Methodology

Our crawler polls Roblox's public experience data on a rolling schedule, recording a snapshot of each tracked game: concurrent players, total visits, favorites, up/down votes, and access price. Snapshots are timestamped and stored so we can chart history and detect changes.

When a value we watch — thumbnail, name, description, price, or the game's last update time — differs from the previous snapshot, we log a change event. Those events power the change timelines and alerts.

Roblox does not publish per-game revenue, so any single-number "revenue" figure is a guess dressed up as a fact. Instead we publish a band — a conservative low and a higher upper bound — derived only from observable concurrent players.

Our current model assumes, per concurrent player over a 720-hour month:

• about half of concurrent players spend at all, and
• effective monetization of roughly $0.01/hr (low) to $0.05/hr (high) per spending player.

That yields a monthly band per concurrent player. The chart below shows how the estimated band scales with live players:

These figures are deliberately blunt instruments for benchmarking relative size — treat them as order-of-magnitude, not accounting.

The breakout radar ranks how quickly a game is gaining players. For each game we look at its recent snapshot window and compute a velocity: the change in concurrent players, the percent change versus the start of the window, and players gained per hour.

We combine two components into a bounded 0–100 score:

• Relative growth — percent increase over the window (weighted 60%).
• Absolute momentum— players/hour, log-scaled so a few giant games don't dominate (weighted 40%).

Scores map to bands: Surging (60+), Rising (25–59), Steady (below 25), and Cooling (declining).

Snapshots are sampled, not continuous, so very brief spikes between polls can be missed. Newly discovered games have little history until the crawler builds it up. Revenue bands are estimates, not disclosures. We keep improving the model and will document changes here.

Questions about the data? See the weekly report for a worked example of these metrics in action.