How would you improve Uber?
+9 votes
in Product Improvement by (725 points) | 4.4k views

3 Answers

+3 votes

Let’s follow the steps:
1. Describe the product
2. Clarify the scope of the question
3. Choose a metric
4. List the user groups and select yours
5. List and prioritize the use cases / pain points
6. List out your solutions
7. Evaluate solutions
8. Define metrics for measuring performance (Optional)
9. Summarize your answer

1. Describe the product
Uber nowadays is more of a logistics company lately. The Uber driver take passengers, parcel & food to their destination. They also offer UberFlex for drivers that don’t have cars of their own. They have big investments in autonomous vehicles innovations. They have a presence in many countries.

2. Clarify the scope of the question
What does improve mean?
Is it a concern in specific markets, countries, regions?
Is there a specific metric or problem that needs to be solved?

Talk about key metrics:
> # of customer rides dropping
> # of drivers being stagnant
> # of food deliveries stagnant
> Challenges with autonomous vehicles & how they can be improved
> Let’s look at breakdown by markets. It is possible that some markets are underperforming while others are thriving. Perhaps some markets need special attention due to competitive pressure.

3. Choose a metric
Let’s say the interviewer wants you to focus on:
> # of food deliveries stagnant in Bay Area over past 2 quarters

4. List the user groups and select yours
> Supply: Restaurants signing up.
> Demand: Customers ordering.
> Logistics: Drivers providing pickup/drop-off service

Let’s say for us the problem is of Supply. So more restaurants need to signup on platform.

5. List and prioritize the use cases / pain points
You will start to collect qualitative & Quantitive feedback about why restaurants are not signing up on the platform.

1> Signup & on-boarding process for restaurants is too complicated.
2> Marketing/Outreach investments in Bay Area might have to increase.
3> External Factors: We should also look at competitive pressure from companies such as EAT24 & DoorDash in Bay Area where UberEats might be falling behind. Maybe competitors are offering some incentives for signup that we should consider.
4> They get enough business already from competitors or foot traffic and they do not need any more business.
5> Internal marketing/outreach is weak

6. List out your solutions
1> Simplify the signup workflow for merchants by reducing fields to fill
2> Simplify the on-boarding workflow for merchants by allowing PDF file parser/import of menu items
3> Simplify image import from phone directly
4> Build a dashboard for restaurant owners that gives them step-by-step instructions on what to do next
5> Purchase/Contract with a third party solution/company that already has relationships with restaurants
6> Hire third party contractors to engage with restaurants and bring them on the platform
7> Hire for internal outreach team to reach out to more restaurants
8> Post an ad in newspaper/billboard/radio/TV inviting restaurants to signup with UberEats. Throw in some incentives.

7. Evaluate solutions
Solutions have to evaluated based on: Strategic fit, Impact vs Effort, ROI, Competitive Pressure. Best thing to do is to assign a score to each of the product ideas. These would then allow you to categorize your solutions against a simpler metric:
Must-have vs. Nice to Have

Based on these criteria the following would make it to the top of the list because these are good long term strategic initiatives that we should be investing in. Also, the time invested in these solutions would give us a benefit across the board in multiple regions. Effort would be low, and impact would be high (assuming that is a regular complaint from sellers that drop off during signup).

1> Simplify the signup workflow for merchants by reducing the number of fields to fill.
2> Simplify the on-boarding workflow for merchants by providing PDF parser for import of menu items.

Let’s say cumulative cost for these two ideas = 24 man months, with a 6 months duration.

Now if the problem is not in the product but in the acquisition pipeline they we would have invested in the ad campaigns first.

8. Define metrics for measuring performance (Optional)
Key metrics were defined earlier:
> Number of restaurants signing up each day/month/quarter should increase as per projections
> Number of orders should increase each day/month/quarter as per projections

9. Summarize your answer
The way to improve Uber Eats is :
> Focus: On supply side by bringing more restaurants to the platform
> Region: Bay Area
> By addressing problem: Signup is complicated & takes too much effort
> Solution: Reducing the number of fields to fill.
> Solution: Allowing PDF file parser for import of menu items.
> Duration: For a development period of 6 months
> Cost: At a cost of 4 developers man months. So 6*4 = 24 man months cost
> Expected Return/Goal: With the goal of acquiring another 1,000 restaurants on the platform.

Hi there,
Thanks for the answer. Everything good until you start the solutions section. Solutions such as “simplifying the sign up flow” is not going to impress the interviewer as it’s not specific enough, doesn’t explain what you’re exactly doing (and how you’re doing it) and, is not unique. I strongly suggest that you stay away from any “tweaking” and tactical solutions and focus instead on new unique features that solve very specific user problems.
Got it.

Here are some unique ideas to get more restaurants on the platform:

1) Build a loyalty program on top of UberEats, so that restaurants can offer loyalty type of promotions to their customers. This would keep customers engaged and a dedicated portion of the site/app could showcase promotions from past restaurants & deals from new restaurants in the neighborhood.
e.g. Buy 5, get 1 free.
Exclusive promotion for today: $5 off entire order over $50.

2) Create levels for merchant capabilities: Silver, Gold, Platinum with varying benefits & commission structures. This will allow merchants an opportunity to differentiate themselves on the platform, perhaps as “featured restaurants”. Offer the platinum service for free to new merchants for first 3 months.

3) Hire professional photographers to take pictures of restaurants and their menus. Offer this service for a fee/free to attract new/notable restaurants to the platform.

4) Offer incentives for restaurant owners to invite other restaurant owners to the platform.
0 votes

This is a more built out version of a 15min breakdown I did with my prep partner Denise (who in turn answered the Improve Twitter question).

Uber is ride-sharing application connecting riders interested in short-notice transportation and crowd-sourced drivers interested in monetizing their time and fixed vehicle assets. To improve the Uber service, I’d like to focus on retention of riders as more frequent riders are more likely to refer friends and drivers will drive for the platform that offers the highest chances of rides and by association the larger user base.

Identifying Uber segments:
Commuters Commuters are people with a daily routine who are interested in getting to and from destinations in a predictable, regular pattern. They are interested in reliability and predictability in pickup to hit a flow-state for getting the day or evening started or wind-down from work. Commuters as it relates to predictability, may be more inclined to cut time short, as there is less natural uncertainty to budget and running late 3min, could cost them an hour in commute.

Recreational Users – Short Trip Rec users have less predictability in their ride planning and may be early or late. They may be at work and need to get to a dentist appointment within a close proximity. Recreational users will be more sensitive to the experience of the ride as it is less routine. Ease-of-use will be important in terms of setting up the logistics.

Recreational Users – Long Trip Users without ready-access to vehicles or transportation options for rides of a significant duration >30min. This could include airport rides or weekend getaways. Time is budgeted for the trip, but the idea of ride-sharing for a long duration is uncomfortable and awkward.

Reflecting on our goal of improving engagement through retention, commuters will be the most active Uber user and by association the most likely to refer and invest in the process. A seamless end-to-end ride offers substantial value to these customers and will keep them invested over exploring alterantives. Solutions to achieve this goal:

Automated Pickups and Route Planning Allow commuters who will run on a fixed schedule to set a recurring pick-up (+-5min) and preferred route. The feature would be paid-for premium feature feature a driver outside waiting. Uber can use this data to inform drivers where the demand for the following day is likely to be to allow the market to adjust to fill gaps instead of the current word of mouth, secret strategies each seem to have. The preferred route, could be compared with traffic conditions on a day-to-day to alert the user early to delays and optimize route. Integration to alexa and google home would support. Drivers would be encouraged to not deviate from app established route and would be compensated for the time spent waiting as a percentage of the premium feature addition. The waiting driver would act as a buffer against being late and offer momentum to the user who walks outside vs open app, assess driver proximity, etc.

Driver Ratings & Passenger Preferences Offer better matching of drivers with riders to ensure more pleasant rides. Most people would prefer to wait another 1-3minutes to ensure their driver is compatible and people love sharing bad uber stories, which is negative word of mouth on the company. The current model of passive ratings could be a more active experience to score drivers on features of cleanliness, small, talkativity, thoughtfulness, speed, forethought, etc. These metrics could be compared with rider profile preferences. Riders would be offer an immediate ride or an optimized ride.

Playlist pairing Allow the driver to indicate if they’re willing to accept music streaming of riders at a preset volume to allow commuters to ease into their ride and mental flow-state. The phones would connect through the application and buffer songs ahead to ensure dropoffs don’t interrupt.

Automated pickups with route planning will boost retention by minimizing unexpected events along the journey, which could lead to churn, especially if other commuting alternatives are available. Scheduling is already a feature, as well as optimization. The route planning feature similar to Google Maps pick and drop would need to built out, as well as the driver demand gap forecasting, and premium subscription purchase UI. The predicates for use route planning would lead to expedited development – 2-4 weeks for beta testing, whereas the demand tool, which I suspect exists internally, would have to have a UI built out, which could phase in based on initial success after 2-3 months of dev time and testing. The luxury experience of having a driver waiting for you is a feature reinforced as a luxury and premium in our culture so monetizing this feature is likely if priced in a way where drivers can make more, riders won’t cancel on account of price, and the platform can increase revenues a fraction and benefit from scale.

Driver rating and passenger preferences offers a way of avoiding what can be a horrible experience and putting the power of choice in the users hands, which comes with responsibility for outcome – If I had just waited I could have avoided that. This would increase the seamlessness of the commuters ride and could trickle down to other recreational categories in a scaled rollout as multiple uncomfortable rides will result in churn to another platform and possibly leave platforms all together. Bundling this feature on top of the premium pickup as an initial platform makes a lot of sense given the additional wait can be budgeted for in the advance notice.

Playlist pairing is tricky feature because of the interplay between rider and driver and setting expectations for the ride with variety of taste and preferences. If rolled out in a way where the user is alerted todays ride offers a pairing feature that would be a small win to start the day. Similar to two, I think this makes a lot of sense to rollout in a premium feature as part of a premium subscription for commuters where the driver can monetize more of the service. Control of the environment is one of the few variables that the driver has control of to make their experience on the platform pseudo-enjoyable and this option being forced, could result in driver churn.

Summary In summary, improving Uber through feature development for retaining of commuters is the most likely to grow the platform and introduce new customers through referral. Introducing a premium option for commuters to have a waiting car, preferred routes, predictive traffic indication along route for the rider, for a small increase will let commuters into their routine easier, benefit drivers through demand localization and monetization. Additional features including premium driver matching and playlist syncing would help the commuter enjoy a flow-state and differentiate the experience from other platforms. With rider volumes comes more drivers and positive reinforcement loop.

Hi Steve,
Thank you for the response. The structure is great. I have a couple feedback for you:
– I think you do a great job of explaining each user group and their needs. I would have just described each user group and after highlighting the user group I’m focusing on, I would get into the pain points and needs of that particular user group. This approach gives you more time to go deep on your user group’s needs, and solutions.
– I really like the first solution. It’s great. I think the other two ideas might be an extension of existing features of uber. for example, uber currently allows you to play songs from your spotify, which is similar to your idea 3. you’ll want to make sure that you are introducing new ideas. one way to ensure this would be to ask the interviewer if they do have a feature that’s similar to what you’re describing.
– It would be a good idea if you listed a few more solutions to have more room to play with in your evaluation step. One idea I would particularly suggest would be to offer an Amazon Prime type of service for the commuters as a way to lock them for using uber every day. could be a fixed annual fee to offer preferred rating for frequent users.
Hope it helps. great answer overall.
0 votes


I took few of the steps from the suggested structure, and expanded on those. I am influenced by the Uber version I am using in India.

Here goes my improvement suggestions:

One of the important users of Uber are casual riders, who does not plan the exact time of travel. They are out with friends, or coming back from a movie or going to a restaurant or going out for some shopping. These riders are not too sensitive to the amount of time it takes for the car to come to pick them up, as they are usually in a relaxed environment. They are not in a hurry. However, they want a good, pleasant experience while riding the car, and the ride should not spoil their otherwise happy time.

The need for such a casual rider would be to: a. Spend less effort booking the car. The act of booking the car should be easy b. The car should be of reasonable quality, with a polite driver, so that the upbeat moods stays

For this user, the current pain points are :
a. Having to re-charge wallets/pay for last ride, before booking the next ride (after they launched the app to book the ride)
b. The driver being unprofessional, does not find his way to his/her location and calls her instead, asking for directions
c. Bad quality car or driver arrives

Solutions: Multiple solutions will relieve the pain points described :
a. Recommend pre-filling your wallet with projected monthly spends (projected using analytics based on past usage)
b. Allow a credit limit to the rider based on past payment track record (so that they do not have to pay at that instant, ride now pay later)
c. Location sharing of the rider with the driver, so that the driver need not call for directions
d. Allow rider to define a minimum rating (as a setting) for casual travels. Any driver not meeting this cut-off will not be booked. Since the casual rider is not pressed for time, he would be ok to wait for well rated car to be available
e. Option for casual rider to allow auto-re-attempts for booking , so that even when request for booking times out, it again re-initiates in few minutes by itself

Evaluation of the solutions:
a. Pre-filling of wallets is a good solution, as it will improve the customer experience by not requiring them to reload everytime they need to travel. It also is profitable for Uber/payment partner, as money comes in earlier (enhancing working capital)
b. Credit limit to customers may be a risky solution, if many users do not end up paying later (maybe shift to competitors). Also it will not be great user experience, when the credit limit runs out and user has to suddenly shell out a substantial amount before taking the next ride. I would rule this solution out.
c. Location sharing is already launched in some versions of Uber. This is a good solution, though given that this use case (of driver calling to get directions) is not a frequent/major use case, I would consider this solution as lower in task priority
d. Minimum rating setting for users is a reasonable idea, as long as it does not increase the number of steps it takes to book a car (for example, introducing a step like – book only cars above your chosen minimum rating – which user needs to tick/untick every time, is not a good idea). A suggestion would be make this a default (if the setting is set). We do not expect the rider to change this setting frequently, unless nature of his rides change from casual to formal commute
e. Auto-re-attempt feature should again be a setting, which rider can set as default. Both d and e extends on existing features of Uber, and may not be too difficult to implement

Conclusion: With analytics driven pro-active recharge of wallets, cut-off for rating and auto-re-attempts for booking till finding a car with required rating, the experience of a casual rider will improve. He/she will have that little extra time to say a nice good bye to their friends. The improvement can be tracked as average ratings that the user gives to such rides (higher is better), and number of clicks that user needs to take before booking (lower is better).

Your name to display (optional):
Privacy: Your email address will only be used for sending these notifications.
To avoid this verification in future, please log in or register.
Your name to display (optional):
Privacy: Your email address will only be used for sending these notifications.
To avoid this verification in future, please log in or register.

Related questions

+5 votes
7 answers
asked Apr 17, 2018 in Problem Solving by bijan (725 points) | 3.3k views
+2 votes
2 answers
asked Aug 7, 2018 in Product Improvement by bijan (725 points) | 682 views
+1 vote
1 answer