Tacter: AI Coach Match Review.
Tacter is a gaming platform that combines content, tools, and data-driven insights for competitive players, with Teamfight Tactics as its flagship product. Despite strong visibility and steady traffic, retention and monetization were below expectations. Users consumed content but rarely returned after playing a match, and there was no paid feature compelling enough to justify a recurring subscription.
Our challenge was to design a product experience that players would actively use after every game, delivering clear, personalized value while creating a sustainable monetization loop.
Problem
Most Tacter users relied on static guides and meta content but lacked personalized feedback to understand their own gameplay. Through conversations with high-level players, content creators, and professional coaches, a recurring insight emerged: many players remain stuck in the same rank for long periods without knowing why.
While a small percentage of players pay for personal coaching to identify mistakes and improve, the vast majority do not have access to actionable, match-specific guidance. Players wanted to improve and climb, but existing solutions failed to explain what they did wrong in their own games or how to fix it. This revealed a clear gap between player intent and available solutions.
Opportunity
Solution & Process
Solution
We identified an opportunity to democratize coaching-like insights by transforming raw match data into clear, actionable recommendations. The system would analyze the full context of a player's match, including final placement, in-game decisions, opponent compositions, and meta data.
By combining this information with our internal database of user-generated compositions and publicly available competitive knowledge, the product could generate concise summaries highlighting key improvement areas after each match.
To ensure the insights were accurate and meaningful, we collaborated with professional coaches and well-known content creators. Together, we defined an initial system prompt and iterated on it over time, progressively refining it into a robust prompt framework that translated expert knowledge into scalable, AI-generated feedback.
Process & Iteration
I joined the project after the initial MVP of the Match Review feature had already been designed and shipped. While this first version validated technical feasibility, early usage quickly exposed major usability, clarity, and insight-quality issues that were limiting adoption and engagement.
Users had to scroll excessively to find their matches, match cards contained too much information, and the experience felt overwhelming. Many users did not clearly understand what Match Review was, how to access it, or why they should generate a new analysis after each game. As a result, interaction rates were lower than expected, especially around generating new analyses.
Rather than treating this as a one-off redesign, we approached the feature as an iterative product process. We established a continuous feedback loop with more than 20 users, including professional coaches, content creators, and players across different skill levels. These users generated analyses from real matches and provided detailed feedback on what felt useful, unclear, incorrect, or misleading. Some insights lacked context, others needed clearer framing, and certain recommendations had to be removed entirely to avoid confusion or misinformation.
Based on these insights, we fully redesigned the experience with a strong focus on simplification, clarity, and speed. We reduced cognitive load, simplified match cards, restructured the interface, and iterated on the underlying System Prompt to improve the quality and reliability of AI-generated insights. This work was executed with a strong bias toward shipping, measuring, and iterating.
The redesigned experience led to a clear improvement in key metrics. The number of generated analyses doubled, users opened more detailed match views, and overall engagement across player profiles increased significantly, reinforcing trust and perceived value in the feature.
Onboarding & Activation
One of the most important insights during this phase was that linking a summoner profile was the main activation driver for the product. Users who linked their summoner were far more likely to request match analyses and eventually convert to paid subscribers.
To reinforce this behavior, we made two major product decisions. First, we changed the app's default entry point. Instead of landing on content pages, users now landed directly on the summoner page, making the absence of a linked summoner the primary empty state.
Second, we redesigned the onboarding flow. Although nearly 50 percent of users were already linking their summoner, we refined copy, visual hierarchy, and messaging to clearly communicate that without linking a summoner, users would miss the core value of the platform. Registration was simplified, and a clear double confirmation was introduced to accelerate time to value and reduce friction later in the funnel.
Web to App Funnel Optimization
Since most users initially interacted with Tacter through the web, we also focused on strengthening the web-to-app funnel. We redesigned the Match History experience on web to position the app as the primary destination for full match analysis and insights.
This included clearer calls to action, mobile-optimized banners, and repeated entry points encouraging users to link their summoner and download the app. These changes significantly increased traffic from web to app and reinforced the full engagement and conversion loop.
Impact
2×
Increase in generated analyses
The number of generated match analyses doubled, indicating stronger perceived value of post-match insights.
Higher engagement
Deeper product usage
Users opened more detailed match views and interacted more frequently across key product features, especially player profiles.
3×
Web-driven app traffic
Web-driven traffic to the app tripled after optimizing entry points and positioning the app as the destination for full analysis.
Learnings