Solutions
Credit scoring, CRM, Core Banking, Analytics, BI, APIs.
About
Careers
Resources
What’s the value of a solution that’s never adopted?
The answer, unfortunately, is nothing, which is why at Rubyx we’ve been keen to address the challenges in adopting our credit scoring solutions. As a result, we’ve developed a proven approach to building trust and getting buy-in from key stakeholders – loan officers – for our credit scoring solutions, so we can make sure our lender clients can fully benefit from the solution we’ve developed with them.
In order for loan officers to trust in the credit scoring solution, they first need a) proper technical understanding, and b) an alignment of results to prove that it actually works – seeing is believing. So on top of technical training to help them, in many cases, move from expert scorecards to algorithmic scoring, we’ve introduced dedicated alignment sessions with loan officers where we can compare the results of our scoring algorithms with their own analysis.
During these sessions, we request loan officers to randomly select customers they consider good, average, and risky. We then compare their assessments with the results generated by our algorithms.Initially, loan officers are sceptical, as is natural when integrating new technology into established workflows. However, over time, they begin to realise that our solution consistently aligns with their own assessments in the majority of cases
Of course, there’s never 100% alignment. In instances where there is a mismatch between a loan officer’s evaluation and our algorithm’s results, we invest time and effort into understanding the underlying reason. And the reason is often that an officer’s personal familiarity and long-term history with a customer has led to subjective assessments that don’t reflect that customer’s objective patterns of behaviour. Uncovering these biases together is invaluable for fostering trust in our deployed solution.
Conversely, if the algorithm produces inaccurate results, it could be due to recent or specific customer activities that haven’t been captured in the data used as input of the model. For example, financial difficulties at another institution might not be accounted for. This is precisely why we provide the opt-out feature, enabling loan officers to exercise their expertise and reject customers whom they perceive as risky despite the algorithm’s recommendation to proceed with disbursement. It’s important to note that customers who have been opted out can easily be opted back in once the situation becomes safer. Furthermore, these sessions prove to be an excellent opportunity to collect feedback and identify potential areas where business rules don’t fully take into account local market specificities. By actively involving loan officers in this iterative process, we ensure that our scoring system accounts for their expertise and local market nuances, enhancing its accuracy and relevance.
These collaborative sessions play a pivotal role in achieving successful solution adoption and establishing trust in our credit scoring technology among the professionals who deal with it on a daily basis. By bridging the gap between algorithms and human expertise, we empower loan officers to embrace our solution with confidence, knowing that it complements their knowledge and adds value to their decision-making processes.
We believe that building trust and promoting adoption go hand in hand when implementing new technologies (including automated loans). By actively engaging stakeholders, addressing biases, and incorporating feedback, we can create an environment where credit scoring algorithms are seen as trusted allies, enhancing the efficiency and accuracy of lending decisions.
