Millions of people apply for new loan, credit card, mortgage, auto/life insurance etc every day. Companies spend lot of time and money in application verification and eligibility determination. Lot of fraudsters also apply and companies need to detect the application fraud to avoid losses. Lot of companies perform fraud determination manually, which is very ineffective, costly and time consuming process.
People submit applications using different channels like website, mail, mobile, agents etc. Few of the application details they need to provide are: Personal details - Name, Gender, Birthday, Email, Phone, Address, SSN ... Household details - Owner, number of dependents Professional details - Occupation, Salary, Office address, Phone ... Asset Details - Car, House, Boat, Motorcycle.. Liabilities - Mortgages, Loans.. Credit worthiness : Scores Health Details : Vital readings, Medical history, Prescription drugs... Once companies collect the application information they process the details to perform following tasks: Determine eligibility of the customer Calculate risk & predicted profitability Estimate worthiness of assets Estimate likely hood of early claims Estimate premium, loan amount, credit line etc. Accept or deny the application
Fraudsters apply for loan, insurance, credit cards etc with the intention of defrauding the financial institutions. They provide wrong or engineered information so once they bust out they can't be caught. They change application details to open multiple accounts or join a legitimate account to build good history. They provide wrong information about their assets, health etc to get benefited from fraudulent claims. They often are part of fraud rings, where many fraudsters collude in a group. Banks, Insurance companies should detect this application fraud early on to avoid the losses.
InsightLake application fraud solution uses a tiered detection process to effectively determine fraud. The solution is designed for different industries like Finance, Healthcare, Insurance etc. Further models and flows can be easily customized using InsightLake's intuitive UI.
Validation step allows checking of application details, whether they are complete and accurate. Business rules/thresholds can be provisioned to accept of reject the application.
Link analysis step allows ops team to perform fuzzy matching with existing fraud application database in an automated fashion. Step detects the linking of application with known fraud entities, and generates alerts and move the application
to fraud watch list.
Solution provides exact or partial matching scores. In case of exact match application can be put into deny list. In case of partial match, thresholds can be placed for acceptance or rejection based on PII field criticality.
InsightLake solution enables ops teams to calculate risk scores for each application in automated way. They can explore the applications easily to investigate similar risky application segments. In case of Property & Casualty or Life Insurance they can use additional models to include the asset risk scores as well.
Prediction step uses machine learning models on new applications with fraud database to predict likely hood of application fraud. Model precision thresholds can be used to put new applications in the fraud watch list.
