Unlocking Potential: Buff Technology's Journey to Data-Driven Decision-Making

The Challenge

Buff, an Israel-based video game reward platform hosting over 1,500 mobile and desktop games with 450,000+ daily active users, had accumulated a significant amount of user behavior and gaming data. However, without an in-house data team, this data sat largely unused. Critical business decisions — marketing budget allocation, game portfolio management, player engagement strategies — were driven by intuition rather than analytics.

The consequences were tangible: marketing spend was distributed across channels without understanding which ones delivered the highest-value players, retention strategies were generic rather than targeted, and the product team had limited visibility into which gaming experiences drove the most engagement and monetization. Buff needed to transform their data from a passive byproduct into an active decision-making asset.

The scale of Buff’s data made the challenge particularly interesting. With 450,000+ daily active users across 1,500 games, the behavioral data volume was substantial. Every gaming session, reward earned, and engagement interaction generated events that could inform business decisions — but without the infrastructure to process, model, and visualize this data, it was just noise. The marketing team was running campaigns across multiple channels without feedback loops, unable to tell which channels attracted users who actually engaged and monetized vs. those who downloaded the app and churned within a week.

Our Approach

We designed a comprehensive analytics transformation program tailored to the unique dynamics of a gaming rewards platform:

• Data Infrastructure: We built the foundational analytics infrastructure — ingesting player behavior data, marketing attribution data, and gaming activity data into a centralized warehouse. This involved processing millions of daily events from 1,500+ games, requiring careful architecture for both cost efficiency and query performance.
• Marketing Analytics: We built dashboards connecting marketing spend to player quality metrics (not just installs, but engagement and monetization). This revealed which channels were acquiring “tourists” vs. committed players — a distinction worth millions in annual marketing efficiency.
• Player Retention Analytics: We implemented cohort-based retention analysis that tracked player engagement over time by acquisition source, game preference, and reward interaction patterns. This enabled targeted re-engagement campaigns based on predicted churn risk.
• Operational Efficiency: We created operational dashboards for the business team covering game portfolio performance, reward economics, and platform health metrics — enabling faster decisions about which games to feature and which partnerships to prioritize.

We implemented a game portfolio health dashboard that tracked key metrics for each of the 1,500+ games: daily active users, session duration, reward interactions, and revenue contribution. This allowed the business development team to identify which games were driving the most value and which partnerships to prioritize for renewal. We also built A/B testing analytics infrastructure that allowed the product team to measure the impact of feature changes on engagement and retention with statistical rigor, replacing the previous approach of launching features and hoping for the best.

Results

• Transformed scattered data into revenue-focused dashboards used daily by the leadership team.
• Identified underperforming marketing channels, enabling budget reallocation estimated at $50K+ annual impact.
• Player retention analytics enabling targeted campaigns based on behavior-driven segmentation.
• Game portfolio analytics providing data-driven guidance on feature priorities and partnership decisions.
• Scalable analytics infrastructure handling 450K+ daily active users across 1,500+ games.

Technologies Used

Python, SQL, cloud data warehouse, BI dashboards, event processing pipeline, marketing attribution integration.

Project Screenshots