Methodology

Store data comes straight from Steam: the official Web API and the store's own catalogue endpoints (app details, prices per region, reviews, tags, categories, player counts, news, achievements and Steam's charts). Nothing is copied from third-party sites.

YouTube coverage is found with YouTube's public search, one request per second at most, and checked with the YouTube Data API. Twitch activity comes from the official Twitch API, sampling every live stream on Twitch every ~15 minutes.

• Catalogue, prices, reviews: every app, daily; prices for 40 store regions.
• Players: the most played games hourly, every game with reviews daily.
• Charts: hourly.
• YouTube: games are searched by popularity, new demos first; followed and attached games several times a day; everything else re-checked weekly to monthly.
• Twitch: a snapshot of every live stream every ~15 minutes.
• “Update stats” (signed-in users) refreshes one app on demand: Steam data first, then YouTube and Twitch.

A 0–100 coverage score with a fixed target: 300 videos about the game uploaded in the last 30 days score 100. Below that it grows steadily — 100 videos score 30, 200 score 65:

v = videos uploaded in the last 30 days
score = 0.3 × v                  (v up to 100)
score = 30 + 0.35 × (v − 100)    (v from 100 to 300)
score = 100                      (v 300 or more)

Videos that have no upload date yet (Shorts before their API check) count from the day we found them, so a freshly searched game is not under-scored.

The badge color shows the level: purple under 25, red 25–50, yellow 50–75, green 75 and up. Videos count when their title names the game (or, for round-ups, the description) — trailers and the developer's own channel are excluded. Demo videos count when they say “demo” or “Next Fest”.

Every ~15 minutes we record every live stream on Twitch and the Steam game it is playing (Twitch categories are matched to Steam apps through IGDB). From those snapshots: viewers and live channels per hour and per day, hours watched (average viewers × hours), and per streamer the hours streamed (snapshots ÷ 4), average and peak viewers.

The Twitch score uses the same rule as the YouTube score, counting different streamers who played the game live in the last 30 days: 300 streamers score 100, 100 score 30, with the same badge colors.

Prices are read from the Steam store in the US and 11 more regions every day and in 14 further regions (Türkiye, Argentina, Russia, Ukraine, Switzerland, Norway, Sweden, South Africa, Taiwan, Hong Kong, Singapore, Indonesia, Thailand, Chile) every week; every change is kept, so the price history and the historical low of a game start on the day we first saw it (November 2026) — older lows are unknown to us. USD equivalents use the European Central Bank's daily reference rates; the "cheapest region" compares those equivalents and ignores Steam's regional purchase restrictions. "Best deals" ranks discounts by cut × review quality × how recent the game is; "New deals" are discounts that started since the last daily pass. Followers can set a target price, a minimum cut, a region and historical-low, release / Early Access exit and new-build alerts on each game; the checks run once a day.

Steam publishes a review histogram for every game: how many positive and negative reviews it received per month since release (and per day for the last 30 days). We collect it for every game with 10 or more reviews and show it on the Reviews tab.

From it we derive estimated sales per month: the month's reviews × the game's owner multiplier (the same multiplier behind the owner estimate above, so the months add up to that estimate), and estimated revenue = sales × the US list price × 0.55 (regional pricing, discounts and bundles). Free games get no revenue estimate. Like every estimate here it is rough by design — it says when a game sold, not exactly how much.

Top 250 uses Steam 250's default Torn score: score = positive / total × (1 − (total + 1)−1/3) — the positive share, damped by a cube-root bias correction so a 95% game with 20 reviews does not outrank a 94% game with 50,000. Games with fewer than 10 positive reviews are excluded; positive = round(reviews × % positive). Shown ×10. The year, platform, price and tag lists are the same score restricted to that group; the arrows compare with the previous nightly build.

Hidden gems: the Wilson 95% lower bound of the positive share minus 0.035 × log10(reviews / 10), so approval counts but popularity sinks a game off the list. Bottom 100: lowest Torn score among games with 100+ reviews.

Players (SteamCharts-style): current players and peaks come from Steam's current-player endpoint (top 1,000 games hourly, every game with 20+ reviews daily); the monthly table averages our daily averages, hours played = Σ daily average × 24; trending = change versus the average of the samples taken 23–25 hours earlier (games with 100+ players); records are all-time peaks since we started tracking a game, not since 2012. Steam users online is Steam's own 5-minute series. Chart positions are snapshotted once a day, so each chart shows movement arrows and days on chart.

Steam does not publish sales. We estimate owners from the review count — the widely used Boxleiter method: owners ≈ reviews × a multiplier. Older games had fewer reviews per buyer, so the multiplier depends on the release year:

Blockbusters get far more reviews per owner, so the multiplier is capped at 30 above 100,000 reviews and at 20 above 1 million. Free-to-play games use the same multipliers (they have no revenue estimate). We show a range of 0.5× to 1.6× the estimate. Revenue (gross) = owners × the US list price × 0.55, a discount for regional pricing, sales and bundles, before Steam's cut, refunds and taxes. Treat both as orders of magnitude, not accounting.

SteamRadar does not sell or show e-mail addresses — not even partly. Members reach YouTubers, Twitch streamers and game developers with Send message: on a profile, or by ticking several in a list and pressing Message selected (up to 1,000 at once).

A creator is reachable when they published a business contact themselves — on their YouTube About page or video descriptions, their Twitch bio and panels, or a website they link — or when they have an account here. We keep that contact private and only use it to tell them a message is waiting.

Cost. 5 messages a day are free (a message to 3 people counts 3). After that each recipient takes one credit; credit packs go from 100 messages for $4.75 to 10,000 for $325.00. Replies are free. Messaging the same person twice within 24 hours is skipped and not charged; people who cannot be reached are never charged; a credit comes back if the recipient opts out before reading.

Fair use. Up to 5,000 recipients a day per sender. Recipients can report a message and block its sender; a sender reported by three different members within 30 days has messaging paused until we review it.

Every developer and publisher name on Steam gets a page (ranking) with totals over its games. The class is a size band on the sum of the estimated lifetime revenue of its games: Hobbyist under $50k, Indie $50k–$5M, AA $5M–$50M, AAA above $50M — the bands Gamalytic / VG Insights use, so the labels are comparable. Studios that publish for others are counted on both the developer and the publisher side; the score is review-count-weighted. Rebuilt nightly.

Steam does not publish wishlists. For upcoming and recently released games we show followers (the Steam community hub count, read daily) and estimate wishlists as followers × 12, the ratio commonly reported across the Steam-data sites; the realistic range is ×6 to ×20, which we show on hover. Wishlist momentum is the follower gain over the last 7 / 14 days from our daily snapshots (history starts the day we began tracking).

Tag and year pages aggregate every game in the catalogue nightly: releases, free share, average and median launch price (the list price before discounts, paid games only), median reviews, review-weighted score, estimated revenue (sum and median over paid games), owner estimates, a price distribution, the most reviewed games and the YouTube / Twitch leaders. Per-year rows need at least 3 releases. The Top wishlisted chart is Steam's own "popular wishlist" store order, read once a day.

For every Steam game we look up how to reach the studio, in this order: the support contact the developer publishes on the Steam store (about two in three indie games list one there), the developer's own website (home, contact, about and press pages; robots.txt respected), and the studio's YouTube channel About page. Social links (Discord, X, Facebook, YouTube) are listed on the game page. Like creators, studios are reached through SteamRadar messages only — the address itself is never shown. Games are re-checked every 90 days; studios can opt out with their Steam store URL on the privacy page.

When someone messages a creator or studio, we e-mail the whole message to the address the profile lists publicly — one e-mail per message, with the sender's name and a Reply button that opens the conversation here (replying needs a free account; the sender never sees the address). Members who were on SteamRadar in the last 24 hours get no e-mail — the message is simply in their Messages; members who were away longer get the same e-mail, unless they switched it off in their settings. A message can have up to 1,000 words. Every e-mail has a one-click unsubscribe link (and the standard List-Unsubscribe header): it stops all e-mail from us and all messages to the profiles listing that address, and cancels waiting messages. Creators and studios can also claim their profile (“Claim this profile” sends a link to the address we have on file) to get messages straight in their inbox.

• YouTube counts are the videos we found, not every video ever made: for huge games the list is a sample.
• Player counts are our samples; peaks between samples can be missed.
• History (prices, players, reviews) starts on the day the crawler first saw the app.