In the lending industry, the needs of customers and banks are straightforward: banks want to increase their clients’ acquisition and retention rates, and customers want to get better credit by way of an easy and comfortable user experience. Both want to do so while preserving the security and privacy of all data. There are new and immense opportunities opening up in this direction, thanks to the use of open banking data and innovative analytical tools.
What it does:
How does N.E.O.S work? First, the “access to accounts” component enables the bank to obtain customer consent, and then connect to any account, thanks to the use of either API - when made available by banks - or web-scraping technology. Second, once access is established and data is retrieved, the NEOS proprietary AI-based Categorization engine assigns a specific category to each of the thousands of transactions present in the selected current account. This process is fundamental to prepare the data for the third phase, in which a score engine processes the categorized transactions and provides credit risk scores and targeted commercial KPIs.
How we built it:
The project kicked off at the end of 2018 with a diverse team comprising consultants, data scientists, and transaction data experts. Availability of a data pool comprising transaction information from over 20 banking institutions Enrichment Engine trained on over 5 million current accounts and 1 billion transactions. Continuous cooperation between the advanced analytics team and data experts to improve the model's learning curve. Categories designed to optimize the assessment of creditworthiness and business opportunities. Continuous learning through a back-end system works proactively on transactions with a low confidence level and on performance monitoring process.
Challenges we ran into:
Education of applicants on the Open Banking benefits, Drop out rate, Lack of Standard API, Market regulation specificity,
Accomplishments that we're proud of:
Significant improvement in the acceptance rate, at parity of expected risk Ability to identify low-risk customers among new to credit subjects Improving the customer experience, thanks to the widget design (totally customized) that gives to the customer the impression of navigate the same web pages Full digital experience, thanks to a landing page, without going to the bank branch
What we learn :
Preempting potential questions and generating a sense of trust in the process are essential to increasing the success rate. The offer of clear and tangible benefits in exchange for access to account data is a determining factor in obtaining user consent. A simple and smooth user experience, which guides customers through their consent and linking bank accounts reducing the drop-out rates during the initial step of the process.
What's next for N.E.O.S - New Evaluation Open Suite:
Audit functionalities enhancement, Widget improvement for a better user experience, Demo website enhancement - creation of an operational N.E.O.S client, Development of automated report and console for monitoring end user widget utilization, Enhancement of N.E.O.S web portal, Improvement of a new set of administration functionalities, Development of a significance Index to identify the "main account", Continuous evolution of Machine Learning categorization models, Taxonomy fine tuning by jurisdictions.