How we estimate Steam wishlists

Steam doesn't publish wishlist counts, so every wishlist number on GAMEDATA.wtf is a model estimate. This page explains what goes into it, how it's calculated, and how close it has been to the real numbers developers announced.

What we measure

  • Steam followers for each game, collected daily.
  • Steam's global wishlist ranking: each game's position on the most-wishlisted chart.
  • A sample of public Steam wishlists: how many sampled users have each game on their wishlist.
  • Wishlist counts developers announce publicly, used as fixed reference points.

How the estimate is calculated

  1. Followers × multiplier. Wishlists track followers closely, so the base estimate is the follower count times a per-game multiplier. Games without enough history get a formula that also uses their wishlist rank, and the multiplier has a fixed floor.
  2. Sample check. The share of sampled users who wishlisted a game gives a second estimate with a 95% confidence interval (Wilson score). If the follower-based estimate falls outside that interval, the sample-based one is used.
  3. Rank alignment. Estimates are adjusted until their order matches Steam's own wishlist ranking, with publicly announced counts held fixed.
  4. Outlier guard. Follower counts that are implausible for a game's rank are clipped to the typical range for that rank.

Sales numbers for released games use a review multiplier (about 50 copies per Steam review), and revenue is price × estimated copies, gross, before Steam's cut, refunds and regional pricing. The week-1 sales projection for unreleased games is estimated wishlists × a conversion rate that depends on the price band.

How accurate it is

We compared the model's stored estimate with 304 wishlist milestones developers announced in their Steam news posts between September 2025 and October 2026, using the estimate from within three days of each announcement.

21.5%median error, all 304 checkpoints
13.1%median error for the 163 games announcing 50,000+ wishlists
56%of estimates within ±25% (170 of 304)
+12%median bias: the model tends to overestimate

Errors are larger for smaller games. The table below lists the twelve largest announcements, including the misses.

GameAnnouncedDeveloper's countOur estimateError
Windrose23 Mar 20261,500,0001,209,282−19.4%
Heroes of Might and Magic: Olden Era24 Apr 20261,500,0001,270,935−15.3%
Windrose22 Feb 20261,000,000861,329−13.9%
Low-Budget Repairs22 Jul 20261,000,000433,110−56.7%
LEGO Batman: Legacy of the Dark Knight13 May 2026950,000971,891+2.3%
LEGO Batman: Legacy of the Dark Knight4 May 2026900,000891,165−1.0%
REPLACED3 Mar 2026850,000766,903−9.8%
The Mound: Omen of Cthulhu13 Jul 2026800,000791,271−1.1%
Cairn5 Dec 2025700,000463,041−33.9%
The Mound: Omen of Cthulhu23 Jun 2026600,000486,455−18.9%
I Have No Change26 Jun 2026600,000343,630−42.7%
KYORA29 Jan 2026500,000277,086−44.6%

Checkpoints were found automatically in Steam news titles (for example "100K Wishlists!"); headlines cross-posted under several games were excluded. Snapshot as of 13 October 2026.

Limitations

  • Steam currently rate-limits follower collection, so some games show "n/a" until fresh data arrives.
  • Estimates are least reliable for games far down the wishlist ranking and for games with few followers.
  • The review multiplier for sales varies a lot by genre and price; treat sales and revenue as order-of-magnitude figures.

Found a wishlist count we got wrong, or announced yours publicly? Email xiphosure@gmail.com and it will be added as a checkpoint.