OPENGYM
Creating a platform to improve gym efficiency for students.
As project lead in Product Studio at Carnegie Mellon University, I oversaw a team of 15 designer, engineers, and product managers and spearheaded the design.
Role
Head of Product & Design
Duration
6 months
PROBLEM SPACE
Gym facilities at CMU are often crowded, making working out and healthy habits hard to maintain.
This turns what should be a positive workout into a frustrating experience.
SOLUTION SNAPSHOT
With Open Gym, students can gather quick and detailed insight into busyness of gym facilities on campus.
Based on feedback from numerous CMU students and our own experiences, the CMU gym often feels overcrowded, turning what should be a positive workout into a frustrating experience.
RESEARCH
Interviewing 14 students.
We validated the problem that the gym's busy-ness impacted CMU students' gym experience.
and found a few key insights:
Over 90% of gym inflows are directed to the Cohon University Fitness Center.
INSIGHT 1
We should prioritize the Cohon University Fitness Center first floor in our MVP.
Students who don't go to the gym often feel a sense of intimidation, especially for the first floor (where most of the weights are).
INSIGHT 2
How can we reduce uncertainty and pressure for beginner gym-ers? Especially for the first floor?
But as we were listening to people's pain points, we saw this reoccurring statement of wanting more information about how many people or machines were available, or the flow of the gym, etc.
So we found that ultimately, a key pain point lied within wanting more context of occupancy of the gym.
KEY INSIGHT
How might we help students identify busy gym times through occupancy data?
✦ SO ULTIMATELY,
Priority 1 (P0): How might we provide more context to CMU students about the busy-ness of the gyms so they can make better decisions before they go?
IDEATION
So within this space, we began to ideate on few feature ideas that could help tackle this question.
IMPLEMENTATION CHALLENGE #1
However, we knew that a big hurdle would be engineering feasibility.
Since our team included engineers with varying levels of experience, we decided to prioritize a few features for our MVP:
For our MVP, we focused on the first three features, knocking out that fourth feature.
Note: It’s worth noting that implementing machine occupancy tracking would be significantly more challenging. We’d need a system to detect whether someone is sitting at or actively using each machine, and that would require hardware or sensing solutions installed across all units.
But, how did we actually collect the data to complete these features?
We knew that CMU administrators would have some form of that data. Considering COVID, they would need to keep track of how many people were in the gym. So, we reached out to them
IMPLEMENTATION CHALLENGE #2
After several discussions with CMU administrators about data privacy and identifiability, we ultimately were able to receive two types of data:
However, manual headcount data would only be updated every 2 hours, and card swipe data was updated every 15 minutes. And we wanted live data.
So, we engineered a third data stream: a physical laser tracker that counted people entering and leaving the gym in real time.
DESIGNING
And this data helped shape some design directions;
DESIGN EXPLORATION
We focused on the details of how to represent the data in an easily scannable way.
For example, one specific design was when students would go on the platform, how would they be able to tell that this gym was more empty versus full?
Initially, this is the design we came up with:
But then, after conducting a competitive analysis of some of other colleges and how they represented occupancy,
we saw a meter design to be more effective and provided more context into occupancy of the gym. This led to this final design:
DESIGN TESTING
This metered design also raised the question of: numerical representation or percentage?
Should the data be represented in a numerical form, so for example, 15 people out of the 50 max occupancy, or 30%?
After testing on 2 participants, we found something interesting:
Percentages felt easier to interpret and easier to feel how busy the gym was.
When participants looked at numbers ("15/50"), it took longer for them to decide whether it was busy or not as busy compared to percentages.
FINAL DESIGNS
Introducing OpenGym. Providing real time data of gyms on campus.
Detailed View of Each Gym
Quick snapshots of busy versus not as busy times, occupancy graph, and general amenities. Allowing for more detailed view of the gym to reduce gym intimidation and easier gym planning.
Login System Ensuring Security
To ensure security of student data, we established a login that goes through the two-step integration system before showing any sensitive data (ex: occupancy of gyms)
IMPACT
We presented our project at the Google office in front of 4 product managers and 1 product designer. Gratefully, after the presentation, we got much praise, including this one!
"With this presentation, you all would pass the product manager interview at Google.” -Scott Bluman, PM at Google
This platform launched as of February 2025.
A few days after launching, we tested are data and found a 95% data accuracy level and had over 80 users during our launch.
Check it out here! ↴
https://cmuopengym.com/
Note: you need a CMU email to login!
MY TAKEAWAYS
Especially within cross collaborative teams, creating these discussions were able to keep us on the same page and hear everyone ideas.
I would have loved to do more usability testing in the early stages. Especially as the designs were being created, like lofis.
This project had such a huge focus on implementation, and it showed me the importance of considering technical and physical constraints early when designing end-to-end products.