Sayali Deshmukh

BACKGROUND

Being a Lyft user, I used to chat with the drivers and fellow passengers about the service, rating system etc. I found that rating system was one of the pain points which was frequently brought up.

Customer is unsatisfied with the driver.                                                Customer gives poor feedback for the driver                          Next day, the driver sees a drop in his rating and is Driver Driver is unaware of any customer dissatisfaction                                                                                                                                            dumbfounded                              

USER RESEARCH

I conducted some interviews with a number of Lyft drivers asking for their problems with the current rating system. I also interviewed some people using Lyft for transport.

Besides talking with the users, I also researched on sites like Twitter, Reddit, Glassdoor to understand what Lyft drivers were saying about their current rating system.

PERSONAs

LYFT DRIVER

LYFT USER

User Journey

For the driver - 

UNDERSTANDING THE PROBLEM

For the driver -

• In most of the cases, driver does not know what was the reason for a low rating

• Generalized rating system based on number of stars

• If rating falls below 4.6, the driver is fired

For the user -

• Time constraints/low interest in writing down comments

• Rating/comments for the previous ride, asked while calling Lyft for the next ride

SUCCESS METRICS

• Lower time taken for giving constructive feedback

          With the previous design, if a customer would want to give any constructive for the driver, they would have to type in their feedback in the comments section. With the addition of                these rating factors, passenger's would have to use the slider and give their rating.

• Increase in passengers who give feedback based on the factors

          With the previous solution, passengers entered the comments, only when they felt strongly satisfied or dissatisfied with the driver. With the additio of these rating factors, more                    people should opt in to give feedback   

• More drivers getting higher rating, over time

         This metric would have to be calculated over time. The assumption is, If this rating system works, the drivers would get constructive feedback on what to improve, and thus they                 could work on these factors. 

SOLUTION

If the customer is happy with the driver, he can choose "Yes" in response to "Would you recommend driver A to others". After that there will be a prompt for Tip which would be optional. Also, there will be a feedback request where the user can select what he did like about the driver. Else he can choose to ignore this.

 If the customer is unhappy with the driver, he can choose "No" in response to "Would you recommend driver A to others". After that there will be multiple options where he can choose what and how much he is dissatisfied with the driver.

FACTORS TO RATE THE DRIVERS ON

Here is the chart which shows the most frequent complaints that passengers have. I chose these factors as an inspiration for providing feedback to the drivers. 

SKETCHES 

I sketched multiple designs for different screens. I then asked some of the Lyft drivers and customers to complete 2 tasks

• Positively recommend driver A
• Give negative feedback for driver A

I then asked the users questions on 

• What they liked
• What they didn't like
• Suggestions for improvement

Based on their feedback, I iterated on the designs. 

DESIGNs

First the user would be asked whether he would positively/negatively recommend the driver. Based on his input, he has to rate the driver based on four major factors 

• Driving 
• Car cleanliness
• Navigation and route
• Driver friendliness

User can then scale these factors based on how much they liked that particular factor. User can also provide additional comments if he wants to. 

RESULTs

Since this was a side project, I tested the hi-fidelity prototypes on people who used Lyft.

I evaluated the result based on their performance for tasks -

• They were happy with their ride
• They were dissatisfied with their ride, since the route chosen was not optimal
• They were dissatisfied with their ride, since the driver was unprofessional

For these task flows, I did a comparison of this designs with the original version. The average time taken for these tasks was reduced by 30%.

Also, for the negative use cases, some users gave a lower star rating without leaving any comment about the route or the driver for the original design. For the designs that came out of this exercise, most of the users chose to rate the ride based on the factor that they were unhappy with, thus providing some constructive feedback for the driver.