As specialists in streamlining KYC & AML processes, we were very pleased to sponsor and speak at ECN's Identity & KYC Summit US a week or so ago. Aimed at driving discussion and inspiring initiative among thought leaders in the field, the event in New York looked at “reworking the identity authentication and KYC processes, to ensure privacy, drive down costs and empower customers”.
Moderating a panel discussion on ‘Combining a User-Friendly Onboarding Process with Robust KYC Compliance’ was our very own Patrick Penzo. The discussion gave rise to some powerful insight, and one key theme was the way in which Machine Learning technology can help keep the two in tension. But don’t worry if you missed - we’ve summarised the key takeaways below.
What is Machine Learning?
Founded almost 70 years ago as a subset of Artificial Intelligence, it revolves around getting machines to learn and act the way humans do. Machine Learning looks set to become the most revolutionary tech of the next century, and is currently most commonly used in three ways: product recommendations (like Netflix or Amazon), Insights (using big data to spot patterns) and process improvements (using machines instead of humans).
User experience at the core
It’s also particularly beneficial in improving the User Experience, especially for Fintech and banking. There is now an increasing need for banks to focus on customer experience first. FinTech startups are now capable of offering quicker, cheaper and more convenient financial services than traditional banks, and as they become more common, banks need to move online in order to compete. It’s thanks to Machine Learning technology like ours that they’re able to do so.
The tech behind FinTechs
Onfido’s Machine Learning technology helps improve the customer experience by bringing formerly manual KYC processes securely online, offering frictionless form filling and preventing fraud. We use Machine Learning in 3 key ways:
Classification: identifying what kind of document is being uploaded by recognising template elements
Extraction: extracting the information in the document to populate the relevant fields, allowing for a more seamless user experience.
Verification: our OCR functionality is able to check internal consistency and detect fraudulent alterations.
A risk vs onboarding tradeoff used to exist - a person may have been impersonating someone else, so it was better to meet in person and do checks manually. This was lower risk, but the slow and inconvenient process increased dropout. Machine Learning has eliminated the risk vs onboarding tradeoff, and banking is now being digitally disrupted as new entrants challenge the legacy banks to serve the needs and expectations of ‘mobile-first’ consumers. That means we can now verify people's identities and keep banks KYC compliant entirely online, without ever needing to meet in person.
Banking the Unbanked
This can also help us bank the 2bn ‘unbanked’ worldwide. Traditionally, only complex and expensive software has been able to scan and compute identity documents. By contrast, Machine Learning technology will help us accurately identify documents which have been uploaded using commodity technology; all you’ll need is a basic smartphone and an internet cafe meaning that even people in developing countries can be background checked and onboarded to financial services online.
Roboadvisors are becoming more prevalent to help improve UX. While people used to have to pay a premium for financial advisors, they are now being brought online with Machine Learning technology that is helping facilitate one to many relationships, giving more people access to investment services.
Also coming to the fore are biometrics, helping improve the user experience is through fraud prevention capabilities. Rather than carrying a calculator-type device or remember an average of 19 passwords to authorise a payment, soon customers will be able to verify their identity on their phone with only their face. By 2020, 7 billion transactions will be authorised by biometrics annually.