I met a friend recently who is a product head at a well-known global fintech. He was very excited about a Figma design shared by one of the vendors. It was a user journey for digital onboarding. To say that I was surprised by this is an understatement. As some of you know, I worked for a Singapore-based company, Advance.ai. We used to and still design really exciting onboarding technology for fintechs. What my friend was showing was something that had already been implemented by a plethora of companies across the world. These companies have onboarded hundreds of companies as well.

What the above interaction highlighted to me was the fact that the availability of tech doesn't mean banks and fintechs have implemented it or are even aware of it. Therefore, this is the first of many posts that intend to highlight some of the key tech and the people building them. The content presented herein is based on my opinion, so please take it as such.

Digital onboarding solutions have gained prominence in the last 4 years. The majority of the countries produced with non-face-to-face onboarding regulations around 2020, accelerated by COVID-19. The technology is termed as electronic KYC, e-KYC, digital identity, and identity verification, which are some of the popular ones. The technology enables banks, fintechs, or any company to onboard customers remotely while still complying with regulations.

The simplest journey, shown below, includes capturing the customer's ID document and taking a live selfie of the customer. The journey is typically completed in less than a minute. While the customer journey looks simple enough, the tech behind it is fairly advanced.

The key capabilities of an e-KYC provider are as follows:

Identity document (ID) capture

This feature helps users upload the ID document photo. This step can include a simple photo upload or complex blur/focus/lighting/ID card type detection features. What works for you will depend on the ID Card type, incidence of fraud, and availability of ID card verification technology. This step also has front-end functionality, so you need to explore the customization capabilities of the vendor.

Identity document verification (IDV)

This is a backend check where you verify whether the uploaded ID document is genuine or fake. This is by far the most difficult step to accomplish owing to a variety of factors:

• The type of ID documents that are accepted (regulatory requirement) - passport, driving license, tax ID, etc.

• Each type of ID card would have many versions - SSN has 6 versions after 2000, the latest being in 2022.

• Each ID card type and version has different security elements - hologram, overlapping printing, background patterns, barcode, QR code, near field communication (NFC), etc. You can notice some of these elements in a sample US passport below.

• Different government authorities issuing ID cards provide different methods to verify the ID card.

You need to map your requirements with the capabilities provided by a vendor. There is no single vendor who is the best at everything. Some of the features that I have observed are as follows:

• Basic:

  • Conduct optical character recognition (OCR) to extract details of the ID card and check if the ID card number matches a certain pattern.

  • Front and back of the ID card belong to the same person (ask me about how a fintech saw a massive fraud that could be stopped with this simple check).

• Advanced:

  • Check if all the printed details are in the place they are supposed to be.

  • Landmark detection - logo, patterns, photo, hologram, chip, etc.

  • Printed font verification.

  • Matching the main face photo to the small face photo provided in some ID Cards (like the one above).

  • Barcode/QR code detail extraction and matching to printed details.

• Innovative:

  • Hologram verification (difficult to carry out in a non-face-to-face setting).

  • Chip reading using NFC.

  • I was amazed when I first saw the demo to detect the thickness of a Japanese ID card.

Some vendors use a team of analysts to manually verify the ID. This helps detect ID fraud that you have never encountered before. This service typically is more expensive and takes time to complete (SLA of 10 mins to 1 hour is standard).

This would be simplified if the government provides an API to verify a particular ID card. Even in this case, there are quite a few differences in how governments have implemented this API.

Liveness detection

This check ensures you are taking a live picture of the customer. The check is supposed to only allow a real-live person to be photographed. To accomplish this, the customer is asked to move their head left/right, forward and backward, open and close their mouth, close and open their eyes, smile, say aloud the numbers displayed, etc. Liveness detection takes a selfie of the customer during the process that can be stored in your CRM database. While smile-based liveness detection below looks cool, I do not recommend it, as the accuracy is not acceptable.

Some countries need fingerprint scanning. This is generally more complicated and there are only handful of players who can deliver this.

The liveness detection technology is fairly mature. While you can test the solution yourself, it is generally hard to replicate what fraudsters do. Therefore, it is best to rely on independent third-party certification providers like ISO 30107-3 Level 1 and 2. iBeta provides certification, and the results are publicly available.

Identity verification

The final check is to verify and match the face from the ID card and the face seen during liveness detection. This is a face biometric comparison technology that is very mature. I have written a detailed blog post on this. While this is an important check, the technology itself is very mature, so the difference between vendors is not very high.

Information security

Given the nature of the data being handled by vendors, it is also important to verify the information security features of the solution. SOC certifications help in determining this.

Best providers across the globe

Below is a list of some companies that provide onboarding solutions that I have reviewed over the years:

• Onfido - so far the best I have seen in terms of features and ease of implementation.

• Jumio - the pioneer who started the IDV revolution.

• Incode - one of the best for the US market.

• Facetec - by far the best liveness detection that I have seen.

• AU10TIX - comes from the border control space. I have heard great feedback about their document verification capabilities.

• Idemia - also comes from the border control space.

• Shuftipro

• Sumsub

• Advance.ai - (my previous company) - the best in Southeast Asia.

• HyperVerge - started in India but has an impressive presence in a lot of countries.

• Persona

• Mitek

• Innovatrics

• Veridium - provides a fingerprint scanning solution through their 4F feature.

• Amazon Rekognition - provides ID OCR, liveness, and face comparison point solutions. Not very customized to the use-case but are easy to access and super cheap.

In the below video, you can see an end-to-end implementation of an e-KYC journey by Bunq (Neobank based out of Amsterdam) using the Onfido solution. The solutions you deploy would look similar to this.

Just to reiterate, these are my opinions based on interacting with fintechs, banks, and vendors themselves. If you have any questions or want to know more, please feel free to reach out to me on LinkedIn.

Thanks for reading,

Ravi