App Store & Google Play Reviews — Only New Since Last Run

App Store & Google Play Reviews — Only New Since Last Run

Get only the new App Store and Google Play reviews since your last run — both stores in one job. No reviewer names (PII stripped). Export to CSV/JSON, send to Sheets, or pull via API. From about $0.10 per 1,000 reviews (see exact events below).

• What you get

  • Newest customer reviews from both stores in a single dataset
  • Only-new deduplication across scheduled runs (mode: new_since_last_run)
  • Per-review fields: store, appTitle, country, stars, title, text, appVersion, reviewedAt, developer reply info, thumbsUp (Google)
  • Per-app summaries in key‑value store: newReviews, avgStarsNew, share1to2StarsNew, unansweredNew, latestReviewAt, currentRating, ratingsCount
  • No reviewer names or profile links — PII is not emitted

• Who it’s for

  • Mobile teams and support leads routing fresh 1–2★ reviews to Slack
  • ASO/review agencies monitoring multiple apps across countries
  • Product managers tracking rating changes per release/country

Input prefill (example)

This is the default example in the UI (Spotify + Calm, both stores, US/GB, daily new reviews):

{
  "apps": [
    "apple:324684580",
    "google:com.spotify.music",
    "apple:571800810",
    "google:com.calm.android"
  ],
  "countries": ["us", "gb"],
  "language": "en",
  "mode": "new_since_last_run",
  "maxReviewsPerApp": 200
}

Comparison (what this Actor optimizes for)

Capability	This Actor	Typical single‑store scraper
Both stores in one run	✅	❌
Only-new dedupe for schedules	✅	❓ (often missing)
Reviewer names/PII	🚫 never outputs	⚠️ often included
Price signal	~$0.10 / 1,000 reviews	varies

Example output (no personal data)

store	appTitle	country	stars	title	reviewedAt	developerReplied
apple	Spotify: Music and Podcasts	us	5	Fonte de pesquisa	2026-10-31T20:53:23.000Z	false
google	Calm - Sleep, Meditate, Relax	us	2		2026-11-01T02:58:44.250Z	false

Pricing (pay per event)

Exact events are defined in .actor/pay_per_event.json:

• Review: $0.0001 per review row (primary)
• App checked: $0.002 per app × country inspected (charged even when no new reviews)
• Actor start: $0.00005 per run start (1 event at 256 MB)

Rule of thumb: ~$0.10 per 1,000 reviews returned.

Recipes

• Daily schedule (only-new)

  • Set mode: "new_since_last_run", keep stateKey empty to auto-derive, and add a daily schedule in Apify Console.
  • Results: each run emits only reviews not seen before for each app × country.

• 1–2★ alerts to Slack (Zapier/Make/n8n)

  • Add a filter on stars <= 2 in your automation tool.
  • Trigger from new dataset items; send message text with store, appTitle, country, stars, title, text, reviewedAt, and reviewUrl (often null by design).

Reliability and behavior

• Uses official libraries with retries: app-store-scraper and google-play-scraper via p-retry (3 retries, backoff), concurrency capped at 2.
• Apple pagination limit: newest ~10 pages (~500 reviews) per app × country; older reviews cannot be paged beyond what Apple exposes.
• State & dedupe: per app × country, bounded ID window; new_since_last_run emits only unseen reviews and advances watermarks.
• Error isolation: per-target failures are logged and do not stop other targets.
• PII protection: reviewer identifiers are never emitted; defensive deletion of userName, userImage, author, profileUrl before write.
• Empty-new edge case: if no reviews are emitted for any target in a run, the Actor exits with an error saying “All apps failed.” This is intentional to surface misconfiguration; a run with at least one emitted review succeeds.

FAQ

• How do I target an app? apps accepts Apple numeric IDs or Google package names (or full store URLs). Prefix with apple: / google: to disambiguate.
• Multiple countries? Yes — use countries: ["us","gb",...]. Apple and Google results are storefront‑scoped.
• Filter by rating? Use minStars / maxStars (1–5). Example: set maxStars: 2 for negative‑review alerts.
• Do you return reviewer names? No — this Actor never outputs reviewer names or profile links.
• Is there a dataset when nothing is new? Summaries are written to the key‑value store; if nothing new is emitted across all targets, the run ends with an error (see edge case above).