Uber Rating System

This is a two week project to improve Uber rating system. In this project, we are trying to solve the the discrepancy of customer rating and Uber rating needs.


Group 6

What’s the Problem?

Interview & Secondary Research

What is “Average”?

Scale average is not equal to the fact average. Customers’ common sense telling them 3 is average, while in fact, the average is 4.8.


4: Praise or Criticism

4 is good enough for some customers. For Uber, less than 4.6 average rating will get the driver deactivated from the system. 4 is not a nice score but a poison.


Who is Paying for Cognitive Bias

Sometimes uncontrollable factors such as surge price and bad traffic situation will be counted into the final rating due to Halo Effects.


 How to Solve the Problem?

Secondary Research

1. Close Pandora’s Box

Five Stars to Three Stars

         The cause of the first and second problem is that user do not understand  their rating on the same view as Uber. To make it worse, different users have their own understandings of “good rating”, which makes it impossible for Uber to adapt to the user.

        How to solve this problem? Should we train the user so that they could rate “wisely”? No, that’s impossible. We choose to collapse the rating scale from five stars to three scale so the meaning of each choice would be clean enough for users to make “wise choice”.

2. Correct Cognitive Bias

To avoid driver’s rating be affected by uncontrollable factors, what we do is not trying to educate customers, but to adopt categorized rating. In this way, Uber is able to differentiate which parts are driver’s responsibilities. What’s more, we want driver to know their strength and weakness, so that they could make improvements.


Group 6