App Review Intelligence

Turn App Store and Google Play reviews into answers: what's breaking, what users want next, which release hurt your rating, and where competitors beat you. Alerts reach Slack before your rating drops.

Most review scrapers give you a spreadsheet of text. This Actor reads the reviews for you:

• Complaint topics: crashes & bugs, performance, login & OTP, payments (incl. M-Pesa), pricing & subscriptions, ads, UI/UX, customer support, update regressions, notifications, privacy & fraud, delivery, content quality. It covers English and common Swahili/Sheng.
• Feature requests, grouped ("dark mode", "offline mode", "export statements") with mention counts and the best example.
• Release impact: average rating and negative share per app version, the change vs the previous version, and regressions flagged automatically.
• Trends: weekly average rating, last N days vs before, and topic spikes ("crashes jumped from 3% to 41% of reviews").
• Unhappy-user phrases that appear far more often in 1–2★ reviews ("otp not received", "crashes on startup").
• Developer replies (Google Play): reply rate, reply rate on negative reviews, median reply time.
• Competitor comparison: health score, ratings, crash share, top complaint, top request, plus topic gaps where you get more complaints than competitors.
• Alerts: recent rating drop, store-rating drop, complaint spikes and release regressions, sent to Slack, Discord, Teams or any webhook.
• Optional AI insights with your own key (OpenAI, OpenRouter, Groq…): an executive summary, issues with evidence, likely causes and fixes, feature priorities and a reply template. Reviewer names are never sent.

Input

{
  "apps": ["https://play.google.com/store/apps/details?id=com.example.wallet&gl=KE", "https://apps.apple.com/ke/app/example-wallet/id1234567890"],
  "competitorApps": ["com.rival.wallet", "id987654321"],
  "appStoreCountries": ["ke", "us"],
  "daysBack": 90,
  "monitorName": "wallet-daily",
  "webhookUrl": "https://hooks.slack.com/services/…"
}

Accepted formats: store URLs, App Store ids (id389801252 or 389801252) and Play package names (com.whatsapp). Add a label with My App (iOS) = id123….

Output

• Dataset, view App summaries: one row per app with store rating, period and recent averages, negative %, health score (0–100), top complaints with examples and trends, feature requests, versions and regressions, weekly trend, reply stats, alerts and AI insights.
• Dataset, view Classified reviews: each review with rating, date, version, country, topics, sentiment, feature request, helpful votes, developer reply and is_new.
• REPORT: shareable HTML with alerts, a comparison table, per-app details and release impact.
• COMPARISON, FEATURE_REQUESTS, OUTPUT: JSON for dashboards and automations.

Scheduling a monitor

1. Run once to build the baseline. The first run sends no alerts.
2. Schedule daily or weekly with the same monitorName. Turn on Only new reviews in dataset if you push rows into a sheet.
3. Alerts fire on: the recent average dropping by ≥ 0.3★, the store rating falling, a complaint topic rising ≥ 10 points, or a new version rating much worse than the previous one.

You pay for each review only once per monitor. Re-analysis in later runs is free.

Good to know

• App Store: Apple's public feed exposes the ~500 most recent reviews per country. Add the countries where your users are.
• Google Play: reviews are filtered by language (playLanguage) and country. Large or frequent runs may need Apify Proxy.
• Sentiment follows the star rating. Text sentiment is used to flag mismatches (e.g. 5★ with an angry text).
• Topic detection is keyword-based: fast, free and explainable, but not perfect. Turn on AI insights for nuanced root causes.
• Reviewer names are excluded unless you turn them on.

Pricing

$0.50 per 1,000 new reviews analyzed + $0.02 per app report, + $0.01 per AI insight (optional). See PRICING.md.