The Challenge
Today, modern “digital” banks in particular offer services such as “digital wallets with attached digital payment cards” or “small loans”, very quickly.
To do this, a decision must be made quickly to facilitate onboarding procedures and quickly start an engagement with the customer.
The request was to improve Credit Scoring during the Onboarding phase and provide risk indicators related to the good and bad payer during the life of the credit.
The Solution
The Bank normally follows the credit history of each customer in order to identify the bad payers. But during Onboarding the “internal” information available to give a score to the customer and predict a behavioral profile to understand if he will be a good payer are scarce.
The solution was to determine a scoring and a behavioral profile with Machine Learning technologies based both on internal data of behavioral similarity, anti-fraud, use of e-money and credit services, and on external data such as Adequate Verification procedures for Anti Money Laundering (AML) purposes, geographical data, Istat data on income, on well-being, etc. or other external credit risk indices provided by Rating Agencies.
Models of different nature Supervised and not have been applied both in the onboarding phase and in the monitoring phase of the life of the Credit.
The Benefits
The main benefit of this solution is the ability to use the results and scores achieved with Machine Learning models during the monitoring of the Customer Base Credit, in order to give added value to the rapid identification of which onboarding procedures to apply to the entry of a new Customer into a new service.
Humanativa’s Role
Realization of the end-to-end Machine Learning service, from the creation of Machine Learning Models, Integration into Business Systems, MLOPS Procedures (with ApiRest & microservices), and Advanced Analytics.