Nobody has published a sales figure, and that was checked rather than assumed
Valve’s news endpoint for App ID 2001760 returns 17 posts between 22 January and 12 August 2026. Two of them thank players — the launch-day “Available Now” post and the v1.0.7 patch notes — and neither carries a number. There is no milestone announcement, no “one million players” graphic and no disclosure inside the window. GAME FREAK is not a listed company reporting unit sales, and Fictions has published nothing. So any figure you find for this game is somebody’s estimate, and the useful question is what it was estimated from. That is the rest of this page.
The review pool has doubled and the verdict has not moved
Three readings, all of Valve’s all-language totals: 2,245 reviews on 5 August at 56.4% positive, 3,301 on 7 August at 57.0%, and 4,489 on 12 August at 57.3%. The pool has exactly doubled in a week and the share positive has moved nine tenths of one point. That matters because the standard defence of a launch-week “Mixed” is that the early pool over-represents people with technical problems and will improve as it fills. It filled, twice, and it did not improve. This is the third reading of a claim `/reviews/` first made on two, and it is the first one large enough to be worth calling settled rather than early.
What a concurrent-player count does and does not tell you
2,034 players were in the game on Steam at the moment of the read on 12 August. Two caveats travel with that number and both are larger than the number itself. First, Valve’s public endpoint reports the current count and nothing else — no peak, no history, no daily shape — so a single read cannot distinguish a quiet Wednesday morning from a decline. Second, and more important here: it counts Steam only. This is the first GAME FREAK title on Xbox Game Pass at launch, it shipped day one on PlayStation 5 and Xbox, and none of those players appear in it. For a game with a day-one subscription release, the Steam concurrent count is a minority of the audience by construction, and treating it as the population would understate it by an unknown factor.
The estimate everyone quotes, and why this page prints a range instead
The only published bridge from a review count to an owner count is a multiplier, and the multiplier is not a constant — it drifts with genre, price, how loudly a game’s players feel about it, and whether the game prompts for reviews. Published estimates for PC releases sit somewhere between twenty and eighty owners per review, which is a factor of four between the ends. Applied to 4,489 reviews that is a band, not a number, and the panel above lets you move the assumption and watch the answer move with it. Articles in the form “X has sold N copies” are almost always this arithmetic with the multiplier left out; the arithmetic is fine and leaving it out is what makes the result worthless.
The completion curve is the one population figure that is a real measurement
Valve publishes the share of players who unlocked each achievement, and those are counted rather than estimated. Nearly everyone who launched the game reached “Purifier’s Mantle”, the first marker; a fraction of one per cent reached “Duty Fulfilled”, the last. The live figures for both, and the whole 46-step curve between them, are on the achievements page — they move daily, so this page names the achievements and links out rather than typing numbers that would be stale by the time you read them. What the curve cannot do is produce a headcount in either direction: a share has no denominator attached, so it describes the shape of the audience and never its size.
What would change this page
A published figure from the publisher, at which point every estimate here becomes irrelevant and the honest thing is to say so and keep the estimate only as a record of how far off the method was. Failing that: more concurrent readings, which turn one instant into a shape, and further review-pool readings, which are the only counted series this page has.