My first ever case study, published December 2019. Done as part of the small, inaugural cohort of Students Who Design, which took me from 0 design experience to where I am today. The case below is lifted directly from Medium, where I posted the original study. Additionally, here's a thread that I wrote with some more thoughts

“What do you want to do for dinner?” my roommate asks, firing up Uber Eats.

Accompanied by a cloud of wafting, white-grey kitchen smoke, imaginary waiters pile into the room with arms outstretched to the sky, effortlessly palming silver platters of salty meats coated in viscous, brightly-colored sauces; stalks of vegetables Jenga-stacked atop each other; and golden-brown, glistening shells of encrusted treasures, freshly pulled out of their dips in vats of bubbling oil. Chinese, Thai, Italian, Mexican, American — the choices were endless. But I knew that the actual options were Boba Ninja, Toss Noodle Bar, Panda Express or Chipotle. After all, we ordered take out from there last week, the week before that, and the week before that.

This is in part because of delivery apps. Uber Eats today is a great way for low-budget college students to order food from restaurants. It also suggests new spots to try, but these other options are often ignored. People want to find new foods that they will enjoy, but they can’t because:

They won’t order outside of their pre-chosen handful of restaurants.

They don’t want to spend money on something they won’t like.

This is doubly inefficient: Uber Eats keeps restaurants that don’t get views and restaurants keep pretty profiles that don’t get orders.

I first thought users didn’t order from Uber Eats’ recommendations because they were bad. But through behavioral interviews, I learned that this wasn’t the case.

I found out that users actually completely ignored recommendation features. They would go for what was cheap or familiar, ignoring recommendations along the way.

I also found that users who said they ordered from many places ordered from 4–5. They created a list of restaurants that were cheap and passed the taste test and only ordered from them. This ensured that they weren’t paying for anything that they wouldn’t enjoy. I learned more and more that users didn’t trust Uber Eats’ recommendations enough. Food was much larger of a commitment than a skippable song recommended by an algorithm.

To explore this, I scheduled a late-night brainstorming with a group of friends.

Most of the results centered around making food discovery more interesting to users. Two takeaways were to make copy more friendly/colloquial and suggest food in newer ways.

My early concepts toyed with this idea of innovating how Uber Eats recommends food. Some examples were using Friends’ past orders and Pairings, foods users enjoyed together. I narrowed it down to these final screens’ features:

I split the screens into two groups:

Recommending: In these screens, I explored the entry points for recommendations. Shown above, there are examples of “pre-order,” “mid-order,” and “post-order” recommendations. I tried to use the social aspect of ordering and food combos in mind to help refer other relevant foods.

Exploring: This screen was more of being able to explore a wide variety of food. I thought that showing every choice through pictures would help users make decisions.

Recommendations are a feature bread-and-butter to other apps. Amazon and Netflix, for example, rely on being able to recommend users items to buy/binge. 35% of what consumers purchase on Amazon and 75 percent of what they watch on Netflix come from product recommendations based on such algorithms. Uber Eats can learn a lesson from this. Recommending appetizing pictures of side dishes or drinks could increase average cart size.

I played around with the display of food recommendations. After considerations, I went with the inlet (highlighted). It wasn’t as intrusive as the other options and created a hierarchy between information.

Surfacing drinks, side dishes or desserts also addresses indecisiveness from scrolling through menus. If this builds Netflix/Amazon-like algorithm strength, users will also trust it more. Two things about this feature that I’d like to highlight are:

The inlet: Uber Eats’ current app has everything on an “overlay.” Seen in the home screen, white tabs all lay on top of a grey background. On top of those white tabs lie smaller modals of restaurants/promotions. This is to establish an intuitive hierarchy and guide users to important information. So, to display “secondary” recommended information, the only way for me to go was deeper into the screen. I did this by splitting the menu to an inlet with the recommended add-ons.

Use of Uber Eats’ green: Rather than using regular font, I accented the price increase with Uber Eats’ green. This is a great visual hint to give users a clear idea of what feedback they will get if they interact with it. This push for interaction design and feedback is something that Uber Eats needs to do more of.

The second feature, the Explore Feed, was a way to get users to discover food.

I chose the screen without text cues (highlighted) since food is innately very visual. I thought using more space for images and relying on visual cues would attract users’ eyes more.

The screen is a non-intrusive way for users to explore and “eat through their feed.” It also allows users to explore all their choices. Two things that I’d like to highlight are:

A food-focused approach: Uber Eats currently separates by restaurants instead of by food. Although this organizes the home screen, food is a visual experience first. The Explore Feed shows all the food on menus to entice users more and encourage “eating with eyes first.”

A “New!” banner: I learned that users used “Past Orders” on the navbar to help decision making. I use this banner to help users realize which food they’ve tried at a glance and also invite them to clear all the “New!” banners, reinforcing the behavior to “eat through their feed.”

A common theme in my interviews was that users thought of Uber Eats as too much of a “functional” app. Its core functionality is too powerful for them to pay attention to other features. Companies like this that don’t leverage community, locality, etc. are often disrupted. I hope the Explore Feed and Recommendations start a movement in Uber Eats to do this!

While working on my features, I learned that Uber Eats ended up shipping a version of them. Uber Eats’ newest update has food-first modals and a recommendation pop-up. This was a delightful surprise! I hope Uber Eats rolling these out shows that my research was spot on, meaningful, and tackled a real user issue.