Like many young adults, I love to hang out with my friends, and by hanging out, I mean eating. Many use Yelp to quickly search for nearby restaurants, read reviews, and decide which restaurant to try.
People read reviews to decide where to eat, but they have a hard time reading and trusting them because
- There are few reviews as it’s time consuming to write one
- Reviews are hard to read because they are bulky and contain unnecessary information
- Reviews may be biased and one bad review can throw people off.
Yelp Reviews Aren’t Being Used
Going into this, my original hypothesis was: If the Yelp review feature was easy to use, users would be more willing to read and leave a review. Therefore, I set out to discover why people didn’t leave a review on Yelp and what was the problem.
User Research — Why People Don’t Review
My goal was to identify key pain points in engaging in Yelp’s review feature. Here are some key insights:
- Users use Yelp to find a place to eat.
“I use Yelp to find a place I want to go”
2. Users believe that most reviews are helpful, but they cannot completely trust them because some could be biased.
“I think the reviews are biased so I don’t really read the reviews.”
3. Unless the user feels strongly about the restaurant or has a personal connection with the owner, the default seems to be not writing a review.3.
“I don’t write reviews because I don’t want to.”
“I am lazy. I don’t feel that I need to write a review.”
Refining the People Problem:
At first, I thought the review feature was not used because people don’t care about reviews. However, there were very few reviews, which limits the accuracy.
People read reviews to decide where to eat, but they have a hard time reading and trusting the reviews
As a result, my hypothesis was partially right. I learned that users want to engage in the review feature. Yet, when there are not many reviews people don’t trust the source.
Figuring Out Which Feature to Implement
I recruited my friend, Cammy Wong, as my brainstorming buddy. After exploring, we decided on one opportunity:
- Improving Review Quality: How might we make people encourage people to read reviews? How might we encourage people to leave more reviews?
Improving Review Quality
The larger problem with improving the review quality is encouraging people to write one in the first place. People don’t trust the ratings when there are only a few reviews.
It is a cycle:
1. When there is no clear breakdown of the star ratings, people are uncertain and feel less motivated to write a review
2. When there are not many reviews, people do not trust the source
3. When people do not trust the reviews, they are less likely to read one.
I came up with two features to improve review quality:
- Banning people who leave toxic and biased reviews from leaving a review
- Providing specific guideline to write reviews
I decided to see how to standardize the current rating system.
How Other Products Implement the Review Feature
Multiple apps and web products ask for ratings and reviews.
Many of the apps with the review feature have a star rating system or “like” feature. However, it’s unclear what the stars or a “like” represent. People have different standards of what is important. Also, the reviews are very long and contains unnecessary information, which discourages users from reading them.
Deciding On Content Requirements
Based on user research, I concluded that the star ratings segmentation can help the readability and trustworthiness of the reviews. There were two main advantages with the detailed division of star ratings:
- Users would have to read less text. The segmented star ratings can replace the bulk of the text. This addresses the readability issue.
- Users would leave more reviews, when they have to write less. This addresses the shortage of review issue.
- Users would be able to locate necessary information more easily. This addresses the trustworthiness issue.
User Flow Diagram
After brainstorming, I discovered that people want segmented star ratings, shorter reading, and less writing.
I decided to break down the star ratings into five sections that matter the most to the majority of users. This includes food quality, service, value, cleanliness and location.
Segmenting Star Ratings
The purpose of the review feature is to provide users more detailed and accurate information and to reduce the burden of writing. //Super Important!
There are two different parts to a review — leaving a review and displaying it.
In leaving star rating, there were two approaches.
- First approach
Exploration 1 has all segmented star ratings on one page. This gives an overview of the entire star ratings and allows the user to quickly leave a review in a split second.
- Second approach
Explorations 2 and 3 have one segmented section of star rating per page. This ensures that users focus on one thing at a time.
- Exploration 2 reminds the users which restaurant they are reviewing.
- Exploration 3 asks questions to get users think about specific factors to consider.
I decided to go with the Exploration 3 because the interaction provides further guidelines for users and breaks down the rating process.
In leaving a detailed review, there were two approaches.
- First approach
Exploration 1 and 2 have a writing section that allows users to freely express their thoughts.
- Second approach
Exploration 3 has a few sample answers so users don’t have to come up with own answers. If they want to, they can still write detailed opinion.
I decided to go with the Exploration 3 because people don’t want to write a whole paragraph. Rather, they preferred to choose from sample compliments and complaints.
In displaying the review, there are two approaches.
- First approach
Exploration 1 displays the star ratings on the main restaurant information page. This gives an overview of the segmented star ratings.
- Second approach
Explorations 2 and 3 have a “Show more” button. The breakdown of the star ratings is not visible on the main screen. If users want to see, they can click on the button to see the ratings. This helps to reduce bias. Users wouldn’t be influenced by other people’s ratings when rating the restaurant themselves.
- Exploration 2 has more detailed questions in leaving a review. This is to provide more context for users to consider when rating.
- Exploration 3 has one question for each star rating segment. This is to reduce the amount of reading.
I decided to go with the Exploration 3 for two reasons:
- Seeing other people’s ratings could bias users’ own star ratings.
- People do not want to read a lot.
Final Interaction for Rating Restaurant
The final interaction includes a simple rating system so that users can leave a review quickly and read those in a more organized way. Users can review in detail more quickly with segmented stars and a few given answers for additional comment.
Understanding Yelp Visual Design
Here is the UI Kit that I generated from my analysis.
What I Learned
For the product, I learned that the simple breakdown could enhance the usability of the review feature as a whole. People do not want to read or write too much. Simpler review system could solve both problems.
In terms of the design process and thinking, it was my first full design process. Through this experience, I learned that there could be multiple approaches to the same problem, and there is always room for improvement. While I devoted a lot of time in initial brainstorming, I concentrated less on exploring in high fidelities. In the future, I would like to spend greater time on doing more medium and high fidelity explorations.
The current approach encourages people to read and write a review. The segmented ratings allow people to find the relevant information more easily in the reviews. Also, they can review more quickly. However, this solution does not fully address whether people can trust the review.
In conclusion, I tackled the problem of readability and writability issues well, but there is more work to be done for addressing biased reviews.
This is a case study for a project in Intro to Digital Product Design.