Steam Review Intelligence 🎮 — Reviews → AI Dev Report

Stop reading thousands of Steam reviews by hand. This Actor scrapes a game's Steam reviews and turns them into an actionable intelligence report written for game studios, publishers, and indie devs — complaint taxonomy, feature & patch requests, refund-risk signals, review-bomb detection, competitor mentions, and a prioritized action list.

Point it at any Steam AppID and get back both the raw reviews (for your own analysis) and a ready-to-read AI report (for your next stand-up).

---

What you get

1. Raw reviews — every review as structured JSON: recommendation, playtime, helpful votes, language, timestamp, and full text.

2. AI Insight Report — the reviews distilled into decisions:

Section	What it tells you
Executive summary	Recent sentiment vs lifetime score, and the 2–3 things that actually matter
Complaint taxonomy	Grouped themes with counts, severity, and a representative quote each
Feature / patch requests	What players keep asking for, ranked by frequency
Refund-risk signals	The first-session failures (crashes, FPS) that trigger refunds
Review-bomb detection	Separates current-build feedback from legacy grudges & protest votes
Competitor mentions	Which rival titles players compare you to, and why
Recommended actions	5 concrete, prioritized moves for the team

Example output (real report)

Run on Cities: Skylines II (200 most recent English reviews):

Summary: Recent sentiment (75% positive) is far above the lifetime "Mixed" score (57%), driven by goodwill toward the new developer. Residual negativity concentrates in three fixable areas: performance on mid-tier hardware (the #1 complaint), the mod platform experience, and DLC pricing. A large share of current negative reviews are launch-era grudges rather than reactions to the current build — making reputation-repair unusually high-leverage right now.

Top complaints: Performance/optimization (31, high) · Crashes & save corruption (16, high) · Content thinner than predecessor (9) · DLC pricing resentment (8) · Traffic AI (7)

Recommended action #1: Make empty-map FPS and first-session crashes the optimization KPI — profile the first 60 minutes on mid-tier hardware. This is the refund funnel, not the graphics ceiling.

Input

Field	Description
appId	Steam AppID (e.g. 949230) or a store.steampowered.com/app/<id>/ URL
maxReviews	How many recent reviews to analyze
reviewType	all, positive, or negative
language	Review language filter (e.g. english, all)
generateReport	Generate the AI report (on by default)

Pricing (pay per event — you only pay for what you get)

Event	Price
AI insight report	$1.95 per report
Reviews scraped	$0.10 per 1,000
Actor start	$0.02 per run

No subscription, no rental fee, platform compute included free. A typical run (500 reviews + 1 report) costs about $2.

Tip: if you set a "Max charge per run" limit, make it at least $2.20 for a full run with the AI report. The Actor always respects your limit — it scrapes and charges only what fits, and skips the report (never charging for it) if the limit can't cover it.

Use it from an AI agent (MCP)

This Actor is callable as an MCP tool via Apify's MCP server — let your agent pull review intelligence on demand.

FAQ

Where does the data come from? Steam's official public review API. Only publicly visible reviews; no personal data behind logins.

How many reviews can I get? Steam exposes reviews via cursor pagination — you can pull thousands per game.

Something broke? Open an issue on this Actor's page. Fixes ship within 24 hours.

---

Built by datakura — data extraction tools that turn messy sources into decision-ready output.