Creating an automated host warning and banning system

Product design > Turo case studies > Quality : Automated host warning

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 2019 my team set out to improve share of 5-star trips. We set out to create automated consequences if hosts fell below a certain overall rating threshold.

Details

Team: Host

Platforms: IOS/ Android/Web

Role: Senior Product Designer (Design lead on a cross functional team)

Timeline: 3 months

Skils: Iterative design, UX design

Problem

We already had ratings and reviews put in place, and previously built the All-Star host program to incentivize good behavior, but there were no automated consequences for hosts who’s ratings fell below a certain threshold.

Our worst offenders would be flagged and removed manually, but this process could take a while and in the mean time, these hosts received no coaching and continue to contribute to terrible guest experiences.

Insights

Guests who leave 5-star reviews have the highest likelihood of returning to Turo.

Goals and metrics

Improve share of 5-star trips
If we can get hosts to improve how they interact with guests and make the guest trip experience more enjoyable, we will hopefully improve the overall share of 5-star reviews and improve guest repeat rate.

Existing experience

The performance page was previously created to display host quality metrics related to interfacing with the guest.

Response rate - how often they respond to a request
Acceptance rate - how often they accept trip requests
Commitment rate - how often they
5-star ratings - % of trips that receive 5-stars

In 2019, we added the Host status component, which is awarded the all-star host badge if a host surpassed a certain threshold in each metric.

Strategy

Build more statuses into the pre-existing All-Star host status module.
Give a banning warning when the Status component falls below a certain threshold.

Close the loop holes
Even after banning hosts, they quickly figured out how to get back in by signing up with a different email or a different phone number. We wanted to make sure they were banned for good.

How might we warn hosts who consistently deliver sub-par experiences and ultimately remove them if they fail to improve?

Warning State iterations

After all that exploration, our systems team recommended a less stylized approach (for now). I disagreed that it wouldn’t fit the styles of the other states, but ultimately went with their recommendation. 🤷‍♀️

Final design

A lot of logic went into when we would begin warning a host and how long we would wait before we banned them. We could only give them generic tips on how to improve because we did not have any structured data on why guests rated hosts sub 5 stars.

Final outcome

This, along with the All-star host project, helped us improve the five-star trips metric by 5% in Q4 of 2019.

We did remove a lot of hosts, but from Turo’s perspective, they had shown little intent in improving to meet our standards, so the business considered this a necessary win.

What I would do differently

At the time, we did not have any resources to do user testing. Had we had those resources, I would have tested the copy and the designs to ensure comprehension.