Yelp Concept: Personalized In-Restaurant Experience

Some users use Yelp to find new restaurants. Others look up new dishes when they are already at a restaurant.

Yelp wants to solve the problem of “what to eat” : letting users explore new dishes at restaurants. But, users hardly engage with Yelp content in restaurants because:

  1. Information is not organized clearly and effectively
  2. Does not offer customized experience for different user preferences

I want to create a personalized in-restaurant experience that is:

  1. Fast and effective
  2. Personalized

Yelp Restaurant Profiles Are Not Effective

The current Yelp restaurant profiles overload users with information. When seated at restaurants, users don’t want to be reading lengthy reviews. Users want to decrease the amount time spent on decision making before ordering.

Understanding Why People Don’t Engage With Restaurant Profiles

User Research

My goal was to learn about how users engage with content on Yelp. Here are some key insights from user research:

  1. Users want a personalized experience on Yelp.
“I want to find new restaurants and dishes that fit my taste preferences.”

2. Users want to quickly make informed decisions about their meal choices.

“I want to know what there is and what is good, fast.”

3. Users want recommendations from people that matter to them.

“I trust my friends because I know my tastes will be like theirs.”

Identifying Personas

Personas generated from interviews and surveys

Generally, there are three types of users:

  1. User who needs friends’ recommendation and validation
  2. User who needs dishes that are specific to taste preferences
  3. User who seeks to try out new dishes

Visualizing Journey Map

Storyboarding the Restaurant Experience

People Want to Engage With Useful Content

At first, I thought restaurant profiles were hardly used because they were not necessary. But instead, people don’t engage because information present is lengthy and ineffective.

People desire filtered information that prioritizes highlights.

As a result, my hypothesis only accounted for part of the problem. Content engagement is different when the user is already at a restaurant.

Figuring Out Which Feature to Implement

Brainstorming Session

I recruited my two friends, Tony Li and Greg Schultz, as my brainstorming buddies. After exploring, we decided on three solution spaces.

I sketched out these ideas to visualize the concepts and conducted user interviews to find out which idea is worth pursuing the most.

Feature 1: Customized Taste Profiles

The goal of this feature is to profile tastes based on history of preferences.

  • Taste profiles can improve future visits to the same or similar restaurants
  • Users can access each others’ profiles and see what their friends ordered

Feature 2: Items Combinations

The core idea of this feature is to increase satisfaction for the entire dining experience instead of focusing on individual dishes.

  • Give restaurants the opportunity to introduce combinations of items that may otherwise never be discovered

Feature 3: In-Restaurant Experience

The goal of this feature is to prioritize relevant information for in-restaurant dining.

  • Users can look up recommended items on the spot
  • Friends recommendations can increase credibility of recommended items

Key Findings

Key Findings For Three Features

Enabling Users to Make Fast and Informed Decisions

After analyzing key findings, I decided to pursue the In-Restaurant Experience feature. This feature is intended to solve the problems of:

  1. Indecisiveness: not knowing which items are more suited for specific taste preferences
  2. Inefficiency: waste of time reading through reviews
  3. Credibility of reviews: recommendations from friends are weighted with higher importance

How Other Platforms Execute Restaurant Profiles

Other applications also prioritize choosing restaurants and neglect the in-restaurant experience.

“Most apps don’t give that personalized experience.”

Determining The Entry Point

After taking analysis from market research, I explored numerous entry points to distinguish my feature from the current restaurant profiles.

I iterated different entry points: location confirmation on restaurant profile, check in toggle, notification pop-up, and home & highlights toggle.

Medium Fidelity Entry Point Explorations

Explorations A, B & C imply that users should be already seated at a restaurant so that they can “confirm their locations” or “check in”. I considered edge-case users who may want to access these highlights & recommendations before coming to the restaurant. If they must confirm location or check in first, this feature would not be useful for them.

I pursued D: it allows users to go back and forth from “Home” to “Highlights” anywhere and at anytime.

Content Strategy — What Makes the Most Sense?

Originally, I named the two toggle options as “Home” and “Highlights”. After receiving feedback, I decided to re-evaluate the content and consider what makes the most sense.

Content Strategy For Entry Point

Because “Home” seems too broad (A & B) and “Reviews” seems too restrictive (C), I decided to pursue option D.

Exploring the Recommendations Interaction

Recommendations Display Explorations

A — Categories show up upon scrolling. Engages users to keep scrolling.

B — Disturbs visual flow of the page. Too many toggle features may be counterproductive.

To maximize engagement, I decided to pursue A. But after receiving feedback from user testing, I began to question my choice of category naming.

Evaluating Recommendation Categories

“Taste Recommendation”: How do you identify “taste”? Is it based on the users’ previous orders and recommendations?

“Friend Recommendation”: Adds a lot of value to recommendations. Users place more trust in their friends’ suggestions.

“Recent Recommendation”: Is there value to recency of recommendations? Is a recommendation posted 5 minutes ago more important than one posted 50 minutes ago?

After re-evaluating and brainstorming more ideas, I realized the three categories I chose do not best solve my people problem. Additionally, there was no effective way to display the categories in a hierarchy that made the most sense. I decided to redefine my categories to be “Appetizer Recommendation”, “Entree Recommendation”, “Dessert Recommendation”, and “Drinks Recommendation”.

Interaction Flow For Recommendations

Medium-Fidelity Recommendations Flow

The recommendations flow is now more efficient and intuitive with the change of categories, but I wanted to explore more iterations with content requirements for each recommendation card.

Visual Design

High-Fidelity Visual Design Explorations

A — Have the option of viewing more photos but the featured image may be too small.

B — Easily scroll through images with clear mental model of “liking” a food item by liking the image. Visual design may not fit with Yelp’s current design

C — Featured image clear enough for users to see without drop down images. Fits in with Yelp’s current design.

As a result, I chose option C, which allows users to compare 3 items (A loses this comparison when photos drop down, B requires scrolling).

UI Design — What Information Best Fits?

Focusing on individual item cards, I determined which components were necessary to best convey information for each recommended item.

Item Card UI Explorations

I asked users which UI design was best for conveying “recommend”. Users immediately ruled out B, as star rating is hard to translate into a Yes or No recommendation. C seems unnecessary and crowded up the card too much. Between A & D, I decided to pursue A because a “like’ icon is more intuitive for recommendation whereas a “heart” icon conveys “favoriting” or “loving”.

Iteration for Item Card UI

I decided to pursue A to separate the popularity metrics and the social metrics. From feedback, users prefer a separation of information from the public and from their friends.

“Like” Icon Placement Iteration

I realized it may be misleading for users who would be inclined to click on the item cell. I decided to center the placement of the “like” icon to make it the main focus of the card.

Final Interaction for Recommendations

Final Interaction for Recommendations

Origami Studio Final Prototype

Understanding Yelp UI

UI Kit

What’s in it for Yelp?

After creating this case study, I realized that there are many opportunities spaces that Yelp could work on to focus more on the “during” and “after” experiences. Look at Grubhub, DoorDash, Zomato, TripAdvisor, and Yelp. All these local business discovery apps target mostly the “before” process. A recommendations feature in Yelp would differentiate Yelp from other local business services, thus increasing Yelp’s competitiveness in the industry. This would also incentivize businesses to streamline their items information and images.

Conclusion

I have always been an avid fan of traveling and discovering local cafes and restaurants, and I love designing to connect people with unique experiences. For these reasons, Yelp has always been one of my favorite brands.

Developing this case study was a great learning experience for me. I learned to iterate and re-iterate as much as I could, actively ask and listen to feedback, and be completely open to scrapping initial ideas. I also learned to think beyond the interface and to consider everything — from the humanities of how people think and use products to Yelp’s business goals.


This is a case study for a project in Intro to Digital Product Design. I am in no way affiliated with Yelp.

Like what you read? Give Cathy Liu a round of applause.

From a quick cheer to a standing ovation, clap to show how much you enjoyed this story.