Carrefour Banque et Assurance is a branch of the Carrefour Group. For over 30 years they has been providing a wide range of products such as consumer credit, savings and insurance, through both physical and digital channels.
Carrefour Banque's financial risk teams decided to reduce residual fraud as much as possible. Their existing tools did not allow them to tackle this project alone. That's why they decided to call on Bleckwen.
Within 10 months, we were able to deploy, in collaboration with their teams, a predictive model to identify requests that were proven frauds and would lead to a payment default.
The results were excellent: a few months after implementing the model, we noted :
- 80% reduction in residual fraud
- 30% reduction in fraud attempts
The positive impact on business value is obvious. In addition, employee time has been freed up for other, more useful tasks.
Fraud is now identified much earlier in our process, and checks are carried out much more quickly. All in all, we have saved the equivalent of 1.5 full-time staff in credit application processing.
As a subsidiary company of Crédit Mutuel Arkéa, Financo pays the utmost attention to its partners to develop financing solutions tailored to their customers' needs, while offering them a smooth and unique experience. Financo has recently become a mission-driven-company, with the aim of contributing to a more sustainable economy.
Concerning Financo's business activity, like many other players in the car and motorbike financing, they are exposed to credit fraud. That’s why Financo called on Bleckwen to combat credit fraud and reduce its financial losses. The team already had a fraud detection model in place, but it was very manual. The aim was to replace this existing model, to enable fraud requests to be automated and identified in real time.
Bleckwen develops AI-powered lending fraud prevention solutions for financial institutions to protect them against financial crime. Our tailor-made models are trained to detect in real-time fraudulent loan applications, allowing lenders and retail banks to:
- Optimise the number of fraud alerts generated while accelerating fraud investigation and decision thanks to our comprehensive fraud scores and AI explainability
- Significantly reduce their fraud losses by an average of 80% and increasing acceptance without adding more friction to the underwriters or loan applicants
- An ensured 400% ROI after the implementation of our solution.
Since 2019, we have been identified as leaders in loan fraud detection. Among lenders and retail banks we already serve; BNP Paribas, Mobilize France (Renault-Nissan alliance) and Stellantis (formerly PSA Finance).
We believe that artificial intelligence is an opportunity to make the financial world a safer and fairer space by protecting financial institutions from fraudsters’ constantly evolving threats and limiting the use of defrauded funds into criminal activities.
To maximize the chances of the project, Bleckwen proposed to Financo to co-build their solution and to verify its feasibility through various audit stages.
This approach combining real-world conditions and a limited scope, has led to the emergence of an analysis and scoring model suitable to both Financo’s needs, and its clients' specific profile. It is now faster to identify suspicious files, with far fewer false positives.
Since the implementation of the API’s solution, the main benefits for Financo are :
- Detecting fraud in real time
- Delivering frictionless customer service
- More effective identification of fraud (false-positives)
- Realigning the objectives of the various departments
In addition to the ML, Financo benefits from privileged information, delivered by Bleckwen, on the behaviour and trends of fraudsters, in order to protect itself as effectively as possible.
Performance figures aside, fighting fraud is also about making the world a better place.It is difficult to know exactly what the criminals plan to do with the stolen money or vehicles. However, following the complaints lodged by the banks, the information received by the police stations allows us to see things a little more clearly:
- terrorist organisations
- prostitution rings
- and arms purchases are the main verticals financed by stolen money
Following a conclusive audit phase, a successful go-live, and a trust-based collaboration, we can only hope to continue in this way for many years.
Fraudsters will constantly innovate their fraud attempts. So, lenders must have a multi-layered fraud shield that is robust and adaptive enough to the evolving nature of such threats.
Without implementing solution like Bleckwen, this quest is a high-performance hoover, on human and financial resources, with relatively few results. With a machine learning-based solution, this maintenance process can be carried out with minimal effort and maximum ROI.
First of all, we win responsive in the ability to detect fraudster profiles since we now have a solution that runs in real time.
We also have better discriminating power and therefore better efficiency in detecting these fraudster profiles that interfere with our grant decisions.
And finally, this project was a catalyst for us to adapt our processes and all the acculturation of our teams, especially commercial teams, around fraud.