Can artificial intelligence and machine learning improve collections department performance? | Flexys

Can artificial intelligence and machine learning improve collections department performance?

9th January 2019 by Paddy Gilling

How emerging technologies can increase the amount of debt collected while driving down costs and maximising efficiency.

A simple definition of machine learning

Machine learning is an application of artificial intelligence that provides systems with the ability to automatically learn and improve from experience without being explicitly programmed. Machine learning focuses on the development of algorithms that can identify data patterns which can then be matched against individual customers or agents data to determine an optimal action.

Flexys Labs

Flexys Labs is the research and development division of Flexys Solutions using deep domain expertise in Machine Learning (2nd generation AI) and Contextual Adaptation (3rd generation AI) to enhance collections performance. The team is driven by a natural curiosity and enthusiasm for process improvement backed up by a rigorous scientific approach to data and analytics. Flexys Labs is lead by research engineer Don Liyanage, who alongside his work at Flexys, is studying for a PhD in Neural Network Reasoning Using Case-Based Reasoning. Don is joined by research consultant Pedro Ferreira whose role includes explaining the business benefits of machine learning to clients across every sector and industry. Postgraduate software Engineer Keiran Sears completes the team.

Accountability and security

By carrying out artificial intelligence research and development in-house, Flexys can monitor every aspect of the process. The Lab works to high ethical and regulatory standards including using fully anonymised data to comply with GDPR. The integrity of the algorithms is protected by a rigorous checks and balances protocol to ensure predictive results are genuine, relevant and continually updated. In addition, the team can work with clients to adapt the machine learning applications to suit changing business demands.

Applying machine learning to the Flexys suite of solutions

The Flexys Cognition solution is a highly scalable microservices platform to automate strategy optimisation, identify customer segments, define sentiment and identify vulnerability. Cognition optimises early-stage collections by segmenting self-cure customers from ‘won’t pay’ customers and directing both to the most appropriate course of action with minimal agent oversight. It also aids the identification of customers experiencing vulnerable circumstances and helps prevent such customers slipping through the net and receiving treatment that does not meet regulatory obligations. Cognition can further be used in late-stage collections to match an account’s characteristics to a DCA’s core competencies to ensure the best collections outcome in terms of cost and debt collected. 

Cognition is delivered as a SaaS cloud solution, which can be implemented and delivering improved collections performance within a matter of weeks. It works seamlessly with legacy systems and modern digital platforms as well as integrating with other Flexys modular solutions to deliver a complete end-to-end enterprise collections system. Cognition provides agents with a dynamically customised journey for each customer. Each step of the customer journey is determined by customer circumstances in full compliance with business and regulatory objectives. Agents can also visualise the best option for next-step engagement with the customer and alternative steps when needed. 

Measuring the benefits of machine learning

Any business that has a collections functionality has large data sets consisting of sales details, billing history, payment records, and customer information. Most organisations keep this data as a system of records for procedural and regulatory requirements. However, holding this data in the system requires additional costs including storage, back and recovery maintenance/procedure, compliance-based data protection and security procedures. Today, organisations can bring machine learning into core business processes that are connected with the business’s data streams to improve their decision-making processes through real-time learning and to turn a ‘cost’ into a benefit.


  • Lower cost to collect through more accurate treatment of customers.
  • Increasing recovery rates from late-stage debts.
  • Optimising values of debt sales.
  • Protecting your business reputation with customers and consumers.
  • Helps rehabilitate customers and build loyalty.
  • Reduce call centre workloads

Fast-tracking new developments

The Flexys Labs research includes a Knowledge Transfer Partnership with Professor Jim Smith, Head of Interactive AI at the University of the West of England.  Together, the UWE AI team and Flexys are developing ‘explainable AI’ capabilities for their debt collection software solutions that will allow organisations to achieve an unparalleled level of transparency.

*explainable AI produces models where specific actions are explainable to a “human-in-the-loop”, without sacrificing performance.

Read more about the Flexys suite of collections solutions at

Book a demo of our digital self-service Collaborate solution