Summary: Uber has created a massive peer-to-peer ride sharing service that is offering service to millions of riders a day. This system relies heavily on trust which comes through its 5-star rating system. But, this rating system is biased and inefficient in capturing (and conveying) the qualitative problems faced by riders. We followed an iterative user centered design process to design a solution that provides more qualitative feedback (than system of stars) to drivers without increasing interactions for the riders.
Whats wrong with "5 star" rating?
Uber’s rating system is really essential as it helps them to root out bad drivers and riders. But in the existing rating system is subjective leading to inaccurate ratings. User’s interview, study of online articles, Uber reports regarding rating system and drivers’ blogs were used to understand the problems with the existing rating system. The problem of inconsistent mental model regarding rating was evident both in primary and secondary research.
"It’s soul crushing when my rating takes a dive. I spend my first waking moments wracking my brain trying to figure out what could have gone wrong with my rides the night before." - Uber's driver blog
Providing specific feedback to drivers while minimizing interactions for the riders.
The main problem that we identified with the existing rating system in Uber is that its ambiguous and inaccurate. The system's ambiguity comes from lack of shared mental model of working of rating system among the passengers. One rider gives 3 stars on average, other 4 and whereas other gives 5, for the same experience. This further leads to inaccurracy. Moreover, drivers don’t get reasons behind the bad ratings. Providing them reason behind bad rating gives them a fair chance to improve themselves which will prevent uber from deactivating good drivers.
Quick qualitative feedback
The design we came up with asked riders if they faced any of the 6 most-frequently occurring problems. The system then reported to driver if he was lagging in any one of the elements. Example, driver was informed if more than 5% of his customers faced problems with his behavior.
We conducted a semi-structured usability testing followed by design critique with other designers to identify problems with the design. A catch that emerged in design critique and user testing was use of negative options. The design encouraged leaving negative feedback. So, to make design practical we worked on removing this bias. These prototypes went through continous loop of design critque and quick-and-dirty usability testing.
Iteration: Round 1
Iteration: Round 2
Final Design Sketches
We took the 6 most common problems faced by uber riders and made those options available upfront for users to select. In 5 star based rating system users had to convert their qualitative opinions into quantitative one. Through our design they can directly state the problem they faced. Additionally, driver would exactly know what problems riders are facing soo that they can improve upon them. The system all this without increasing any effort on the riders end.
In case of a problem
Reporting the driver