MAS1S, AI-powered sports analytics app
A tennis analytics application, powered by AI, designed to enhance the performance of amateur tennis players.
My role
Project lead and UX researcher & designer
Team
1 AI researcher & 1 tennis coach
Client
Variable Latente
Timeline
06/23 - 09/23 (3 mos)
Introduction
"Variable Latente" is a groundbreaking startup using advanced artificial intelligence to enhance tennis performance analytics.
We address the challenges of current options, where amateurs must choose between costly solutions or rely on mobile phone cameras for less precision.
In response, our product introduces a groundbreaking business model, deploying high-quality cameras within tennis clubs, allowing them to recoup investments through revenue-sharing via subscriptions. This ensures superior statistical accuracy and simplifies the user experience, requiring just a few effortless clicks to start match recordings.
Challenge
Overcome the limitations of existing tennis performance analytics solutions that forced amateur enthusiasts to choose between costly, high-precision options or more affordable ones reliant on mobile phone cameras.
Proposed Solution
An app that integrates an intuitive user interface, along with the functionalities of an AI-powered sports analytics app, to simplify the recording process, enhancing match analysis, and providing a comprehensive and user-friendly tennis performance tracking experience.
Success metrics
Time on task: TOT < 30 sec Task success rate: TSR > 90%
Tools
Figma | User interviews | Affinity maps | Workshop | Usability Hub | UserTesting
Our user
〰️
Our user 〰️
The concept
〰️
The concept 〰️
Introducing MASIS
〰️
Introducing MASIS 〰️
The process
〰️
The process 〰️
SWOT analysis
Due to the significant complexity of AI-powered sports analytics apps, I decided to conduct a SWOT analysis on applications Swing Vision, Seven SIx, Dartfish, Kinovea and Tennis Commander to obtain a comprehensive assessment of the current state of these products and to gain a holistic understanding of the business landscape.
-
1. Swing Vision's AI features for automatic data collection and real-time ball tracking.
2. Dartfish's ability to provide audio feedback and personalized coaching.
3. Kinovea's extensive tools for motion analysis and annotation.
4. SevenSix Tennis's AI-powered swing analysis for various shot types.
5. Tennis Commander's real-time performance analysis and match-making features.
-
1. Swing Vision is not available on Android devices.
2. Dartfish is relatively expensive compared to some other options.
3. Kinovea is limited to Windows operating systems.
4. SevenSix Tennis currently supports a limited range of shot types.
5. Tennis Commander is still developing some planned features.
-
1. Increasing interest in tennis and technology-driven training methods presents growth opportunities for all these apps.
2. Expanding their range of supported devices and shot types can attract a broader user base.
3. Offering more affordable plans or freemium models could attract budget-conscious users.
4. Collaborating with professional players or tennis organizations can enhance their credibility and user base.
-
1. Competition in the tennis analytics app market is intensifying, potentially leading to pricing pressures.
2. Rapid technological advancements could make current features obsolete or require frequent updates.
3. User data security concerns could impact user trust and adoption.
Interviews
Goal
To gather valuable insights and user feedback to inform the design and development of the AI-powered tennis analytics application. The research aims to understand users' needs, preferences, pain points, and expectations in order to create a user-centered and effective application.
Assumptions
We assumed that users bring their mobile phones to the club, they want to measure their performance and access statistical data from their matches to improve their game and they have not used tennis analytic apps previously.
Challenge
The interviewees used a large number of specific tennis-related technical terms that I didn't understand. In some instances, I had to ask them for clarification to contextualize the conversation and maintain the flow of their narrative.
Tool
Conducted 10 remote interviews with tennis enthusiasts, ages from 25 to 66, where the majority, approximately 70%, consist of male participants, and then consolidating the results into an affinity map, to distill and organize user interview data into actionable insights and patterns.
What people are saying…
Workshop
As a result of the qualitative research, we have discovered significant potential for redefining the project scope to meet the needs of MASIS users. To achieve this, I conducted a workshop with the clients to agree design decisions and determine whether to include them in the first iteration or not, depending on the current capabilities of the technology they have designed.
Why didn't I conduct quantitative analysis?
While I was keen on quantifying which game statistics MASIS users prefer (absolute values or percentages) and their preferred graph visualization (bar graphs, trendlines, etc), I opted against surveys. Given our research assumptions based on the founders' experience that users hadn't used such apps before, surveying could be complex and risk incomplete or biased responses, participants may perceive the survey as akin to a mathematics exam. Instead, I incorporated these questions into interviews to provide context and clarifications if needed.
During research, it became evident that users want to enhance both their technique and gameplay strategy. To ensure a user-centered design, I shared my findings with the client to explore the possibility of incorporating new features into our app to address these needs. Unfortunately, technology still doesn't have the capability to generate our design solutions, and we will have to postpone it for the second product iteration.
Information architecture
I carefully plan the navigation system during IA creation to ensure a seamless user experience. Visual hierarchy helped me organize and prioritize content, ensuring users can easily navigate and understand the importance of each element. Information architecture wasn't a one-time task; it required continuous iteration and refinement. Given the substantial volume of research-derived information, I consistently conduct tests and evaluations on the IA to ensure it aligns with users' needs and expectations.
Design validation
〰️
Design validation 〰️
Usability test
Goal
Identify design issues, uncover improvement opportunities, usage preferences and gather KPIs: TSR > 80% TOT (record video) < 30 sec
Tools
5 moderated usability test, both online and in-person, with users from Club de Tenis Alicante (4/1-male/female ratio, aged between 28 and 52 years).
Assumptions
The prototype has been designed in English, and language proficiency was not considered in the selection criteria, leading to bias in the TOT data collection.
TOT > target
The application was designed in English; however, for usability testing, language proficiency was not considered as a selection criterion. As a result, the TOT calculation had to be repeated with native speakers since, in the initial test, participants would pause to read and understand the instructions in English.
The result
〰️
The result 〰️
Main user flow
Initiating a match video recording
Initiating a match video recording requires just 15 seconds and six simple clicks.
Matches
Users are interested in accessing match statistics presented both by individual sets and as a complete match overview.
According to our users, the most effective way to improve in tennis is by watching themselves play, analyzing their strategy, and identifying whether they have exploited their opponent's weaknesses. They will have access to video recordings of their matches, from which downtime without gameplay has been removed.
The usability test revealed that users wanted to see a summary of the match highlights.
Users find value in understanding the specific court positions where they receive shots and place the ball as a means to enhance their technique.
By knowing the spin of their shots, they can work on improving the weaknesses in their technique. Activating this feature allows users to identify whether their shots fall into slice (star), topspin (triangle) or flat (circle).
According to our users, coach feedback is crucial for improvement. Shots are categorized by types, and users can share a specific shot with their coach, who can then provide feedback through voice annotations or messages in response.
Trends and Stats comparison
For potential MASIS users, the ability to select a specific time range for comparing statistical data is crucial. Similarly, they wish to track the progression of their skills across matches to quantify their game improvement, which will serve as motivation and aid in planning training sessions with their coach based on areas of improvement.
Share a video with the coach
For our users, their coach's guidance is crucial for enhancing their gameplay, so they can seek feedback from their coach on specific shots by sharing a clip, with the added flexibility of attaching a message or voice note. The coach can then provide feedback on the clip and respond through the app.
What I have learned as a designer and researcher
Discovering the uniqueness of tennis enthusiasts, they share common patterns in their motivation to play, even though they have unique needs.
The importance of selecting the right tools depends on the problem we need to solve.
There were unanswered questions during the interviews (about data visualization preference), and it would have been appropriate to conduct preference tests before the first usability test.
Effective communication with stakeholders prevented redesigns and facilitated a user-centered prototype.
My proposal for future iterations
100% of interviewed participants highlighted the importance to improve technique and strategy, so the priority is to design the virtual coach feature (to improve technique and strategy), followed by the social network and goal planning.
Real on-court testing with real QR code scanning should be conducted for TOT calculation. Calculate learnability as well.
Potential accessibility issues with heat maps must be explored (users mentioned were a bit small).
Given the complexity of the data and the unique user needs, users should be able to customize data visualization.
