Uber Eats is a food delivery platform that helps you get food from your local restaurants to your destination in 3 simple steps - Browse, Order, and Track.
An order could get cancelled even after the driver was already delivering it. The whole journey just stopped there, leaving a starving angry user alone.
We focused on order cancellations initiated by Uber Eats, drivers, or restaurants after pickup.
The target group is Uber Eats users of all levels in North America.
Conducted interviews with users and drivers, as they both played an important role in the journey. Would have interviewed restaurant owners if possible. Besides, I had done competitive research and online surveys to understand behaviours.
Why did my order get cancelled? I need a reason.
Bell / Expert
I don't think I can get anything out of the chatbot.
Ada / Addict
I need to know what I don’t know, not what I already know.
Grace / Expert
After spending so much time making up my mind on what to eat. I rather stick with what I was ordering.
Carman / Occasional user
My hypothesis was users would not reorder after order cancellation. But the research result revealed that 100% of users tend to reorder. Instead of solving how to make users reorder after cancellation. I pivoted the focus to improving the reorder journey in terms of speed and efficiency. Great realization in the early stage.
3 key insights
Lack information of order cancellation
Feel undervalued since no options were given before involuntary cancellation
Need real-person customer support
Before moving to handcrafting the design, I articulated the insights from the research into a feature pyramid which is the value pillar of my design.
Reordering involved 9 screens and 8 decisions asking the user to start over the journey.
I reduce the number of screens by 67% to 3, and several decisions by 75% to 2 in the new reordering path. This helps users to reorder faster and lessen their negative feelings.
A user receives an order cancellation update from Uber Eats, then tries to reorder the same food.
A user receives an order cancellation update from Uber Eats, then tries to reorder different food from a different restaurant.
A user receives an order cancellation update from Uber Eats, then looks for support.
3
4
30 - 55 mins
Google Meet
Low-fidelity prototype
To simulate the real scenario as closely as possible, the test started with a notification from Uber Eats on a locked screen so the participant will go through the real process as it would.
Prototype for usability testing
I don't think I can get anything out of the chatbot.
Ada / Addict
Why did my order get cancelled? I need a reason.
Bell / Expert
After spending so much time making up my mind on what to eat. I rather stick with what I was ordering.
Carman / Occasional user
I need to know what I don’t know, not what I already know.
Grace / Expert
3 key insights
Lack cancellation details
Tend to order something else
Live chat is not helpful
Participants did not seek customer service after the cancellation, while what we have learnt in the research is that 76.5% of respondents loved to contact Uber Eats’s support.
Besides, most of the participants felt that quick access to Ordered Before is meaningless to them in the usability test, but 58.8% of respondents in the online survey indicated the need for this. Surprise!
Consolidating all feedback from all the user research and usability test, I carefully iterated the mid-fidelity design to adapt to users' needs.
Major iterations
Replace quick access of Ordered before with Similar restaurants
Replace live agent support with cancellation details
In the usability test, most users found live chat unhelpful and liked more cancellation details. They also felt neglected due to a lack of information.
Add a Details button
This provides a detailed summary with one click: driver’s name, reason, and photos. Reference number for follow-up.
Although users preferred restaurant suggestions they had ordered before in the online survey, they wanted new information that they don’t already know during the usability test.
Fast delivery and Similar restaurants toggle
A quick suggestion for similar restaurants based on the user's immediate needs. Users often order similar food after cancellation due to previous decision-making. With this toggle, users save tremendous time in searching.
I’ve learnt that 77% of participants indicated that they would contact Uber Eats’ support team after order cancellation. But no one thought of that when they were given the simulated situation during the usability test.
When users can feel the changes but are not able to tell the difference. That’s the best iteration as long as the changes could improve the whole user experience.
Making iterations on a product with many keen users needs careful consideration. It’s important to take small steps in publishing changes helps preserve the user flow they used to have, so this has less impact on the digital experience they expected.