Frequently asked questions

StatKraken currently covers Armor Games, Coolmath Games, CrazyGames, Poki, Y8, Yandex Games, and YouTube Playables. More are in preparation and appear here the day they go live — the list on this page is read from the live platform registry, not written into it, so it cannot fall behind.

Most platforms have their full catalog re-read daily; the largest catalogs, Y8 and Yandex Games, are read in rotation over a few days. Each game page shows the exact time its reading was taken, and a freshness badge that judges it against that platform's expected interval rather than against the clock. That distinction matters for platforms we read in rotation: on a catalog large enough that it takes several days to work through, a three-day-old reading is current, not stale, and the badge says so.

A chart needs at least two readings taken at different times. A game discovered yesterday has one, so its page shows the single value it has rather than a line drawn through one point. The same applies to a game on a platform we have only recently started reading: seven-day growth needs seven days of history, and until that exists the figure is shown as unavailable rather than as zero growth. Game pages state how many observations they are based on, so a short series is never presented as a long one.

The public, logged-out pages of the platforms themselves — the same HTML any visitor gets. Most portals publish their game metadata as JSON-LD or a similar structured block, and that is what we read; where they do not, we read the catalog listing their own browse page calls. No private APIs, no partner feeds, nothing behind a login, and no personal data about players or developers.

Because those are different facts. Poki does not publish play counts, so a Poki game has no play count — it is not a game with zero plays. A blank means the platform does not measure it, or has not measured it yet. We never fill one in with a zero, never sum a blank into a total, and never rank a game last for a metric its platform does not report.

They are exactly what the platform displayed at the moment we read the page, stored unrounded. If a figure here does not match what the platform shows now, either the platform has changed since the reading — check the timestamp on the page — or our parser has broken, which is worth telling us about. Snapshots are never edited after the fact, so a correction fixes the parser and everything computed from the series, not the history.

No. History starts on the day we first read a platform, and each platform page shows that date. There is no backfill: the portals do not publish past values, and reconstructing them would mean publishing estimates as if they were readings.

Partly, and the site is careful about which parts. Catalogs, discovery dates and breakout scores compare cleanly. Ratings do not: each portal rates on its own scale — 0 to 5 stars, 0 to 10 points, 0 to 100 percent — computed from a different population, so we always show a rating in its native scale and never publish one averaged across platforms. Growth is the same story: on platforms that publish play counts it is plays gained, on the rest it is votes gained, and those are not the same unit. Every growth figure on the site is labelled with what it counts.

A 0-100 composite that flags a game gaining traction unusually fast: 50 percent seven-day growth, 20 percent velocity — how much of the month's growth landed in the last week — 20 percent acceleration against the previous seven days, and 10 percent recency. It is normalized within the set it was computed over, which makes it the one figure that is comparable across platforms even though the raw growth behind it is not.

Three lists per platform: Top, by the best engagement metric that platform publishes; Trending, by seven-day growth; and New, by recent discovery. They are recomputed after each day's read, from stored snapshots, using published formulas — heuristics, not a black box. A list can legitimately be empty: Trending needs a week of history, so a platform we started reading recently has nothing to put in it yet.

Yes. Every game has its own page with the full series we have for it, charted at the real times each reading was taken. Every developer has a page per platform plus a combined view across all of them, listing their catalog with the aggregate figures that are safe to aggregate.

A StatKraken rating comes from visitors to this site, on one 1-to-5 scale, and is the only score here that means the same thing on every platform. A platform rating comes from that platform's own users on its own scale. The two are always shown separately and are never averaged together, blended, or substituted for one another.

Because not enough people have rated it yet. A score built from one or two votes says nothing, so nothing is shown until a game clears a minimum. Reviews work the same way and are moderated before they appear.

No, and there are no accounts to make. Ratings are anonymous and limited to one per game per device. Nothing about you is stored beyond what is needed to enforce that.

The underlying data is available for business and partnership use. See business API details.

Nothing. There is no paid tier, no sign-in, and no advertising on the site. If that changes, the free browsing experience stays free.

We read only public pages while logged out, take no personal data, identify our requests with a self-describing user agent, honor each platform's robots.txt and rate limits, and store aggregate game-level figures. That is a defensible posture rather than a settled question, and we would rather say so than pretend otherwise. If you operate a platform and want your data out of this site, email us and we will act on it.

Not yet; there is no verification flow. If a page about one of your games is wrong, or you want it corrected or removed, email us and we will handle it directly.

Post it on the feature request board — anyone can add one and vote on the others, and it is what we work from.