Designing a Rating System for Meaningful Matches



Overview

This project demonstrates how I found the user’s need for filtering people and created the rating system for the dating app JOOPI, which increased 30% user growth after coming out.
    Info

    • Timeline:  Nov 2025 - Dec 2025
    • Team: Iris
    • My Role:  Design strategy,  Prototyping


    There are already lots of dating apps in the market. How could we create something different?

    After discussing with the marketing team, we realized that the key is always the people, not the app. 
    With that being said, it’s important to help the user select the right person.

    A gap appeared 
    After doing a bunch of street interviews with the marketing team,
    we found that people sometimes had bad experiences with the date, but then they just unmatch and move on to the next one


    Why don’t we keep the bad experience away from the other users?

    That’s how I thought about the rating idea. Rate the profile so we can know if we will promote them or not.

    But then more questions came out...
    • Is it legal to make the data public?
    • How will we show the rating to the other users? Scores? Keywords?
    • In what way can users rate others?
    • Who is eligible for the rating? 
    • How to trigger the rating in the app?
    • Can users see each other’s ratings?
    • Can the user still rate if they are unmatched?
    • Will the rating be too subjective?
    • Should the user themselves see the scores?
    • ...

    Solutions
    To address these questions, the marketing team, SWE and I discussed a few times to settle down the logic

    To protect the user’s data
    Instead of showing the score, we only show the 3 most rated tags on their profile

    To avoid the rating being too subjective
    The survey questions only focus on the objective sides, such as profile accuracy and the person’s traits (if their apprearance match their photo, if they are being on time and polite, etc.)

    To avoid rating others randomly
    The survey only works if the users have met in person. The survey pop-up appears 5 days after being matched (the typical meet-up time).
    The recommendation algorithm changes if they rate randomly.

    To incentivize the users to complete the survey
    The user will receive a Spark (AKA rose) after the survey

    To make the rating result more accurate
    The keywords will be visible when they have more than 7 ratings
    And the person themselves cannot see the rating


    Having the solutions in mind, I created the rating flow

    Pt 1. Rating

    Pt 2. Viewing the rating
      Impact
    • Helped increase the users by 30%
    • It became the highlight of the app and attracted users from social media



    What I learned 💁🏻‍♀️
    • Empathy - Think from the user’s perspective, how can the user benefit from a feature
    • Think about different scenarios to make the logic thorough
    • The insight trully comes from the users