Analytics platform: How the EOS Group is optimizing its international debt collection processes.
Debt collection works differently in every market, but despite this all EOS national subsidiaries of EOS can learn from one another. The key is the new analytics platform. Thanks to machine learning, every group company can learn the appropriate courses of action for their collection processes.
- International knowledge management: collectively smarter thanks to the analytics platform.
- The EOS national subsidiaries use the platform to optimize their debt collection processes.
- The platform has already been used successfully in a pilot project in France.
What can be so difficult about debt collection? For outsiders, the process seems straightforward. If you don’t pay you get repeated reminders until payment is received. If not, you’ll find yourself answering to the courts. “It is absolutely not that simple,” says Joachim Göller. “Do I send a dunning letter? An email? Do I make a phone call? How much extra time do I give the debtor to pay? The best action to be taken next in a collection process can be completely different from case to case.” Or from country to country: “What works in Germany can be entirely the wrong thing to do in France.”
What we are building here is something like an EOS ‘super brain’.
Optimizing processes and becoming collectively smarter
There’s probably no other company with better data about what works where than EOS, with its 26 national subsidiaries. And it’s Göller’s job to make this data even more useful. Göller is the Head of the Center of Analytics (CoA), a team of data scientists and IT experts that is working on an analytics platform.
Through the platform, the CoA is providing a technology that local analysts can use to develop and deploy their data models quickly and easily. In addition, the colleagues in the analytics community consult closely with one another on models and best practice.
The vision is for the national subsidiaries to become collectively smarter by handing over their data – anonymized of course – to the central platform. With the help of machine learning algorithms, the platform can then evaluate thousands of country-specific collection processes and detect patterns, e.g. to determine which collection step works particularly well under which circumstances. These insights then flow back to the EOS national subsidiaries for use in their own debt recovery processes. “What we are building here is something like an EOS super brain which will be fed with data from our international colleagues. The systems in the various countries send a query to the AI of the analytics platform, which can then provide information, for example, on individual payment amount and probability of collecting a specific receivable.”
Analytics solution in pilot phase in France
A milestone has now been reached. Since last October, EOS in France has been connected to the analytics platform. Our colleagues in France work with the OYO (Optimize Your Operations) core receivables management system, which is tailored to market conditions in France – for every defaulting payer from an initial test client they can now retrieve information on the expected payment in real time, on the basis of which they determine the best actions to take next. “Previously they could only make predictions for entire customer segments,” says Marianne Hügel. In her capacity as Senior Manager Business Development & Consulting she is working with her team and colleagues in Western Europe to implement innovative and analysis-driven processes in the region. “We want to provide all colleagues with technologies and methods that make even faster and more reliable predictions.”
The initial experiences have been very positive, says Hügel: Already there is evidence about how good the new process is. The “Triumph” project tested out the combination of analytics platform and analytical process control for a specific client. Only the calculated score determines which steps are to be taken and when the collection process is stopped. The outcome of this first benchmarking is that the data-driven collection process was more efficient and cost-effective. The number of cases was gradually expanded, and meanwhile this client is managed entirely based on the predictions of the algorithm.
By creating important synergies with our colleagues worldwide we will change the way we solve problems in the same way as we use new technologies. We are going to become much more efficient.
The system keeps learning – and so do the people using it
“This pilot project means a lot to EOS in France because we are very interested in synergies within the EOS Group,” says Laurent Redois, Head of Statistics at EOS France. “No country can be so successful alone, and it is important that we share expertise and technology.”
Nicolas Cabaj, Head of Department of Statistical Studies at EOS France, is also delighted about the knowledge-sharing process. “The close collaboration has taught us to work with modern tools and methods. Although all members of the project team work in different cities, we managed to achieve quick progress and effectively share information.” Laurent Redois is already looking to the future. “We aim to stay competitive, so we are moving away from processes set in stone towards a constantly evolving work flow that is regularly reviewed and compared and is based on the latest state-of-the-art in technologies and analytics. In the same way that our models learn we are also constantly learning.”
Please don’t hesitate to contact us if you’d like further information.
Photo credits: Luis Alvarez / Digital Vision / Getty Images, EOS