Since June 28 we've been recording a daily snapshot of the live player count for every game our network tracks — 104 games with a reading on all ten days, straight from Roblox's public games API. Roblox only ever shows you now; the history evaporates unless someone keeps it. We keep it. Ten days is the blink of an eye, and it's already more interesting than I expected.
The headline is concentration. Among the 79 tracked games that had at least 1,000 players on day one, the median game lost about 8% of its live players across the window. The fourteen big games — 50,000 players or more — posted a median gain of about 4%. Same ten days, same platform, opposite directions. The big get bigger, the middle drifts, and the aggregate number that people quote ("Roblox concurrents keep growing") can be true while most individual games quietly shrink.
The movers
The gainers are a lesson in what actually moves a Roblox population:
- Blox Fruits: 200,286 → 279,469 (+40%). No mystery, just gravity — the biggest trading game on the platform having a big summer.
- Pet Simulator 99: 116,398 → 163,776 (+41%), and here's the tell: over the window its listed title changed to lead with "[WORLD CUP]". The game strapped itself to the tournament with an event, and players followed. On Roblox the square-bracket title prefix is the marquee, and it works.
- Dress To Impress: 76,125 → 112,885 (+48%), plus Tower of Hell (+62%) and Evade (+61%) further down the size ladder — school's out across most of the northern hemisphere, and the drop-in games are drinking it up.
The fallers are just as instructive: Berry Avenue (−47%), Welcome to Bloxburg (−51%), and Flee the Facility (−64%) all bled players over the same stretch, and Adopt Me — still the second-biggest game we track — slid 18%, from 392k to 323k. Nothing "happened" to these games, as far as we can tell. That's the point. On a platform where attention is the currency, no news is the news: a game that isn't shipping an update against summer competition is a game losing share to the ones that are.
An honest caveat about our own thermometer
One reading per day is a coarse instrument, and I want to be straight about its main flaw: our snapshot fires when the daily build runs, not at a fixed hour, and Roblox concurrents swing hard across a day. Some of any single game's ten-day move is really an hour-of-day artifact. Two things keep the findings above honest anyway: the day-one and day-ten snapshots we compared happened to land at similar times of day, and the patterns we're leaning on — the big-versus-median split, a +40% move with a visible event attached — are far larger than the intraday wobble. Still, we treat anything under about ±15% as weather, not climate. (Pinning the snapshot to a fixed hour is on the list; the frozen RAP feed taught us to distrust our own instruments first.)
Why we bother
Player counts are the games-side half of the same project as our limiteds history: Roblox's APIs answer "what is it now?" and never "what was it?" — so any question with a trend in it (is this game dying? did that update work? is the market thinning?) is unanswerable unless someone banks the data daily. That's what the trending board on our sister site runs on, and it's why the limiteds market page exists even while its price feed is frozen. Ten days in, the games ledger already distinguishes an event-driven spike from a summer tide from a slow leak. At a hundred days, it'll tell you which games' declines were seasons and which were funerals. Nobody else will be able to tell you that, because nobody else was writing it down.
The trade takeaway, since this is BloxToolbox: population is the demand side of every in-game economy we track. A game gaining 40% of its players is minting new traders who don't know the price of anything; a game losing half its players is a market where sellers outnumber buyers a little more every week. You don't need a value list to act on that — you need a calendar and someone keeping score. As of today, someone is.