Sport Analytics Software Case Study

Overview

Responsibility

I was responsible for redesigning performance analytics software for tennis professionals with an integrated ML solution. As a Product Designer within a cross-functional team, I collaborated closely with ML, Backend, and Frontend engineers to ensure seamless implementation. My involvement in this project continued from Q4 2023 to Q3 2024, covering the full design and development support lifecycle.

About Product

It uncovers player’s patterns and translates them into game-changing insights. These insights empower players to build strategies using a data-driven framework, ultimately increasing their chances of winning.

Before, Tennis Experts manually reviewed match videos to analyze essential data points such as court positioning, tactical situations, shot selection, and more. This process took over 8 hours per 3-hour match, posing significant challenges in terms of time and scalability.

🎯 Challenges

Understanding expert workflows: Identifying how experts analyze matches and which parts of the process could be automated to maximize value.
Balancing usability and innovation: Ensuring that the proposed improvements complemented the existing app, essential for the MVP phase, maintaining a user-friendly interface and seamless adoption of the new workflow.
Preserving human input: While many tasks could be automated, it was vital to leave room for human experts to review results. This balance allowed experts to focus on providing personalized, qualitative insights while leveraging ML for repetitive tasks.

📈 Impact

Empowered experts to focus on delivering higher-quality insights.
Reduced the time required to process matches up to 70%

Discovery

Through workshops, we collaboratively defined the product vision and desired outcomes using the Lean Canvas Framework. These sessions enabled alignment among stakeholders while clarifying the broader objectives. Additionally, we talked directly to analytical experts to gain a deeper understanding of their current workflows, pain points, and areas for improvement.

A key takeaway from this research:

Identification of two primary user roles: the expert, responsible for analyzing the game, and the player, who seeks actionable insights to enhance performance.
Experts review the match first, and on the second time, they manually set the dots on the court schema to track the ball position aligning with the seen on the video.

Leveraging the Customer Journey Mapping approach, I outlined an optimized workflow that addressed user needs while accounting for technical constraints. This process helped to visualize how the new solution could seamlessly integrate into existing routines while delivering maximum value to both user groups.

Review and Adjustment:

Once processing is complete, the expert reviews the ML-processed annotation. They can set advanced filters for specific shot types (e.g., backhand shots) and adjust detected elements like player identity, ball position, or court placement if inaccuracies are found.

“

Before automation, experts manually checked ball positions on the court projection and watched the video separately on another device. The new system automates ball landing detection, significantly reducing manual effort.

Completing Report:

After making adjustments, the expert saves the work, and a detailed report that visualizes ball placements is generated. This report includes advanced filtering options based on shot type and placement, allowing experts to focus on analyzing patterns and adding personalized notes.

The end-user (the player) receives the report, typically via their coach, enriched with expert notes and recommendations. Players can also interact with the data by selecting specific shots and viewing their replays within the video for deeper insights.

Usability Testing

“

Visually more appealing, vibrant, and refined compared to the previous version. The navigation is intuitive and straightforward.
‍
– Feedback from Usability Testing Sessions

Before the MVP launch, it was conducted a Usability Testing. Specifically, we have tested the reports page that would be available for external clients. We were interested in whether potential users can easily find all they usually need on the completed report page after navigation changes compared to the previous app version. The 5 sessions were facilitated with tennis players and tennis experts.

✍️ Main Takeaways:

All the users understand the basic concept of the shots on the court projection and the color’s meaning. Users with a background using the initial app had a good knowledge of each of the forms of the shots on the court, when new ones used description in the filters.
Some improvements in navigation and filtering were applied. However, most of the users found them self-explanatory and straightforward.
All of the users are more likely to click the shot on the court to see it in the real play. It’s one of the core features, we were testing.

💡 My Learnings and Insights

Cross-functional communication is essential, especially in understanding how learning works. This alignment helps in designing feasible interfaces and delivering faster.
Preserve what works: It’s crucial to identify genuine pain points and gaps without disrupting what already functions well. Retaining familiar patterns and terminology ensures adoption and ease of use for existing users.