Revolutionizing the concept of collections

Implications: Data, Machine Learning, IA, Banking, Financial Services, Advanced Analytics

Understanding default as an important moment in the relationship with clients

The context

The collections area is one of the main business levers of a financial entity, however the circumstances of its procedures are such that they’re guided by a standardized approach with technologies that can be improved upon. It was a huge challenge: their ability to collect was falling by 1.5% per month and they were unaware as to why.

Our approach

To reexamine the automation systems of the entity’s collections and recovery model with a very open mind: simplifying it, adopting the client’s point of view, co-creating solutions with the team, and applying exponential technologies to generate efficiency with a direct impact on EBITDA. 

call center recobros The Cocktail

The project

We uncovered the real reasons behind what was causing the deviation in collections and recovery, identified which ones were key, and acted on them by coming up with more than 100 initiatives for change.

We prioritized 15 concept trials that generated the highest impact and launched them to control groups with real clients.

We sped up the implementation process of the trials, setting up parallel systems to the official ones, allowing us to achieve results in 6 weeks: we put fake webs,  independent accounting systems and new call center scripts into production, and created a predictive model alternative to that of the entity by using machine learning technology.


We went from a standardized collections system, to a new client relationship model, guaranteeing the personalization of the debt journey.

A new predictive model based on machine learning, capable of improving client categorization and demonstrating that:

  • More than 93% of Low Risk clients pay without intervention.
  • Specific communication items at the time of contracting services reduce the probabilities of falling into arrears by more than 62%.
  • Small action alerts reduce falling into arrears by almost 20%.

In short, we’ve achieved a relationship model that’s much more cost efficient, and that has a positive impact on the client: more EBITDA and improved NPS.