During our KNOW Identity Digital Forum, Making Money Move, our stacked lineup of panelists received so many important questions they couldn’t answer all of them on-air. Never fear! We followed up with our expert speakers and captured their thoughts on your burning questions.
That “X” Factor
To kick off the event, we hosted a cross-industry conversation on “KYX,” a term expanding customer verification to the needs of banking, healthcare, telecoms, and beyond. In “That ‘X’ Factor: Cross-Industry Applications of KYX“, we heard from Jason Aspinwall of Advent Health, Franklin Garrigues of TD Bank, and Philipp Pointner of our partner, Jumio. Their discussion sparked some additional interest from our audience, which Pointner expanded on further.
OWI: How does the work of AAMVA, ICAO, and within ISO 18013-5 fit into the future of Identity Verification of an individual in the context of digital identity?
Philipp Pointner, Jumio: Any government issues form of digital identity will be very welcome by the Identity industry. It takes away the need to scan and analyze a plastic or paper ID and makes it usable online in a more direct fashion. It will be a highly fractured ecosystem where different countries are issuing in other formats and standards. As an identity solution provider, it will be our job to make sure those variants that see user adoption are accepted by us with the same ease of use as the scanning of plastic today.
OWI: In the context of Digital Identity, does Self Sovereign Identity become the answer, and to assure success, who must be the anchors of SSI credentials?
Pointner: I’m personally a big fan of SSI. The privacy and data ownership that it grants to the individual consumer is really important. That said, the current concepts are quite far apart from the needs that businesses and governments have and how identity is regulated.
It seems that the vision of the end goal of SSI is very compelling, but too little is invested in finding out how we can go from where we are today to that final stage.
OWI: Why does Jumio think the mobile phone will be the device to store an individual’s ID?
Pointner: To be precise, I meant the phone will be the device we use to authorize and control our digital identities. I didn’t mean to make a claim on where it will be stored. The phone is going to be the conduit because of convenience and user adoption. I think in the short term, there might be 3rd party identity apps. In the long run, it will be on the operating system level. Why do I think that? All digital identity schemes that require specialized hardware are not going to work. It is too cumbersome for the end-user. The phones are what everyone is carrying around all day long.
OWI: What is the one thing you are excited about in ID Proofing? What is one thing that keeps you up at night?
Pointner: I’m excited that companies start to understand that data-driven KYC alone is no longer enough and needs to be paired with credential driven KYC. I am constantly thinking about how to further eliminate online identity fraud.
A Picture is Worth 1,o00 Extracted Words
Following that session, our partner Microblink took the stage with “A Picture is Worth 1,000 Extracted Words.” Their discussion between Managing Director Izet Ždralović and Capital One’s VP, Identity Services, Pranav Khanna, tackled how document verification innovations can reduce friction for consumers and manual interventions for companies. Microblink provided additional insights to follow-up questions from attendees.
OWI: Often, we talk about verification for consumer solutions in banking, but how can identity solutions transition and evolve to verify business clients and customers?
Izet Ždralović, Microblink: Verifying a business is not too different from verifying a customer. A major part of it is user verification of the Ultimate Beneficial Owner (UBO), the person who owns more than 25% of the company. Here, we can still employ all of the techniques we’ve discussed, including ID scanning and document authentication.
However, business verification entails other steps we have to take to minimize risk and perform our due diligence. Applicants are often asked to submit documents that verify the company’s registration, and that information has to be periodically updated. These records shouldn’t be mailed or brought into the office – a business client can simply scan them in through a mobile or web app.
This way, we’ve drastically improved their onboarding experience, but we’re now also able to verify the extracted data against relevant registries.
OWI: What’s your suggestion on testing: how to keep it lightweight without too much tech work before realizing it’s an effective solution?
Ždralović: The answer to that question largely depends on what you’re testing. As a rule of thumb, you should start with your users and the way they’re planning on using a piece of technology. Download a demo app or develop a sample app in-house and start testing different usage scenarios.
Once you’re happy with the first version, roll it out as an MVP to a small chunk of your user base. Gather feedback from the early adopters, improve and repeat with a beta release. Once you’ve locked up the features, resolved bugs, and polished UX, you can proceed with a public release.
At Microblink, we’re trying our best to give people a chance to thoroughly test our technology before going to production. First, we have free demo apps for both Android and iOS devices. These apps aim to showcase the best of our technology and are free to download.
Second, we keep the integration sample code of all our products as bare-boned as possible so that developers can try them out in their app in minutes. But remember that, for any further refinements to usability and functionality of your final app, you’ll always have to roll up your sleeves and test for any defects and inconsistencies. High-performance software operation can only be achieved through detailed testing.
Learning the Machine
Finally, in our closing session, Learning the Machine, experts Stacy Stubblefield & Dusan Bosnjakovic from TeleSign dove into how machine learning can streamline fraud prevention processes.
OWI: Any thoughts on where to go for accurately labeled training data?
Stacy Stubblefield, TeleSign: Typically, our clients use their own data for this. If they already have accounts with verified phone numbers, they’ll select accounts that are relatively aged (maybe 6+ months) and apply labels based on whether the user has been found to be fraudulent or not. Suppose the client does not already have verified phone numbers on accounts. In that case, they can start verifying phone numbers while using our generic Score model and then customize the model when they begin to know the results (several months later). Also, note that our generic score model uses labeled training data across our client base to score numbers. Therefore, by using the generic model, clients are in effect using labeled training data (it just isn’t customized for that particular client).
Dusan Bosnjakovic, TeleSign: I’ll just add that labeled data is time-consuming to produce, often guarded by companies for privacy and security reasons, and varies in quality and meaning from company to company. This is why it is challenging to obtain a lot of high-quality training data. TeleSign’s approach is to create incentives for customers to provide the data by constantly improving our solutions and by creating customizations for customers based on that data.