Steam Reviews Scraper & Monitor (by language)

Steam Reviews Scraper downloads Steam user reviews for any game — review text, recommended/not recommended, playtime, helpfulness votes, Steam Deck and early-access flags — in any of Steam's 30 review languages. It also builds a per-language score breakdown (total, positive, negative, positive % and Steam's score label for every language) and has a monitor mode that returns only the reviews posted since your last run. It uses Valve's public review endpoint: no login, no Steam API key, no proxies.

What does Steam Reviews Scraper do?

• Takes one or more games as Steam store URLs or app IDs (https://store.steampowered.com/app/1245620/ELDEN_RING/ or 1245620).
• Downloads reviews newest first for the languages you choose (english, japanese, schinese, koreana, brazilian, ... or all).
• Optionally adds one language summary row per language so you can compare how each localization is received.
• In monitor mode, remembers the newest review it has seen per game and language, and on the next run returns only newer reviews.
• Writes everything to an Apify dataset you can export as JSON, CSV, Excel or HTML, or send to Slack, Google Sheets, Make, Zapier or a webhook.

Who is it for? Use cases

Who	What they do with it
Game developers & publishers	Schedule monitor mode on your own game and get new (especially negative) reviews in Slack after a patch or a sale.
Localization teams	Compare the positive ratio per language. Real example from a run on 2026-09-24: ELDEN RING is 93.9% positive in English but 75.6% in Japanese.
Community managers	Read every new review in the languages your team covers.
Market & competitor research	Pull reviews of competing titles and feed them to an LLM to extract complaints and feature requests.
Streamers & content creators	Check how a game is received in your audience's language before covering it.

Why use this Steam reviews scraper?

• Per-language score breakdown — one row per language with total / positive / negative / positive ratio / Steam's label ("Very Positive", "Mostly Positive", ...) / share of all reviews. Steam's store page only shows the overall score and your own language.
• Monitor mode — returns only reviews posted since the previous run. The state is kept in a named key-value store in your own Apify account.
• Privacy-safe by default — reviewer SteamID, persona name and profile URL are not collected unless you turn on includeAuthorIdentity.
• Language shortcuts — ja, jp, 日本語, en, zh, ko, pt-br, es, de, fr, ru are accepted as well as Steam's own codes.
• Cost cap respected — if you set Maximum cost per run, the Actor stops cleanly before exceeding it.
• No proxy, no browser — lightweight JSON requests only, 256 MB memory by default.

What data do you get?

Review rows (type: "review"):

Field	Description
appId, appName	Steam app ID and game name
reviewId	Steam recommendation ID
language, languageName	Language the reviewer selected on Steam
votedUp	true = recommended, false = not recommended
text	Review text
votesUp, votesFunny, helpfulnessScore, commentCount	Community reactions
playtimeAtReviewHours, playtimeForeverHours	Playtime when the review was written and in total (hours)
steamPurchase, receivedForFree, writtenDuringEarlyAccess, primarilySteamDeck	Purchase and context flags
authorGamesOwned, authorReviewCount	Reviewer's library size and number of reviews (no identity)
developerResponse	Developer's reply, if any
createdAt, updatedAt, scrapedAt	ISO 8601 timestamps (UTC)
changeType	"new" in monitor mode
authorSteamId, authorName, reviewUrl	Only when includeAuthorIdentity is on

Language summary rows (type: "languageSummary"): language, languageName, totalReviews, positive, negative, positiveRatio, reviewScoreDesc, shareOfAllReviews. Only languages with at least one review are listed (up to 30 rows per game).

The dataset has two views in the Console: Reviews and Language breakdown. A RUN_SUMMARY record in the default key-value store lists per-game counts and errors.

How much does it cost to scrape Steam reviews?

Pay per event. Platform usage is included — you only pay these events:

Event	Price
Actor start	$0.00005 per run
Game checked (each game fetched successfully in a run)	$0.001
Result (each review or language-summary row in the dataset)	$0.0008 ($0.80 per 1,000)

Worked examples (calculated from the prices above):

Scenario	Rows	Cost
1 game, latest 100 reviews	100	$0.081
1 game, 1,000 reviews + language breakdown (30 languages)	1,030	$0.825
10 games × 500 reviews	5,000	$4.01
Monitor 20 games once a day, a day with no new reviews	0	$0.020 per run (≈ $0.60 / 30 days) plus $0.0008 per new review

Set Maximum cost per run in the run options to cap spending.

How to scrape Steam reviews

1. Open the Actor and click Try for free (an Apify account is free).
2. Paste one or more Steam store URLs or app IDs into Games.
3. Pick languages (or leave all), set Max reviews per game, and optionally Only reviews from the last N days, Review type or Purchase type.
4. Turn on Add per-language score breakdown if you want the summary rows.
5. Click Start, then download the dataset as JSON, CSV or Excel.
6. For alerts: turn on Monitor mode, save the input as a Task, add a Schedule (e.g. hourly) and an Integration (Slack, email, webhook).

Input example

{
  "games": ["https://store.steampowered.com/app/1245620/ELDEN_RING/", "413150"],
  "languages": ["english", "japanese"],
  "maxReviewsPerGame": 500,
  "daysBack": 0,
  "reviewType": "all",
  "purchaseType": "all",
  "languageSummary": true,
  "monitorMode": false
}

Input	Default	Notes
games	— (required)	Store URLs or app IDs
languages	["all"]	Steam language codes or the shortcuts above
maxReviewsPerGame	500	Shared across the selected languages, filled in the order you list them. 0 = no limit (big games have 1M+ reviews)
daysBack	0	Only reviews from the last N days (0 = no limit)
reviewType	all	all, positive, negative
purchaseType	all	all, steam, non_steam_purchase
languageSummary	false	Adds the per-language rows
monitorMode / monitorId	false / —	Only new reviews since the last run
includeAuthorIdentity	false	Adds SteamID, name and review URL
requestIntervalSecs	1	Pause between requests (0.5–10 s)

Output example

Real output from a run on 2026-09-24 with the input above (maxReviewsPerGame: 3).

A review row:

{
  "type": "review",
  "appId": 1245620,
  "appName": "ELDEN RING",
  "reviewId": "236042749",
  "language": "english",
  "languageName": "English",
  "votedUp": true,
  "text": "Best game since skyrim 2011 nothing can compare to it even now in 2026",
  "votesUp": 0,
  "votesFunny": 0,
  "helpfulnessScore": 0.5,
  "commentCount": 0,
  "steamPurchase": true,
  "receivedForFree": false,
  "writtenDuringEarlyAccess": false,
  "primarilySteamDeck": false,
  "playtimeAtReviewHours": 2.1,
  "playtimeForeverHours": 2.1,
  "authorGamesOwned": 0,
  "authorReviewCount": 14,
  "developerResponse": null,
  "createdAt": "2026-09-24T13:04:56+00:00",
  "updatedAt": "2026-09-24T13:04:56+00:00",
  "scrapedAt": "2026-09-24T14:18:52+00:00"
}

Language summary rows from the same run:

[
  {"type": "languageSummary", "appId": 1245620, "appName": "ELDEN RING", "language": "english", "languageName": "English",
   "totalReviews": 572456, "positive": 537582, "negative": 34874, "positiveRatio": 0.9391,
   "reviewScoreDesc": "Very Positive", "shareOfAllReviews": 0.4951, "scrapedAt": "2026-09-24T14:18:52+00:00"},
  {"type": "languageSummary", "appId": 1245620, "appName": "ELDEN RING", "language": "schinese", "languageName": "Chinese (Simplified)",
   "totalReviews": 250435, "positive": 227064, "negative": 23371, "positiveRatio": 0.9067,
   "reviewScoreDesc": "Very Positive", "shareOfAllReviews": 0.2166, "scrapedAt": "2026-09-24T14:18:52+00:00"},
  {"type": "languageSummary", "appId": 1245620, "appName": "ELDEN RING", "language": "japanese", "languageName": "Japanese",
   "totalReviews": 6014, "positive": 4548, "negative": 1466, "positiveRatio": 0.7562,
   "reviewScoreDesc": "Mostly Positive", "shareOfAllReviews": 0.0052, "scrapedAt": "2026-09-24T14:18:52+00:00"}
]

Monitor mode (new Steam reviews only)

1. Turn on monitorMode (optionally set a monitorId to name the watch list).
2. The first run fetches up to maxReviewsPerGame newest reviews and remembers the newest review timestamp per game and language.
3. Every later run returns only reviews written after that, marked changeType: "new". If nothing new was posted, the dataset is empty and you pay only the start and game-checked events.
4. Add a schedule and an integration (Slack, email, Google Sheets, webhook, Make, Zapier).

The state is stored in a named key-value store in your own Apify account. Changing the games or languages (without a monitorId) starts a new watch list.

Tips

• Start with maxReviewsPerGame: 100 to check the output before large runs.
• List languages in priority order — the per-game limit is filled in that order.
• Use daysBack for "reviews since the last patch" pulls.
• The language breakdown needs about 30 small requests per game (one per language); turn it off if you only need review text.
• Use reviewType: "negative" with monitor mode for a complaints-only alert feed.

Integrations & API usage

Python:

from apify_client import ApifyClient

client = ApifyClient("YOUR_APIFY_TOKEN")
run = client.actor("panda_studio/steam-reviews-monitor").call(run_input={
    "games": ["1245620"],
    "languages": ["english", "japanese"],
    "maxReviewsPerGame": 200,
    "languageSummary": True,
})
for item in client.dataset(run["defaultDatasetId"]).iterate_items():
    print(item["type"], item.get("text") or item.get("positiveRatio"))

Node.js:

import { ApifyClient } from 'apify-client';

const client = new ApifyClient({ token: 'YOUR_APIFY_TOKEN' });
const run = await client.actor('panda_studio/steam-reviews-monitor').call({
    games: ['1245620'],
    languages: ['english'],
    maxReviewsPerGame: 200,
});
const { items } = await client.dataset(run.defaultDatasetId).listItems();
console.log(items);

cURL (run and get the items in one call):

curl -X POST "https://api.apify.com/v2/acts/panda_studio~steam-reviews-monitor/run-sync-get-dataset-items?token=YOUR_APIFY_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"games": ["1245620"], "languages": ["english"], "maxReviewsPerGame": 50}'

MCP (Claude, Cursor, VS Code and other AI agents): add the Apify MCP server with this Actor as a tool:

{
  "mcpServers": {
    "apify": { "url": "https://mcp.apify.com?tools=panda_studio/steam-reviews-monitor" }
  }
}

Is it legal to scrape Steam reviews?

This Actor reads the same public reviews anyone can see on the Steam store, through the review endpoint that Valve documents for developers (store.steampowered.com/appreviews/<appid>?json=1, "Getting user reviews" in the Steamworks documentation). The path is allowed by Steam's robots.txt, no login is used, and requests are spaced (1 s by default, with back-off on HTTP 429). Reviewer identity is not collected unless you enable it. You are responsible for how you use the data — for example, check your local privacy law before storing reviewer identities.

FAQ

How many reviews can I get? As many as Steam serves through the endpoint, newest first, 100 per request. Popular games have hundreds of thousands per language, so set maxReviewsPerGame or daysBack.

Do I need a Steam API key or account? No.

Can I search by game name? Not yet — use the store URL or app ID (the number in the store URL).

Why is there no reviewer name? Privacy by default. Turn on includeAuthorIdentity if you have a legitimate need, e.g. replying to reviews of your own game.

Is the review language reliable? It is the language the reviewer selected on Steam, the same one Steam uses for its own language filter.

Are the language-summary rows charged? Yes, each row is one result ($0.0008), so a full breakdown is at most 30 rows ≈ $0.024 per game.

What happens if an input is wrong? An entry that is not an app ID or store URL, or a game Steam returns an error for, is listed under errors in RUN_SUMMARY; the other games still run. A valid-looking ID with no reviews simply returns 0 reviews.

Can I get reviews between two exact dates? You can limit to the last N days with daysBack. Arbitrary date windows are not supported.

Changelog

• 0.1 — Initial release: multiple games, 30 languages, per-language breakdown, monitor mode, privacy-safe defaults, cost-cap aware stopping.