You're the PM of Uber. The ride rating is 20% less than the industry standards. How do you solve this?

Question

You're the PM of Uber. The ride rating is 20% less than the industry standards. How do you solve this?

Asked at Walmart over a year ago

1.4k views

Problem Solving category

Asked at

1.4k views

How to answer Problem Solving questions

Interview Guide

Top Problem Solving interview questions

Invite members
Invite by email

Add another

Tushar · Answer 1 · 2020-12-28T16:27:42+0000

Mission: Ubers mission is to connect riders and drivers and provide a convenient and safe mobility solution

Clarify:

Is this specific to a particular geography? IVR: USA
What segment does this metric relate to - Uber pool, UberX, Uber Luxury, Uber accessible? IVR: Lets focus on UberX
I assume that the ride rating is different from app rating here and the rider is asked for the ride rating everytime they complete a ride? IVR: Yes
I assume the industry standard rating is the source of truth here scouted out from competitors and research companies? IVR: Yes
Has there been drop in ride rating suddenly or it has been there for a long period of time? IVR: Long period of time
Is the pattern same across web/app (iOS/android)? IVR: Yes

Thanks for clarifying. The objective to find the potential reasons for the low ride rating for UberX in USA and provide potential solutions to bring it at least at par with industry standards. Does that sound right? IVR: Yes

I would discuss the journey, factors that may be causing this affect and then discussion potential solutions and metrics to measure the success.

User Journey:

Rider opens the app
Punches the location where they need to go
Selects the UberX option with estimated pricing and wait time
confirms the ride
Waits for pick up
completes the ride and gets off
Rates the ride experience that would be same as rating the driver
1. Optional feedback that user can punch in

I would focus on 5, 6 and 7 for ride rating.

Strategy/Factors:

Verify the way we measure the metric and how it propagates to the dashboard - this is to ensure that we are looking at the right numbers for ride ratings
Analyze the data per demographics (age, sex, income level, loyalty tier), vehicle type, year, driver experience (trip tiers-uber/total driving)
Analyze the ride wait time, ride duration/miles for low ratings and check if there is co-relation
If uber is providing promotions credits for a bad experience, look at patterns if some customers are trying to game the system by putting in bad ratings to get credit
UI for accepting rating in-app is butchered or not displayed correctly - less likely since Uber has been around for a while
External: Regular protests from taxi union, women and child safety organizations that had negatively impacted rider psychology towards uber. We should also see an affect in other metrics of total rides, churn rates, uber app rating
External: Other competitors have reduced their costs and uber riders feel that the ride is not worth the extra money they spent. We will also see a reducing in no. of rides/churn rate if this is the case

Solutions:

Mitigating wait time -
1. Improvements in algorithm to connect with nearly drivers with higher accuracy I-H, C-H
2. increase Uber maps accuracy to enable easy driver-rider meetup I-H, C-H
Train/terminate the drivers that have consistently lower ratings. Look at more granular data (if available) such as rash driving, not following instructions, very talkative, bad attitude/rude I-H, C-M
Terminate users that are consistently gaming the ride credit by placing poor ratings I-H, C-L
Set up third party vehicle inspection for cars of certain brand, age, lower ratings for safety, comfort, cleanliness I-M, C-L
Give an option to riders to opt for a longer/more expensive route that has better roads/smooth ride to the destination I-M, C-H
Give an option to riders to choose the driver type, vehicle type. This would increase the wait time and rider budget I-M, C-L
Provide promotional credits to users who provide feedback with more granular details contributing to their ride experience I-M, C-L
Conduct A/B test for UI changes on the ride rating page if butched on some app versions/browsers/mobile OS. I-L, C-M

I would first look at 2, 3, 1 & 4 as my primary solutions and then proceed to other solutions.

Metrics:

Retention:
1. %/# frequency of rides W/M
2. %/absolute change in average wait time
3. Churn rate excluding the users who were terminated
4. Ride effective score thru lens of
  1. Rider: age, location, sex, income level,
  2. Ride: wait time, trip duration, vehicle type, brand, vehicle age
  3. Driver: age, sex, driving experience, no. of trips, average rating, uber experience

I started with the goal of finding the cause for low ride rating on uber compared to the industry. I looked at some internal and external factors and focused on training and terminating some drivers, conducting vehicle inspections and reducing wait times. I focused on ride effective score via various pillars to measure the success of my solutions.

Thank you. Please provide feedback. (Total time=40 mins)

I was having a hard time in finding the factors that are causing the low ride rating.

Kaushik · Answer 2 · 2021-02-27T15:10:11+0000

Clarifying Questions

When we say ride rating, it is the rating given by the rider for a ride taken right? - Yes
Is this specific to a certain geography or location, has Uber existed in this market before or are we just getting started? - India, have existed in India for a while now
When we say that the ride rating is 20% lower than the industry standards, which time period has this been measured for? - last few months - not a sudden dip
Do we know the reason for the lower ratings - from input provided by the rider when the final number of stars is being given? - car cleanliness, driver politeness, the experience of using the app - no inputs

Ecosystem level changes

Check overall ecosystem level metrics that could be causing bad ratings

Campaign against Uber
A campaign against Cab rides
A campaign being run by competitors to malign Uber
Marketing campaigns to provide low costs than are offered by Uber
Political issues resulting in a bad name for Uber
A new range of cars launched by a competitor

Product-Related Changes

Have we made any changes to the way in which the riders provide ratings? Initially, all the five stars would be selected, if that has been changed to selecting no stars.

Number of ratings provided might have reduced
Only those receiving a lousy service might be providing a bad rating
Ratings are being asked via a notification much longer after the ride is completed, irritating the user and the user ends up giving a bad rating

User Experience Related Changes

If not for ecosystem-level changes, there are various factors that could have affected Uber's overall rider experience, and discuss a few remedies for each one of them.

Booking a ride
1. Opens the app and searches the location - takes time to find the right location
2. Checks the list of the riding options available
  1. the waiting time could be very high
  2. the prices shown could be very high
  3. or both
3. Books the ride
  1. The ride gets automatically canceled when unable to find the ride (happens multiple times)
  2. There could be a strike that is going reducing the number of cars on the road
Waiting for the ride
1. The waiting time for the ride is much higher than normal - there is a differential between time predicted and time of actual arrival
2. Too many contact rides - where the rider has to call the driver to either provide instructions or check the status
3. App crashes while waiting for the ride - the rider has to refresh the app multiple times before the actual location becomes apparent
4. Too many multi-contact rides - the rider had to contact the driver many times while waiting for the ride
During the ride
1. The car isn't clean
2. The facilities in the car wouldn't work properly
3. Had a tiff with the driver on the way
4. Took longer than imagined to reach the destination
5. Took a longer route than normal to reach the destination
After the ride
1. Got dropped off afar from the actual destination
2. Didn't have cash to pay properly

Problem Validation

Identify the correlation between the metrics signifying one of the above issues with the low-rated rides.

Depending on the problem identified, if in the In-app experience, I'd focus on various solutions for this problem.

Solutions

Improve the search experience
Recruit more drivers to Uber to reduce wait times and prices
Improve training modules for drivers to help with using the product and reducing multi-contact rides
Car Cleaning checks for the drivers to help them keep the car clean
Review of the prediction mechanisms for arrival times
Improve precision of pick up and drop locations

Several of these solutions should help in gradually increasing the end-user rating for uber. Since we don't know why the current ratings are down, the correlation is what we depend upon for implementing the solutions, in order of the size of impact and confidence of improvement.