App Release & Customer Reaction Monitor

Answer the product question that a raw app-review scraper cannot: did this release improve customer experience, or introduce a regression?

The Actor joins structured release records with dated review records. When at least three reviews carry the exact app version, version attribution wins; otherwise a configurable before/after date window is used.

Each release report includes

• review counts before and after
• average rating before, after, and the change
• sentiment direction
• emerging, resolved, and recurring complaint categories
• regression signals and issue-spike flag
• IMPROVED, REGRESSED, MIXED, STABLE, or INSUFFICIENT_DATA conclusion

Composition-first input

Paste records or provide Apify dataset IDs. This Actor intentionally does not scrape Apple App Store or Google Play itself, so it can work with different upstream collectors and does not depend on brittle authenticated/private endpoints.

Quick start

{
  "appName": "Acme Mobile",
  "releases": [{ "version": "2.0.0", "releaseDate": "2026-08-01", "releaseNotes": "New sync engine" }],
  "reviews": [
    { "text": "Stable and fast.", "rating": 5, "date": "2026-07-28", "version": "1.9.0" },
    { "text": "Crashes after the update.", "rating": 1, "date": "2026-08-03", "version": "2.0.0" }
  ]
}

Pricing

The intended pricing is pay per release analyzed, with platform usage included. The run summary is free. Exact current pricing appears on the Store page and charge limits are respected.

Limitations

At least three reviews on both sides are required for a directional conclusion. Version labels and review dates come from upstream data and are not independently verified. Keyword-based issue categories are transparent but may miss synonyms, sarcasm, and non-English feedback. Correlation around a release is not proof that the release caused the reaction. Always inspect underlying reviews before shipping a rollback or product decision.

Reviewer identity is ignored; no personal-data profiling, outreach, private scraping, credentials, or CAPTCHA bypass are used.