Steam Reviews Scraper — Sentiment, Playtime & Review-Bomb Detection

Extract Steam game reviews for any app — as clean, structured JSON, built directly on Valve's own public review API. No login, no browser, no proxies, so it never breaks the way headless-browser scrapers do.

What you get per review

{
    "review_id": "180325487",
    "app_id": "730",
    "app_name": "Counter-Strike 2",
    "recommended": true,
    "text": "Been playing since 1.6, still the best.",
    "language": "english",
    "created_at": "2026-06-01T14:22:03+00:00",
    "updated_at": "2026-06-01T14:22:03+00:00",
    "votes_helpful": 12,
    "votes_funny": 0,
    "weighted_vote_score": 0.63,
    "comment_count": 1,
    "steam_purchase": true,
    "received_for_free": false,
    "written_during_early_access": false,
    "playtime_at_review_minutes": 4820,
    "playtime_forever_minutes": 9143,
    "playtime_last_two_weeks_minutes": 620,
    "author_num_reviews": 8,
    "author_num_games_owned": 143,
    "author_steam_id": "76561198012345678",
    "primarily_steam_deck": false,
    "hardware_os": "Windows 11",
    "hardware_cpu": "AMD Ryzen 5 5500",
    "hardware_gpu": "NVIDIA GeForce GTX 1650"
}

Built-in analytics (free, not billed)

When includeSummary is on (default), one extra summary record per app is pushed alongside the reviews:

{
    "type": "summary",
    "app_id": "730",
    "app_name": "Counter-Strike 2",
    "steam_review_score": 8,
    "steam_review_score_desc": "Very Positive",
    "steam_total_positive": 1264535,
    "steam_total_negative": 214396,
    "steam_total_reviews": 1478931,
    "sample_size": 200,
    "review_bomb_alert": { "detected": false, "flagged_days": [] },
    "top_pros_keywords": [{ "word": "gameplay", "count": 14 }],
    "top_cons_keywords": [{ "word": "cheaters", "count": 9 }]
}

• Review-bomb spike detection — flags days where review volume spikes far above the sample's median, a signal used to catch coordinated review-bombing or a patch/controversy blowing up
• Pros/cons keyword extraction — top recurring words split by recommended vs. not-recommended reviews, a quick read on why players like or dislike the game
• Steam's own review score attached to every run, so you can track sentiment drift over time on a schedule

Why this scraper

• ✅ Never breaks — built on Valve's official public review API, the same one the Steam store website itself uses
• ✅ Richer than a UI scrape — hardware specs and playtime at the moment of review aren't visible on the Steam store page itself, only through the API
• ✅ Multi-game in one run — track your own game plus competitors side by side
• ✅ Fast & cheap — no browser, no proxies, no login
• ✅ Review-bomb and sentiment signals out of the box

Use cases

• Indie devs & publishers: monitor sentiment after every patch, catch a review-bomb or cheating wave early
• Game marketing / ASO agencies: mine competitor reviews for feature gaps and messaging
• Market research & investment analysis: track public sentiment on gaming studios' flagship titles over time
• AI & NLP pipelines: feed structured review text into sentiment/LLM analysis

Input example

{
    "appIds": ["730", "1245620"],
    "reviewType": "all",
    "filter": "recent",
    "language": "all",
    "purchaseType": "all",
    "maxReviewsPerApp": 200,
    "includeSummary": true
}

Filters

• reviewType: all / positive / negative
• filter: recent (newest first), all (Steam's helpfulness-ranked all-time top reviews), updated
• language: any Steam language code (english, schinese, german, ...) or all
• purchaseType: all / steam / non_steam_purchase

FAQ

Is this legal? The actor reads Steam's publicly available review API — the same data anyone sees on the store page. No login, no private data.

Where do I find an app ID? In the game's store URL: store.steampowered.com/app/730/CounterStrike2 → 730.

How many reviews can I get? As many as Steam has for that filter — pagination follows Steam's cursor until exhausted or your maxReviewsPerApp cap is hit.