In short
- Three things are measured: what players say on Steam, how recently the game was patched, and how today's price compares with its history.
- Fixed rules turn those numbers into one of seven verdicts. There are no hidden weights and no model; the rules below are the code.
- When there is not enough data the verdict says so (“Insufficient evidence”) instead of guessing.
What is measured
Player reviews
The share of positive Steam reviews, overall and in the most recent reviews. A game needs at least 500 reviews to get a verdict. 80% positive or more is well received, 92% or more exceptionally so, and below 40% is mostly negative.
Patch activity
How long ago the game last got a patch or an update post. It only counts where it changes the decision: after a patch in the last 30 days the latest reviews may not reflect it yet, and with nothing for more than 180 days updates may have stopped.
Price against its history
Today's Steam price against the lowest price on record and the lowest of the last 12 months (from IsThereAnyDeal). Within 5% of a recorded low, or 50% or more off, counts as a good price.
From the numbers to a verdict
These are all the rules. A game that fits one of the first three gets that verdict; for the rest the score and the price decide.
- Fewer than 500 Steam reviews → Insufficient evidence.
- Latest-review sample (≥50) below 65% positive and 10+ points under the all-time score → Wait for fixes.
- All-time score below 40% positive → Not for you.
- Score 80%+ and a good price (free, deep discount, near a recorded low) → Buy now; a good score at a price that has been much lower → Buy on sale; otherwise Worth considering. An exceptional score (92%+) at a fair price (little to gain by waiting) also → Buy now.
- Free-to-play games: unless one of the rules above (Insufficient evidence, Wait for fixes, Not for you) applies, the verdict is Free to play — the question is whether to try it, not whether to buy it.
- Score 40–79% (mixed to positive) → Worth considering at a good price, otherwise Buy on sale.
- Patch freshness: a Buy now is downgraded to Worth considering when the last recognised patch is over 180 days old AND the game is online-only (no single-player mode, so updates matter) or recent reviews are 10+ points under the all-time score. For Wait for fixes, a patch within 30 days means recent reviews may not reflect it yet; none for 180+ days means a fix may not be coming. An old patch alone never lowers a verdict, and an unrecognised patch history is never treated as stale.
- Confidence counts available evidence (review volume, recent sample, price history, patch feed) and never exceeds Medium without editor review.
Is the price good?
The Steam price is compared with the lowest price ever seen on Steam (or, if unknown, the lowest at any tracked store). Price signal: within 5% of the all-time low, within 10% of the 12-month low, or 50%+ off → good. If waiting could save under 25% or under $5 → fair (little to gain). Otherwise: on sale but cheaper before, or never discounted beyond 30% → fair; no sale and much cheaper before → wait.
Real examples
Taken from the catalogue as it is right now, so they change when the data does.
Steam reviewers rate it Overwhelmingly Positive (96% of 883,259 reviews positive). Last patch or update post: Jan 14, 2026 (293 days ago). Price: rarely discounted deeply.
Read the full verdict →
Steam reviewers rate it Very Positive (88% of 1,899,870 reviews positive). Last patch or update post: Sep 17, 2024 (776 days ago). Price: price check unavailable.
Read the full verdict →
Only 489 Steam reviews were found; PlayVerdicts needs at least 500.
Read the full verdict →
What never changes a verdict
- What members of this site think. The members' rating is shown next to a verdict, but it is never part of it.
- Who sells the game. No store, publisher or link pays for a better verdict.
- Votes on “Was this verdict helpful?”. They tell us where to look; they do not change a verdict.
- Your taste or your PC. A verdict is the same for everyone; the “For you” and “Can you run it?” parts of a game page are separate.
Confidence
Every verdict shows how much evidence is behind it: the number of reviews (1,000 or more counts for more), a large enough sample of recent reviews, a price history and a patch feed. Two or more of those give Medium confidence, fewer give Low. It never goes above Medium without an editor's review.
How fresh the data is
The site refreshes its data in the background, a few games at a time: reviews and patch posts about every 12 hours, the Steam price about every 6 hours and the price history every 24 hours. Each verdict shows when its evidence was last checked.
What a verdict is not
- Not a review score and not advice: it is a guide to the decision, and a good game can still be wrong for you.
- Steam only for now: prices are Steam's, and a game sold elsewhere or on a console is judged by its Steam page.
- Not always right. A patch can be missed (posts are recognised by their titles) and an edition can differ from the history we have; where we know that, the page says so.
- If a verdict looks wrong to you, tell us which game and why. Report a wrong verdict
The data sources, what is not measured yet and how the site is run are on the About page. Did our price calls hold up? See the track record →