Steam Reviews Scraper

Extract Steam game reviews at scale from up to 20 Steam apps per run. Built on Steam's official public review API — no browser, no proxies, no blocking, no login.

Pay only for what you get: $0.10 per 1,000 reviews.

What you get

Every review is returned as a flat, structured record ready for analysis:

Field	Description
appId, appUrl	Steam app the review belongs to
reviewId, reviewUrl	Stable identifiers and direct link
recommended, sentiment	Thumbs up/down as boolean and as readable label
text	Full review body
language	Review language (english, spanish, schinese, …)
authorPlaytimeForeverHours	Total hours played — filter out drive-by reviews
authorPlaytimeAtReviewHours	Hours played when the review was written
authorNumGamesOwned, authorNumReviews	Reviewer credibility signals
votesUp, votesFunny, commentCount, weightedVoteScore	Community reception
steamPurchase, receivedForFree	Whether the copy was bought on Steam or given away
writtenDuringEarlyAccess, primarilySteamDeck	Context flags
timestampCreated, timestampUpdated	Unix timestamps

Use cases

• Game studios — track sentiment after a patch, a price change, or a launch.
• Publishers and QA — mine negative reviews for recurring bug and performance complaints.
• Market research — compare reception across competing titles in the same genre.
• LLM and NLP datasets — large volumes of opinionated, labelled text with an explicit sentiment column.
• Refund and churn analysis — cross playtime against sentiment to see where players give up.

Input

{
  "appIds": ["570", "730"],
  "maxResultsPerApp": 100,
  "sort": "recent",
  "reviewType": "all",
  "language": "all",
  "purchaseType": "all",
  "minPlaytimeHours": 0
}

• appIds — numeric Steam app IDs or full store.steampowered.com/app/... URLs. Up to 20 per run.
• maxResultsPerApp — cap per game (default 100).
• sort — recent, updated, or most_helpful.
• reviewType — all, positive, or negative. Going straight for negative reviews is the cheapest way to find bugs.
• language — Steam language code (english, spanish, german, …) or all.
• purchaseType — all, steam, or non_steam_purchase to exclude key-activated copies.
• dayRange — look-back window in days; applies when sorting by most helpful.
• minPlaytimeHours — drop reviews from players below this many hours.

Finding an app ID: it is the number in the store URL. https://store.steampowered.com/app/570/Dota_2/ → 570.

Output

One dataset item per review. Export to JSON, CSV, Excel, XML or pull straight from the Apify API. Chain it into your own pipeline with the Apify SDK, webhooks, or a scheduled run.

Pricing

$0.10 per 1,000 reviews. You are charged per delivered review, so a run that returns nothing costs nothing beyond the start.

Notes and limits

• Uses Steam's public appreviews endpoint, which is rate-friendly and does not require authentication.
• Filters such as reviewType, language and purchaseType are applied by Steam; minPlaytimeHours is applied locally after fetching.
• Very large catalogues: run the actor per batch of games rather than one enormous run, so partial results are stored progressively.

Feedback

Found a missing field or a bug? Leave an issue on the actor page — fixes usually ship within a day.