App Store & Google Play Reviews Scraper with AI Triage Tags

Scrape app reviews from the Apple App Store and Google Play, and get every review back already triaged: topic, severity, churn risk, actionability, competitor mentions, plus your own yes/no labels. No API keys, no prompts to write, no second tool to run the reviews through.

Most review scrapers hand you thousands of raw rows and leave the reading to you. This one answers the questions a product team actually asks: what is broken, how bad is it, who is about to leave, and what can we act on?

What you get for every review

Field	Meaning
topic	One of bug, performance, account, pricing, ads, design, feature_request, content, support, praise, other
topicConfidence	0–1, how clearly the review belongs to that topic
severity	0–3: 0 no problem, 1 minor annoyance, 2 significant problem, 3 blocking / lost money or data / already quit
churnRisk	0–1 probability the reviewer has stopped or is about to stop using or paying
actionable	0–1 probability the review contains something specific a team can act on
mentionsCompetitor	0–1 probability a competing app or service is named
labels	Your custom yes/no questions, each answered with a 0–1 probability
plus	store, appId, country, date, rating, version, title, text, reviewId

Scores are probabilities, not labels: filter churnRisk > 0.7 or sort by severity and pick your own cut-off. Non-English reviews are tagged too (English is the most accurate).

Example output

{
  "store": "appstore",
  "appId": "324684580",
  "country": "us",
  "reviewId": "appstore:14560912345",
  "date": "2026-09-17T16:18:01+00:00",
  "rating": 1,
  "version": "9.1.82",
  "title": "Unfortunately",
  "text": "I have over 2,000 songs downloaded. The playlist lags nonstop and plays the same songs...",
  "topic": "bug",
  "topicConfidence": 0.81,
  "severity": 2.5,
  "churnRisk": 0.93,
  "actionable": 0.88,
  "mentionsCompetitor": 0.04,
  "labels": { "mentionsOffline": 0.91 }
}

Custom labels

Ask anything that has a yes/no answer, in plain English:

"customLabels": [
  { "name": "mentionsDarkMode", "question": "Does the review mention dark mode or the color theme?" },
  { "name": "afterUpdate", "question": "Does the reviewer say the problem started after a recent update?" },
  { "name": "refundRequest", "question": "Is the reviewer asking for a refund?" }
]

Each label comes back as a probability under labels. Up to 20 labels per run.

Input

• Apps – App Store ids or URLs (324684580, https://apps.apple.com/us/app/spotify/id324684580) and Google Play package names or URLs (com.spotify.music). Mix both stores in one run.
• Max reviews per app – newest first. Apple's public feed serves at most 500 reviews per country; Google Play serves more.
• Countries – two-letter codes; each country has its own reviews.
• Only reviews since – skip older reviews. Combine with an Apify schedule for daily or weekly monitoring that only pays for new reviews.
• Custom yes/no labels – see above.

Pricing

Pay per result, nothing else:

• $1.50 per 1,000 tagged reviews (all built-in tags included)
• $0.20 per 1,000 reviews for each custom label

Reviews that could not be tagged are dropped and never billed. The run stops cleanly when it reaches the maximum charge you set.

What people use it for

• Product managers – a weekly digest of the highest-severity, most actionable reviews instead of reading everything.
• ASO and app marketing agencies – topic breakdowns per client app and per competitor, per country.
• Support and CX teams – route account, pricing and refund reviews to the right queue; reply first where churnRisk is high.
• Competitive research – run it on a competitor's app and see what their users complain about and which rivals they mention.
• AI agents and dashboards – typed, numeric fields that drop straight into a sheet, a BI tool, or an agent's context without another LLM pass.

FAQ

Do I need an App Store Connect or Google Play Console account? No. Only public reviews are read; it works for any app, including competitors'.

Are reviewer names included? No. Only the review content, rating, version and date are returned.

How fresh is the data? Reviews come straight from the stores at run time, newest first.

Something looks mis-tagged. Open an issue with the reviewId. Tag definitions are tuned continuously.