FanDuel: KYC Improvements

Background

FanDuel acquired 3 million new users in 2022. 

But it could have acquired more. 

FanDuel operates in the heavily regulating gambling industry, and it is legally required to verify users’ identities before they can use its products.

The problem: Only 60% of new users who land on the verification screen successfully complete identity verification.

Of all new users who land on that screen:

  • 18% don’t even attempt to verify their identities and leave the app

  • 22% fail verification (they either don’t retry after failing once or fail all 3 attempts and don’t contact customer support)

In 2022, FanDuel decided to optimize its identity verification flow, also referred to as “know your customer.)

The goals were twofold:

  • Increase the number of users who begin the verification process

  • Improve the number of users who successfully verify their identities

My role

Content design

  • Since I came in mid-project, I partnered with the product designer to ensure that any existing UI copy was consistent with our writing guidelines. 

  • I pushed for educating users at certain points in the flow about why FanDuel needed certain information. The process requires a lot of personal information, and my hypothesis was that more educational copy would put users at ease with providing this information rather than abandoning the process..

  • I worked with product management to ensure that error messaging and warning text was clear and helpful so that users knew exactly the information and format needed.

Approach

The team decided to replace the single-page form users had to fill out with a “stepped” flow in which users provided a few pieces of information at a time over several screens. The hypothesis was that breaking the flow into smaller steps would prevent users from feeling overwhelmed when filling out their information. 

An added benefit of this stepped approach was being able to better track where exactly users fell off during the verification process — a decision that would help further iterate on the verification flow after it launched. 

Since the verification process was changing from one primary screen to multiple screens, the team decided to test one element at a time, with the assistance of the Analytics team (which leads optimization experiments at FanDuel). If the test was successful, the design would carry over to the next test, in which a new hypothesis was tested. That way, the new flow would see iterative improvement with each new experiment.

Even though we were changing the flow of identity verification, the process is fairly rigid. We couldn’t change what information we needed to collect due to federal regulations. This constrained our designs somewhat, however, it also focused my work on how well are we explaining the process to users.

Results

Learnings

Sometimes more explanation was helpful; sometimes it was not.

  • Telling users why we needed their correct birthdate helped keep people from randomly entering numbers

  • Providing a more detailed explanation as to why we needed a user’s full Social Security number after they failed verification actually dissuaded some users from attempting to re-verify.

  • The takeaway is to be mindful of when users are reading for comprehension and when they are scanning the text, and adjust your content accordingly.

Test and then test again

  • Testing one element at a time helped us isolate what was working and not working with the new design. Introduce too many variables, and it’s hard to know what affected the outcome.

  • Being able to A/B test the old and new verification flows in real-life scenarios was invaluable for this project. We could have tested design prototypes with users, but their behavior would have been different than it was when they were entering their actual personal information.

Before and after screenshots of the verification failure modal

in-progress designs for a warning message before a user tries to input an invalid date of birth.

The old identity verification screen for FanDuel Sportsbook (Sorry for the poor resolution, but it’s the only way I could get the whole form in one screenshot.)

The new “stepped” identity verification flow. Since I don’t live in a state where Sportsbook is legal, I had to show the Figma designs rather than live screenshots.

(left) The old modal if a person failed identity verification the first time. (right) My new modal, which tested poorly among users and needed to be rewritten.

Previous
Previous

LinkedIn