Trip Ventures

Restaurant Discovery App

An app to discover trending restaurants around the world

An app to discover trending restaurants around the world

TL;DR

Built a community of over 10K daily active users by introducing a mobile app that recommends restaurants based on social activity.

B2C
Mobile App
Shipped

Role

  • Product design
  • UX research

Background

The Trip Ventures team tasked me to design a mobile experience for an app that enabled customers to discover trending restaurants around the world. I was brought on as a UX researcher and designer to design the experience.

Hypothesis

Social activity is a stronger affinity indicator than customer reviews.

Historically, the main paradigm for finding restaurants has been based through "review-centric" apps, like Yelp, Facebook or Google.

The Trip Ventures team believed there’s an opportunity to disrupt the category by providing a discovery experience that recommends restaurants based on their social activity (aka number of posts from location) rather than their amount of customer reviews.

Research

I interviewed people between the ages of 18-45 in the area of Miami to better understand the current solution landscape and their needs.

Key findings

  1. Overall sentiment is that current apps are “good enough” to find options on the go but are “poor” when it comes to finding “new trendy spots and hidden gems”.
  2. Participants felt recommendations were “too generic” and yearn for more personalized suggestions that felt more like “friend recommendation than a search result”.
  3. Around 85% of interviewees indicated they “don’t read reviews” or “don’t put a weight on them” when deciding where to go. All participants indicated they rely on social media and friend recommendations to discover new restaurants.
  4. When presented with restaurant recommendations based on social activity and customer reviews, 90% (27/30) of participants indicated those based on social activity were “more relevant to them”.
  5. Younger participants, mostly students and professionals between 18-30 years old, indicated the restaurant discovery is “very important” pain point compared to their older counterparts who expressed it was “somewhat important”.

Opportunity

The current market for restaurant discovery apps is failing to meet the needs of young customers seeking trendy culinary experiences. 

By providing an app that offers personalized, social-driven recommendations, we have the opportunity to carve an audience of young “foodies” we can monetize.

How might we empower young "foodies" to effortlessly discover new trendy restaurants and hidden spots?

Exploration

We alined on a strategy of developing an MVP for the Miami market to validate the opportunity and stress-test UX concepts before a wider release. Based on the research findings, we aligned on the following solution principles:

📸   Photo-centric

App should show, not tell.

The app experience should showcase images of the restaurant to enable users quickly discern the restaurant ambiance and quality.

✨   Personalized

App should mirror the experience of a “friends recommendation”.

A friend  wouldn't overwhelm you 30 options. Instead they'll  recommend a cozy off-the-beaten path sport or ask what you in the mood for to recommend the best option possible.

👫   Socially-driven

Experience of finding and booking restaurants should be inherently social.

App should empower groups of friends to decide where to eat and prioritize activity from friends and influences.

For the MVP, I explored a few UX concepts to address the main job: find a new restaurant to visit later.

Recommendations UX Concepts

Before starting with development, I created clickable prototypes and conducted user tests with 12 participants to understand which UX concept resonated most and identify potential friction points.

Hover to PLAY ⏵
Tap to PLAY ⏵

Grid

Feed

Carousel

Ultimately, the “Carousel” concept resonated the most with users. 100% of participants completed the primary task and 10/12 choose the concept as their preferred option. InMost importantly,

I observed this UI pattern encouraged users to consider the  recommended restaurants more carefully and had the added benefit of more real estate to leverage for restaurant information and photos.

From there, I crafted a lo-fi wireframe to map key workflows for the MVP:

Low-fi Wireframes

Launch & iteration

With some directional validation, we moved forward with the launch of a closed beta. We targeted food enthusiasts and restaurant owners in Miami to keep our feedback loop tight and start building rapport with the community. While the overall response was very favorable, several key areas for improvement were identified:

🎛️   Control

Users want more control over their recommendation to be able to find options that fit a particular ambiance or cuisine.

How might we empower users to find the restaurant that fits their need?

🤝   Trustworthiness

Users aren't sure they can trust the recommendations because they don't know who or what's behind them.

How might we establish transparency and credibility in our recommendations?

🖼️   Photo Quality

Users are reluctant to select restaurants with limited or low-quality photos, because they're not able to assess ambiance and food quality.

How can make sure users get the  info needed to choose their ideal restaurant?

I designed and iterated on few elements of the experience to address these pain points before expanding to other major markets:

In-app guides

Despite positive feedback, some users struggling to discover the swipe gesture to browse recommendations. To address this, I introduced in-app guides to explain explained key features or gestures when users open tabs for the first time.

"Mood"  filters

Users indicated filtering by cuisine was a must-have. After further investigation, I found their struggle was the app recommendations did not match their particular needs. For instance, they would never go to the same spot for a romantic dinner versus eating with their family.

To streamline search, I introduced controls to filter by cuisine and ambiance to help users to find suitable options.

Distance filters

Some users requested a map-based interface, but upon investigation, I discovered their pain point was the inability to control their search range. Depending on the restaurant quality, people are willing to travel further. To address this, I introduce distance-based filters to enable users to adjust their search on their willingness to walk or drive.

Photo-only sharing

At our core, our app is the “anti-yelp”. Overtime, I noticed the post sentiment started becoming more negative. Ultimately, the ability to attach text to a post encouraged the behavior we were trying to replace.

I re-designed the publishing workflow to only allow the sharing of photos to encourage positive posts. In the end, the number of photos shared remain steady and sentiment improved with the introduction of the new experience.

Social proof

I introduced social proof indicators to help users understand why restaurants are being recommended and generate more trust. This includes displaying the number of visitors and affinity labels such as trending, new, local favorite, hidden gems and Michelin-rated to highlight exceptional restaurants.

Solution

Key workflows

Onboarding

Sign up experience to collect minimum details and permissions to personalize experience.

This included favorite cuisines, camera roll and location permissions. With these permissions, the app analyzes the camera roll, matching photos to restaurants to create a “taste” profile based on visited restaurants.

As the user continues visiting restaurants, our recommendation algorithm improves.

Home

Carousel experience to enable users to quickly swipe between restaurant recommendations.

The UI design focused on delivering instant value to the user by providing personalized recommendations limited to the user’s current location.

The experience displays only one recommendation at a time and limits recommendations to only 10 restaurants.

This approach favors quality vs quantity and encourages users to carefully consider each restaurant recommendation.

Guided Search

Conversational interface to control their restaurant recommendations.

Once users have swiped across their default recommendations, they’re shown a guided workflow to enable them to filter restaurants based on cuisine, distance and timing.

Similarly to how a friend would ask “what you’re in the mood for”, the conversational experience asks a set of basic questions to streamline the filtering process if none of the recommendations meet their needs.

Feed

Social feed experience to enable users to view photos of the dining experiences of their friends and the broader community.

The core of the experience was the feed and profiles, which allowed users to discover experiences beyond their immediate location and follow other influential “foodies” in the community.

The goal of the experience was to open the app for “casual” browsing (aka reactivation) and encourage sharing of their own experiences (aka engagement).

Photo Sharing

Publishing experience to enable users to share their dining experiences.

The experienced leverages our photo-matching engine to enable users to browse photos more efficiently by grouping them by restaurants.

To encourage positive vibes, I streamlined the publishing experience to only allow sharing of photos and tagging posts with keywords.

Push Notifications

Notification experience to proactively encourage users to share photos from their dining experiences.

The experience leveraged our photo-matching engine to notify users whenever they’ve taken a photo of a new restaurant and they’ve not shared yet.

This allowed us to increase the number of posts in our graph and improve retention by resurrecting users who’ve not shared in a while.

Outcomes

The MVP was designed, built and launched in around 3 months. As a result, we:

  1. Attracted 2K daily active users with close beta in the Miami market within a month of release
  2. Increased user base to 10K daily active users upon expansion to add'l major markets

Appendix

High-def mockups