About the data.
Recorded observations, with estimates clearly marked.
Game and group tracking
Player counts, visits, favorites and votes come from Roblox’s public APIs. Game collection aims for one observation every ten minutes. Group membership and the combined top-100 player count are sampled approximately every 30 minutes. Responses can be delayed or unavailable; lines connect recorded observations without inventing missing readings.
Current CCU is the most recently recorded concurrent player count. Recorded peaks cover only Ropeak’s observation period. Visits include repeat visits. Approval is positive votes divided by all votes; rating changes are percentage points.
Each combined-player observation covers the chart games present at that moment. Membership can change. It is not a count of the entire Roblox platform. Group rankings cover the largest communities in our tracked index and are not a guaranteed census of every Roblox group.
Playtime
Estimated average playtime divides observed player-minutes by added visits over the same continuous window, up to 24 hours. It needs at least six hours, sufficient recorded intervals, 100 added visits and observed players. Gaps and visit-counter corrections restart the window. This is an estimate per visit, not private creator session analytics.
Monthly revenue estimates
The charts show modelled creator earnings in USD, not disclosed game revenue, player spending or profit. The model combines Roblox’s published creator-earnings benchmarks with the game’s returned Top Earning position. Creator rankings are an imperfect proxy for games: one creator can operate several games and earn through other products.
Within each rank band, the model distributes earnings using a 1/r curve and divides the annual result by twelve. The displayed half-to-double range is a sensitivity illustration, not a measured confidence interval. Actual earnings can fall outside it. The full calculation and primary sources are available by selecting an estimate on Top Earning charts.
History and comparisons
Histories begin with real observations. We never invent earlier data. Longer histories can use aggregated observations to keep storage manageable. Comparisons use a shared time window and player-count scale. Missing samples do not mean zero players.