managing large portfolios of non-performing loans must balance recovery performance with operational efficiency and regulatory expectations. Every month, thousands of debt positions need to be assigned to collection agencies and individual collectors, a decision shaped by multiple business constraints, available resources, and recovery priorities.
Traditional rule-based approaches struggle to optimize these decisions because they evaluate constraints one at a time rather than considering the allocation problem as a whole. As portfolios grow, striking the right balance between recovery potential, workload distribution, geographical coverage, and operational efficiency becomes increasingly difficult.
Rulex developed a solution combining explainable AI with mathematical optimization to identify the most effective recovery strategy for each debtor while generating the best overall assignment across the collection network.
From recovery strategies to optimized assignments
Working alongside the bank’s internal debt collection team, Rulex began by analyzing historical NPL data to understand which recovery strategies had delivered the best results over time. The objective was not simply to predict which debtors were more likely to repay, but to identify the recovery approach most likely to maximize success for each individual case.
Using Rulex’s explainable AI, the platform uncovered the patterns associated with successful recoveries and translated them into transparent if-then rules. Instead of producing opaque predictions, every recommendation could be understood and validated by business users, giving debt collection teams greater confidence in the suggested recovery strategy.
Once the best strategy had been identified for every debtor, the focus shifted to portfolio allocation. Rulex’s proprietary optimization solver evaluated the portfolio as a whole, weighing recovery priorities, collector capacity, geographical distribution, debt concentration, and operational constraints simultaneously to identify the best overall allocation across agencies and collectors.
Because optimization criteria remained fully configurable, business users could easily adapt the allocation logic to changing business priorities, distinguish between mandatory and flexible constraints, and compare multiple what-if scenarios before generating the monthly assignments. Rulex Platform supported the migration from the bank’s existing environment, ensuring a smooth transition without disrupting operations.
The project also established the foundation for future NPL management through Rulex Studio’s interactive interfaces. Combined with Rulex’s product training services, this enabled the bank to progressively manage, refine, and evolve the solution independently as business needs changed.
Building a more effective NPL management process
The solution developed with Rulex Platform transformed the entire process:
- Every debtor was matched to the recovery strategy with the highest probability of success, based on historical NPL data.
- Large portfolios were optimally assigned across agencies and collectors, balancing configurable business objectives with operational constraints.
- Monthly assignment preparation time and production cycles were significantly reduced.
- Over a three-year period, the bank achieved substantial growth in both recovery performance and economic value.
- Through Rulex’s training services, the bank became fully autonomous in maintaining and continuously improving the solution.
- The project positioned the bank among the first financial institutions in Italy to enter the NPL market.
Tech point
Take a closer look at the core Rulex Platform capabilities behind this NPL management solution
01. Explainable AI for recovery strategies
Rulex’s proprietary explainable AI automatically generates transparent if-then rules directly from historical data, replacing opaque predictive models with explicit decision logic. Every recommendation can be inspected, validated, and refined by business users, improving trust, governance, and regulatory transparency.
02. Build/Solve
Build/Solve is Rulex’s proprietary optimization solver designed to model complex business problems involving multiple objectives and constraints. Business users can define optimization goals, distinguish between mandatory and flexible constraints, and generate the solution that best satisfies the organization’s overall objectives rather than simply identifying a feasible allocation.
03. Configurable optimization
Optimization models can evolve together with business priorities. Through Rulex’s no-code environment, business users can update objectives, modify constraints, and compare multiple what-if scenarios without rewriting optimization models or relying on software development.
04. Data orchestration and migration
Rulex Platform collected and harmonised NPL data while supporting the migration from the bank’s existing environment, creating a reliable foundation for optimization with minimal operational impact.
05. Interactive Interfaces
Rulex Studio provides custom interactive interfaces that simplify portfolio management, scenario evaluation, and optimization monitoring, giving business users greater visibility and operational control.
Why Rulex for NPL management
Managing non-performing loans is not only about identifying the best recovery strategy, but also about executing it efficiently across thousands of debt positions while balancing multiple business constraints. Rulex Platform combines explainable AI with mathematical optimization to help financial institutions improve debt recovery, optimize portfolio allocation, and give business users full control over the decision-making process through a transparent, no-code environment.
Explore more financial services case studies powered by Rulex Platform
Build an Early Warning System
Timely identifying signs of potential risk in open​ banking data.

Automated Wealth Management
Automating the creation of tailored portfolios.

Predict Customer Churn
Timely identifying clients at​ risk of churn to implement corrective actions​.

Dynamic Pricing for E-commerce and Retail
Optimizing pricing with explainable, human-supervised automation.

Rethinking Insurance Fraud Detection with Explainable AI
Accurately identifying fraudolent claims with limited historical data​.

Optimize Non-Performing Loan Management
Improving NPL strategy and allocation across collectors and territories​.

Frequently asked questions
How does Rulex improve NPL management?
Rulex combines explainable AI with mathematical optimization to support both strategic and operational decisions. Explainable AI identifies the recovery strategy most likely to succeed for each debtor, while the optimization engine determines the best overall portfolio allocation across collection agencies and individual collectors, taking multiple business objectives and constraints into account.
Why is the Build/Solve task imporant for NPL management?
The Build/Solve task lets organizations model objectives and business constraints – violable and non-violable alike – in plain language, explore alternative scenarios, and generate optimized decisions without manually evaluating thousands of possible combinations. Rather than simply finding a feasible solution, the Build/Solve task is designed to identify the decision that best satisfies the organization’s overall objectives.
Can business users adapt the allocation process?
Yes. Rulex Platform enables business users to configure optimization criteria, distinguish between hard and soft constraints without programming. This allows allocation strategies to evolve as business priorities and portfolio characteristics change.
Why is explainable AI important for NPL management?
Rulex’s explainable AI helps financial institutions understand why a particular recovery strategy is recommended for each debtor. By generating transparent if-then rules rather than black-box predictions, Rulex enables business users to validate recommendations with confidence before optimizing portfolio allocation across collection agencies and collectors.