How SIB works
Last updated 2 October 2026. Version 2026-10-02.
SIB is a public count of what recent players say. Verdicts are our opinion based on public data. They are not advice, and they are not for sale.
Developers cannot buy a verdict
Paying SIB, or advertising with us, never changes a Buy, Wait, or Skip. There is no sponsored score and no quiet bump for a launch week. The same rules decide every game. An AI summary can explain that verdict in plain English. It cannot change it. SIB may earn a commission on affiliate links. That commission never changes a verdict, a score, or a ranking.
- Buy
- Wait
- Skip
What we read
For each game we call Steam's public store endpoints. No API key, no login, and no player profiles. We keep the review text and the facts attached to it — playtime, votes, language, whether it was a Steam purchase — and we throw away Steam IDs and profile data before anything is saved.
The chart is Steam's own review histogram: one call per game, weekly for busy games and monthly for older ones, covering a year or more. The percentages for 30 days, 90 days, and 12 months are counted from that histogram. Direction compares the last 30 days with the 90 days before them.
Quotes come from a sample of recent English reviews, up to 1,000 on a full backfill, after the filters below. A later run only asks for reviews newer than the ones already stored.
Genres and the console list come from IGDB, matched by Steam app id. Platforms mean where you can play the game. They are not a second sentiment score. Every percentage on SIB is from Steam (PC) players. IGDB data is used under the Twitch Developer Services Agreement for non-commercial use, and we link back to IGDB.
What we drop
- Under 60 minutes played at review (the cutoff is 60 minutes).
- Under 20 characters of text (the cutoff is 20 characters).
- Received for free.
If a review matches more than one rule, we count the first one in that list.
How the trend is labelled
Each ISO week in the last 12 weeks gets a positive percentage and a review count. A week with fewer than 10 kept reviews is marked low data and drawn faded.
Direction compares the last 30 days with the previous 90 days. A move of 5 points or more is Improving or Declining. Anything smaller is Stable. If either window has fewer than 10 reviews, the label is Not enough data. Periods under that count are faded on the chart and marked low data.
How the verdict is decided
Buy, Wait, and Skip come from the last 30 days of Steam reviews, plus the direction label. A game needs at least 10 reviews in that window. Fewer than that, and the card says there is not enough to call.
- Skip when the 30-day positive score is under 70%.
- Buy when it is 80% or higher and the direction is Improving or Stable.
- Wait for everything else that has enough reviews. That includes a score from 70% up to just under 80%, and a strong score that is Declining or still short of trend data.
The paragraph under the verdict, when it is there, is an AI summary of recent English Steam reviews. It explains the verdict in plain English. It has to agree with the rule. It does not see SIB community reviews, votes, prices, or anything a developer paid for. It cannot change the verdict. If there is no summary yet, that block stays empty.
Report an error if a summary is wrong or misses the point. Include the game name.
Creator coverage
Creator videos are links we chose to show. They sit apart from the verdict. They are not a score, and they do not change Buy, Wait, or Skip.
