Last week was very exciting at Socure as we announced our latest funding round. In the short time since this news went public, I’ve received a number of questions about Socure. People are inquiring about things such as what do you solve for, what is your market position, can you share your future goals, how do you utilize AI, etc.
Before I try to answer some of these questions, let me say that I hate the term “artificial intelligence.” It suggests that because it’s artificial, it cannot exceed the accuracy and performance of a human brain. While this has been debated academically, there is objective proof that the machine has advanced far beyond the human in certain situations.
What exactly does Socure do?
I love that Socure’s value proposition is so simple to explain. The first time that you as an individual interact with an entity (bank, retailer, government entity, or healthcare payer), we predict whether you are who you say you are—or a fraudster pretending to be someone else. Because this decision is made at the point of origination (opening the account) and has a known outcome, a simple regression test can prove whether or not our solution is more accurate than your current solution. It’s staggering how much more accurate the machine is than the human when it comes to predicting viability of a person’s identity.
In regression testing with all of the 140+ businesses that use Socure to verify identity, WE HAVE NEVER FAILED TO CAPTURE AT LEAST 50% ADDITIONAL THIRD PARTY FRAUD and frequently capture 80% more fraud than the human-developed rules engines that we replace. And yes, you read that right... Socure's machine learning platform ALWAYS captures between 50-80% more third party fraud than any of the incumbent providers that rely on rules-based logic to make their decisions.
How do we achieve 50-80% more third party fraud capture?
Third party identity fraud costs businesses at least $16 billion each year in the United States alone. This statistic indicates that the incumbent fraud detection and identity verification solutions that have been in use for the past decade or more are just not as effective as they need to be.
Enter Socure. Unlike the traditional serial-rules-based incumbents, we don’t just rely on the half dozen personal data elements you provide in your application when trying to open a new account. We’re not dependent on single-source, static databases. We leverage thousands of online and offline data elements, many collected from the digital footprint that you leave behind every day. The machine correlates these and detects fraud patterns that human-programmed “if-then-else” rules could never come close to achieving.
Why does it matter that Socure is so much more accurate?
Besides the need to safely accept first-time customers applying for loans and credit cards, and eliminating third party identity fraud, financial institutions must provide a great customer experience. The best identity verification solutions keep out bad actors without subjecting the good ones to unnecessary delays and information requests. A poor user experience means people walk away, which in turn means lost revenue. Furthermore, when applicants can’t be auto-approved, a manual review process means increased operating expense.
At Socure, we develop machine learning models that address these issues and solve the "Digital Identity Problem" on a scale required by the top 10 financial and retail institutions. But don't take my word for it. As our data scientists would say,"Words are just words...but data is proof!" And the data shows that our solutions have been able to eliminate at least 50% more third party fraud than the incumbent providers while achieving auto-approval rates as high as 92%.
How are you solving this problem (or “What makes Socure so special”?)?
In the world of Fintech, the word “disruption” gets tossed around a lot—often without merit. But at Socure, we disrupt the traditional methods of identity verification. We are a data driven company. All of our decisions have their foundation in the data. We source thousands of online and offline data signals created from the consumer's PII and broader digital footprint, then apply AI and machine learning methods to draw correlations (between data points) that are virtually impossible to identify with traditional statistical analyses. Are all your profile elements real? And if they are, do they belong together?
The industry incumbents continue to rely on legacy programming and human-based rules (and have for decades). These solutions are easily defeated as they rely solely on the consumer’s PII data—data which in many cases is easily available to criminals through the many data breaches that have occurred over the past few years.
So how does this differentiation result in market momentum for Socure?
Due to 10-20x measurable ROI provided by our machine learning platform, virtually all of the top US financial institutions are either already customers or are evaluating our results versus their incumbent provider. We have already processed 130 million American identities and expect to reach the full sample size of the adult American population by the end of 2019. At this point, our models will be very hard to beat. Our solution not only outperforms the incumbents in test, but gains accuracy in production as the machine learning models learn from our customer's feedback. A proof point is that we have not lost a single Fortune 1000 customer in the past 2 years. Our mission is to become the SINGLE trusted source of identity verification and virtually eliminate third party identity fraud.