A Yelp Case Study: Re-evaluating where and how we chose to eat.
Do you ever find yourself searching way too long for a place to eat? Are you frustrated when you keep seeing the same chain restaurant pop up? The all too well known “Where should we eat” question can be a burdensome task for some (including me). Even after scrolling and reading reviews for 15 minutes, I often still don’t find the right match. This decision fatigue can be contributed to Yelp’s overwhelming and tedious design layout. Over 10 weeks, I completed an extension case study to improve users’ decision-making process on Yelp.
Problem statement: How might Yelp efficiently help people find businesses (especially small and local spots) that align with their values?
Solution:
1. User Research: Who uses Yelp and what do they want?
Research Methodology: A mixed-method case study with participants between the age of 18–30. Includes online surveys, virtual and in-person interviews, direct observation, and desktop research. I recruited two participants for in-depth interviews and collected 38 responses from the survey. The goal was to grasp the basic needs of Yelp users and find exactly what was hindering them in their search.
- The Survey
The online survey solicited a broad range of user needs so that I could better gauge users’ characteristics and their typical interactions with Yelp. There were two recurring and distinct problems based on the responses: 1) Finding a business requires repetitive actions and forces users to keep track and store information themselves 2) Users are overwhelmed by the mass search results and are unable to distinguish between local, established, and large businesses.
Below are some general stats about Yelp’s UX from the survey.
- The Interviews
My one-on-one interviews provided an in-depth analysis of the app’s features as well as insight into users’ thought process of finding their desired business on Yelp. The following pain points were collected:
Results often don’t meet expectations or needs from the search function.
Users spend much longer on the app than desired.
Filters for search results are confusing, cluttered, and unreliable.
The restaurant summaries often lack users’ desired information.
- Personas: What type of user am I targeting in this extension?
Based on my mixed-methods research, I created two types of people that would be the typical user for this case study.
Persona 1: Samantha Sweeney prioritizes the social and aesthetic aspect of a restaurant while also valuing sustainable practices like veganism. It’s important that she finds local spots that aren’t touristy, staying cogent of a product’s cost to value. She needs a tool that can satisfy her multifaceted needs, beyond just the type and cost of food. Samantha is a prototypical example of young professionals looking for authentic experiences in a city with lots of options.
Persona 2: Carlos Fuentes is looking into stepping out of his daily routine and needs help doing so. He wants an app that will help his decide which businesses are worth his time but also environmentally friendly. He finds it important to explore and discover new business in is area, but is also environmentally conscious and would like to visit sustainable businesses.
- Competitive Audit
In this SWOT analysis, I compared Yelp to review-based competitors: Zomato, Tripadvisor, Amazon, Google Maps.
The takeaway: Zomato and Tripadvisor offer extensive filtering options and even make it easier to find a restaurant close to a specific location or landmark. This allows users to have flexibility when filtering and enable them to find restaurants or businesses that best meet all of their needs. Additionally, a clean layout leads to concise interfaces that help expedite the decision making process. Apps like Amazon or Google Maps have a lot of text and and different features that can be overwhelming to the users. However, having features like their side-by-side comparison charts allow users to easily compare products within the same page and reduce the amount of clicks. A successful product should be able to combine this comparison feature with a clean interface.
2. The Design Process: Creating a hi-fidelity prototype.
Note: I initially started out with two types of flow charts that developed into the LoFI stage but for simplicity’s sake, I will only follow the final version’s development (Prototype B).
UX Flow Chart
The first step in my prototype process was to make a chart of a user’s workflow in Yelp. This UX flow utilizes and expands Yelp’s current preference feature.
When searching for a general term, the user can further narrow down the search by selecting the “See More of What You Love” option (currently exists within Yelp). The current feature, however, attempts gathering broad interests that are unrelated to the search item. In this user flow, this feature would show related terms/preferences to the searched item. The user will see and compare recommendations and accuracy scores (similar to Netflix) of several businesses that pertain to their preferences. Since Yelp is review based, the customer’s appraisal of a business determines the accuracy score.
UI Sketch
After the flow chart, I assembled rough sketches visualizing what each step would look like.
Low Fidelity Prototype
The main focus of this extension was to create a filter that would recommend businesses based on past searches and visited establishments. The filter is shown at the beginning of the results page and if the user decides to activate it they can set the percentage of similarity they want their selected business to be based on their preferences. View the model here.
User testing: This prototype went through multiple rounds of user testing before moving on to the next step. The participants found it to be flexible, while providing the freedom to explore results with purpose. They did find frustration with the information organization and were confused at certain navigational points, such as exactly when to implement the accuracy filter. The model needed to balance personalized suggestions with user freedom. Having a simple yet aesthetically pleasing interface also had a positive impact on the users’ experience.
You can find the full notes and observations here (Prototype B).
High Fidelity Prototype
After many iterations, revisiting, reviewing, and testing, I finally reached the end of the process: a clickable high fidelity model.
This finalized prototype allows users to search businesses with more ease. The on boarding feature personalizes the search process and makes finding the right match quicker and smarter. The accuracy feature ensures that the results are relevant and useful to the user’s needs. Finally, the simplified layout makes Yelp’s UX less stressful and more navigable.
3. Final Thoughts
This was my first thorough case study of an application and boy did I learn a lot! I use Yelp regularly so I had a lot of fun re-evaluating its features and accessibility. It was definitely hard to specify which approach and problem to tackle after doing a wide and extensive user research. Here are some things I learned along the way:
You can never have do too much testing. The more people to test your prototype on, the better perspective and scope in the end. I wish I had more time to complete other user tests, but the time constraints of this project wouldn’t allow for that. Each time I interviewed or gathered an opinion, I gained better insight into the user’s perspective and subsequent issues I didn’t see before.
Clean and Concise is usually always preferred. In every step of the process, the overwhelming feedback was geared toward keeping the interface simple and clean. While it may not seem so, this aesthetic allows for way more opportunities and functionality in the space.
The path from start to finish is jagged. Thinking about the initial mindset I had going into this project is funny to look back on, now that I’m at the end of the process. I went through many twists and turns through each stage and especially after user testing. The idea I had at the beginning was no where near where I finished. And that’s a good thing!
Thanks for reading!