Photo: Zuko.io Images / Wikimedia Commons (CC BY 2.0)
Steam's review percentage is the share of a game's written reviews that are thumbs-up positive rather than thumbs-down negative. It isn't a star rating or an average score out of ten, it's a plain ratio: if 87 out of 100 reviewers clicked thumbs up, the game shows 87% positive. Steam then attaches a word label to that number, something like "Very Positive" or "Mixed," based on both the percentage and how many reviews exist. Noun Town currently sits at 87% positive across more than 590 reviews, used as a worked example throughout this guide.
Short answer: Steam's review percentage is simply positive reviews divided by total reviews, shown as a percentage, not a 5-star average. A game needs at least 10 reviews before Steam shows a score at all, and the word label attached to that percentage, like "Very Positive" or "Overwhelmingly Positive," depends on both the percentage and the total number of reviews.
Every Steam review starts as a simple choice: thumbs up or thumbs down. There's no 5-star slider, no score out of ten, and no half marks. A reviewer picks one of the two options, optionally writes something explaining why, and that's the entire rating mechanism behind every percentage you see on the store.
The displayed percentage is just that ratio scaled to 100. A game with 1,000 reviews where 870 are thumbs up and 130 are thumbs down shows 87% positive, the exact same math whether the game has 20 reviews or 200,000. There's no weighting by review length, how detailed the write-up is, or how experienced the reviewer's account looks.
This is a deliberately blunt system compared to a 5-star average, and that's mostly by design. A binary vote is harder to fudge with a slightly-above-middling 3-star rating that doesn't really say whether someone liked the game or not, and it maps cleanly onto the one question most shoppers actually want answered: did this mostly work out for the people who bought it?
Alongside the raw percentage, Steam shows a short text label summarising it. Valve has never published the exact thresholds officially, but the labels have stayed consistent for years and are easy to reverse-engineer from enough store pages. Both the percentage and the total review count decide which label a game gets.
The review count matters as much as the percentage here. A game can only reach "Overwhelmingly Positive," the top label, once it has at least 500 reviews behind it, which is part of why newer or smaller releases tend to top out at "Very Positive" even with a near-perfect ratio.
Screenshot from the Noun Town Steam store page, showing one of the game's vocabulary scenes
Not every account can leave a review that affects the score, and not every review from an eligible account counts forever. Steam restricts voting rights to try to keep the percentage meaningful.
To leave a review, an account needs to own the game through an eligible purchase, have roughly 5 minutes of recorded playtime, and not be a "limited" account, broadly, one that has spent at least $5 on Steam over its lifetime. Reviews written on activated third-party keys or during free weekends are visible on the page but excluded from the percentage calculation itself.
Reviews also stop counting under certain conditions after the fact. If a purchase is refunded, the review tied to it is removed from the score, since the reviewer technically no longer owns the game. Since March 2019, Valve also excludes reviews it identifies as off-topic review bombing, a sudden wave of negative reviews tied to something unrelated to the game itself, such as a developer's unrelated business decision, rather than the game. Genuine criticism of the actual game still counts, no matter how harsh.
Most Steam store pages actually show two separate scores once a game has been out long enough: "Recent Reviews" and "All Reviews." They can tell noticeably different stories about the same game.
"All Reviews" is the lifetime percentage, every review the game has ever received, back to launch. "Recent Reviews" only counts reviews posted in the last 30 days, and only appears once a game has been live for at least 45 days with enough reviews landing in that recent window. Valve introduced this split in 2016 specifically because a single lifetime percentage can quietly hide a lot of change.
A game that launched broken and picked up a wave of early negative reviews can clean up its act through patches and updates, and the recent score reflects that turnaround within days or weeks. The lifetime score, dragged down by those early reviews, can take months or years to recover, if it ever fully does. Checking both numbers, not just the headline one, gives a much better sense of whether you're looking at a game's reputation today or its reputation from launch week.
Across 590+ Steam reviews, placing Noun Town in Steam's "Very Positive" band
Noun Town currently sits at 87% positive across more than 590 Steam reviews. Using the thresholds above, that places it comfortably in the "Very Positive" band, above the 80% floor for that label and well short of the 95% needed for "Overwhelmingly Positive," which also requires at least 500 reviews before it's even reachable.
That 590-review sample sits inside a much larger pool of players across the Noun Town series as a whole, over 200,000 at last count, and the game has picked up 3 awards along with shortlist nominations for 7 more since launch. None of that is part of the Steam percentage calculation itself, but it's useful context for reading a score like 87%, since it reflects a genuinely sizeable, ongoing player base rather than a tiny early-access sample that could swing wildly with the next ten reviews.
A review percentage is a useful shortcut, not the whole picture. It compresses thousands of individual opinions into one number, which is exactly what makes it fast to scan and exactly what makes it easy to misread.
Review count matters just as much as the percentage itself. An 87% score built on 590 reviews, or a far larger sample across a whole series, reflects a broader range of players, playstyles, and languages learned than a smaller or newer title with the same headline number. Checking the recent score against the all-time one, and skimming a handful of the actual written reviews rather than just the percentage, usually tells you more about whether a specific game fits what you're looking for than the number alone ever will.
If you're weighing up whether a language learning game like Noun Town is worth your time based on its score, the free demo on Steam is a faster way to find out than reading reviews alone, since you get to form your own opinion before spending anything.
Curious what 87% positive actually plays like? Try the free demo yourself.
Try Noun Town on SteamIt's the percentage of a game's total reviews that are thumbs-up positive. An 87% score means 87 out of every 100 reviewers rated the game positively, with the rest marking it thumbs down.
No. Steam has never used a 5-star or 10-point scale for user reviews. Every review is a binary thumbs up or thumbs down plus optional text, and the published percentage is just the share that were positive.
Based on how Steam's labels have behaved for years, since Valve has never published the exact cutoffs, "Overwhelmingly Positive" generally requires 95% or more positive reviews across at least 500 total reviews. It's the highest of Steam's nine review labels.
At least 10. Below that, the store page shows a note that there aren't enough user reviews to generate a score yet, rather than a percentage or label.
"All Reviews" is the lifetime percentage across every review a game has ever received. "Recent Reviews" only counts reviews from the last 30 days, and only appears once a game has been out for at least 45 days with enough reviews in that window. The two can tell very different stories, a game that launched rough but improved through updates can show a strong recent score next to a weaker lifetime one.
Only accounts that own the game through an eligible Steam purchase, have roughly 5 minutes of recorded playtime, and aren't limited accounts, broadly, accounts that have spent at least $5 on Steam. Reviews from activated keys or free weekends are visible but don't count toward the score.
Since March 2019, Valve excludes reviews it identifies as off-topic review bombs, mass reviews tied to something unrelated to the game itself, from the default score. Genuine criticism of the game, even harsh criticism, still counts normally.
It stops counting. Reviews tied to a refunded purchase are removed from the score, since the reviewer no longer owns the game through a valid purchase.
Generally yes. 80% positive sits at the bottom edge of Steam's "Very Positive" band once a game has fewer than 500 reviews, and is still a solidly favourable score at any review count. Context matters too, 80% across 10,000 reviews reflects a much larger, more tested consensus than 80% across 12.
Not necessarily on its own. A niche game with 15 enthusiastic reviews can show 100% positive while a larger, more divisive game sits at 85% across 50,000. Reading a handful of the actual written reviews, not just the headline percentage, usually tells you more about whether a game fits what you're looking for.
Further reading: Valve's own announcement introducing the recent-versus-all-time review split explains the reasoning behind it directly. The FTC's warning on fake reviews covers the broader problem of manipulated ratings across online shopping generally. The Steam (service) Wikipedia page has background on the platform, and the Wilson score interval page covers the statistical method some ranking systems use to account for small sample sizes.
← Back to blog