The colossal amount of data being created and used globally has paved the way for increased fraud and data exploitation across industries. In the finance industry, companies involved in auto-lending and mortgage are struggling to address the growing misrepresentation of facts on applications, including falsified identities, overstated incomes, counterfeit pay stubs and other types of prevalent fraud, that are causing extensive losses for these organizations. PointPredictive helps automotive lenders, mortgage lenders, and online retailers tackle fraud and reduce first payment and early payment default losses that are attributable to misrepresentations on applications. PointPredictive’s CEO, Tim Grace, explains how they leverage artificial intelligence (AI) and machine learning to help companies deal with these challenges.
What are the major predicaments faced by customers and how does PointPredictive alleviate those challenges? Also, where do you see customer traction?
The biggest challenge for our customers is the limited number of tools available to identify factual misrepresentation and fraud in the applications that lead to early defaults. The other challenge is the high false positive rates in current fraud-scanning tools that lead to increased consumer friction in the application process and increased operational costs for lenders. PointPredictive’s solutions identify both fraud and misrepresentation early in the application process with very low false positive rates, making it possible to reduce losses and, remove friction for low risk applications and lower lender and dealer operational costs. Our solutions help lenders with applications at both ends of the risk spectrum–for high-risk applications, we provide specific indicators of fraud mitigation actions to take, while for low-risk applications, we create opportunities to streamline the underwriting process to remove friction for consumers and dealers and increase the likelihood of conversion into a funded loan.
We are currently getting great traction in auto lending. This industry is prone to multiple types of fraud attack: synthetic identity, overstated income, falsified employment details, etc. These lenders are looking for fraud and misrepresentation assessment solutions as well as laser-focused point solutions that are holistic and accurate.
Could you explain your key offerings to different types of lenders?
Powered by machine learning and advanced analytics, PointPredictive solutions support lenders in their efforts to counter fraudulent activities. Our Auto Fraud Manager solution has the ability to instantly scan each application to precisely identify misrepresentation and fraud that is likely to result in an early default. The risk of an application is measured on a scale ranging from 1 (low risk) to 999 (high risk); when the score exceeds a lender’s tolerance level, the lender takes preventative action to verify the authenticity of the information provided by the applicant.
To effectively deal with regulatory compliance issues, PointPredictive’s DealerTrace solution continuously evaluates auto dealers to identify those that pose a high risk to lenders.
What drives our solutions portfolio is our talented team. Our data scientists and fraud strategists are the foundation of our value proposition to our customers
The process of scanning each application received ensures that the dealers are constantly checked for authenticity and that their DealerTrace risk scores are current. This risk score assesses both current dealer risk and historical dealer risk in the context of the current application.
In addition to these two holistic solution offerings, our customers have asked us to create point solutions that focus on specific fraud types. We have recently rolled out two of these. Our Synthetic ID Alert™ solution can score every application and determines if the borrower or dealer has created a potentially fraudulent (synthetic) identity (one where the individual identity components are valid, but the combination does not correspond to a real person). Our Income Validation Alert provides a quick assessment of the likelihood that the income provided by the applicant is materially overstated. Both solutions enable lenders to reduce their spend on high-cost external verification services for identity or income by solely focusing on only those applications identified as high-risk.
When organizations share their confirmed fraud data and details of how they were affected by a certain fraud, this information helps other organizations to be wary of falling victim to the same fraud scheme. Our Fraud Data Exchange was built to foster industry-level information sharing, where auto lenders contribute to a shared industry watch-list based on prior fraud experiences. Lenders share phone numbers, addresses, and other information associated with systematic fraud schemes. PointPredictive manages the repository and provides alerts to contributing participants while ensuring that all the contributed data remains private and secure.
What keeps you ahead of the competition in this competitive market?
Leveraging machine learning to solve misrepresentation and fraud risk sets us apart from the competitors who usually rely on checking databases to identify inconsistencies. Database checking is very prone to high false alerts due to typographical issues or incorrect database data. Machine learning ensures accuracy in risk analysis, limiting database-checking to only those applications that are identified as high risk. In head-to-head trials, we have proven that our ability to identify early payment default attributable to fraud/misrepresentation is greater than 50 percent (more than twice that of current solutions). We have benchmarked our scoring technology at close to 5,000 transactions per second; this means that lenders can confidently insert our solutions into their real-time underwriting workflows without delaying their loan decisions. Additionally, our solution has a proven track record of identifying all different types of fraud schemes owing to the broader spectrum of risks covered through our holistic machine-learning approach. Finally, our solutions are built specifically for the industries they serve and therefore, utilize a wide array of specific data from the application records for those lending types.
What drives our solutions portfolio is our talented team. Our data scientists and fraud strategists are the foundation of our value proposition to our customers.
Could you share a case study that highlights your value proposition?
One of our clients, a large subprime auto lender, was losing $100 million per year on loans that defaulted within the first six months. This was primarily a result of the fraudulent misrepresentation in applications that were being approved and funded. In a retrospective test using our Auto Fraud Manager solution, we identified nearly 50 percent of these losses that could have been avoided. For this lender, it equated to an annual savings of $49 million and dramatically increased their anticipated profit margins in new loans going forward. While focusing on less than 10 percent of the applications, the lender can reduce their losses related to misrepresentation and fraud losses by over half and create strategies to reduce friction for consumers and dealers in the low risk population to increase loan capture rates.
How does the road ahead look for PointPredictive?
From a revenue standpoint, we believe we can double our revenue each year for the next five years by focusing on our existing markets with our existing solution portfolio and new, value-added services. From an R&D perspective, we are always exploring improvements in our underlying predictive technology as well as identifying new markets that can benefit from our machine learning approaches to fraud and misrepresentation.