Helping Turo uplevel their guest verification flow
Product design > Turo case studies > Guest verification
Synopsis
Turo is the largest peer to peer car sharing company in the US based out of San Francisco. It serves as a marketplace for people who own cars (hosts) to share their cars with people who need a car (guests).
In spring of 2020, the pandemic had ravaged Turo’s bottom line and we needed to shift the entire company’s focus on improving our profitability. My team took on a very ambitious task of attempting to reduce our insurance premiums through a better guest verification process.
Platforms: IOS/ Android/Web
Role: Design lead (sole product designer)
Core Team: 1 Lead product designer (me), 1 Product manger, 1 Engineering lead, 5-8 engineers.
Timeline: 6 months
Skils: UX/UI, Journey mapping, Design strategy, content strategy, prototyping, user research
Company problem
Turo needed to find a way to reduce our insurance premiums, a very expensive line item in our business. The insurance company partner we worked with told us exactly how to do this:
Get better at ensuring the right person gets in the car.
Should be simple enough, right? 😅
Our existing verification vendor was a less than ideal experience and cost Turo
We had currently only used a verification flow in certain circumstances, such as when a guest books a Turo Go trip, or a very expensive car, or if they tripped up our risk checks. But the experience had a number of issues:
Our existing vendor plugged into Turo and offered very little customization, which made it feel clunky and off brand.
Up to 34% of the Licenses that are scanned result in unreadable errors.
The current vendor only used agents to verify the licenses, which lead to a longer wait time for guests up to 3 minutes — A terrible conversion killer for our main booking flow.
The vendor’s services were very costly and not competitive with other services that appeared to be better.
Part one
Integrating a new vendor and developing the flow
Once we had made the assessment that our current vendor was not suitable to be used on our entire population of new guests, we researched and landed a new vendor which was the leader in it’s field. The biggest draw was the automated software approach to verifying the users and the promise of much higher success rates, especially with people of color.
More customization
Another of the benefit of integrating the new vendor, meant we had much more control on the experience and the customization. I spent months working with their team to understand what they recommended as the most optimal flow for conversion.
Guidance for quality images
The biggest message we took away from the vendor was that most failures occur because the images captured are low quality, either due to low light, glare, blurry images, low contrast, or articles of clothing obscuring the selfie.
We had to design guided steps to help users avoid mistakes, which would result in re-taking photos.
We also had to design other flows to allow the user to take images manually, in the event the camera was not able to detect the license.
Platform specific quirks
Each platform SDK had quirks and slight differences, which made it challenging to design the same experience. I often had to make sacrifices in order to cater to the unique challenges each engineer shared with me. Some examples were:
IOS SDK had different animated guides on the camera than android
The Web License scanner had challenges and the recommendation was to use a QR code to get the guest to finish the flow on their mobile device. I had to design an entire flow around that.
Web had an entire set of error states that had to be accounted for that were unique to that platform.
Part two
Defining the flow location
A large area of contention across stakeholders was where to place this sub flow within our booking flow. The booking flow itself was already very long and there was fear that adding this sub flow would add so much friction that it would hurt conversion too much.
Collaborating with other teams
At the time of this project, there were 3 other teams working on different aspects of the same booking flow:
Adding new requirements in guest onboarding to add credit card before seeing final price
Adding covid 19 training badges to host information on the vehicle page
Experimentation with protection package selection.
I foresaw this as a potential risk, and formed a working group with the other designers to ensure we were all aware of the changes and stayed in sync.
We conducted usability testing of the new booking flow to understand potential impacts on conversion
Insights
Too many repetitive steps
Most users felt that the individual steps were easy but overall process was long due to repetitive or “out of order” steps.
“It wasn’t hard to do but it was pretty complicated and took a long time.”
- P2
“We did driver’s license already and this added passport. Why are they asking for DL again? That's confusing.”
-P4
“Taking the pictures [was repetitive]. Like taking pictures of the ID multiple times, choosing a profile picture, and taking a picture again. I don’t know if that’s something to set up a profile and then you rent a car or you have to do that each time.”
- P3
Sometimes the final decision is out of your hands
Despite bringing forth research and advocating for a better location that had fewer repetitive steps in the booking flow, at the end of the day, the cost was the biggest reason for placing the Automated ID Verification steps at the end of the booking flow.
At the time, we still had quite a bit of churn once users arrived at checkout. This was due to the lack of pricing transparency prior in the flow.
The business did not want to spend money on license scans that were less likely to lead to an actual booking.
Final outcome
We launched the new vendor and new flow as an A/B test and had a 10% holdout group with the previous vendor for comparison. After several false starts with the vendor SDK, we learned that the new flow was flagging and rejecting more people than we expected. Further analysis was being done when I rolled off the project.
In the event that we would continue to invest in this flow, I put forth a number of recommendations on how to further optimize screens, but none of the recommendations have been acted on thus far.
Regarding my contributions
Every designer at Turo works on a cross-functional team dedicated to core metrics. I was the design lead and sole designer, with a product manager and Engineering team as my counter parts. We pulled in additional support from brand designers, copy writers, marketing, legal, data science, research, Verificatons, ops, etc. to help shape the direction of the feature.