Google Maps, TripAdvisor, Booking, App Store & Google Play Reviews to Fix List + Reply Drafts

Ran a review scraper for Google Maps, TripAdvisor, Booking.com, the App Store or Google Play? Pick its dataset here and get back what your customers keep complaining about, ranked by how much it costs you in stars. For each place or app you get:

• What to fix first: the top complaint themes, each with a concrete action and the evidence behind it
• Complaint and praise themes: how many reviews mention each one, their average stars, and whether the theme is rising or falling compared with the previous period
• Example quotes, copied word for word from the reviews
• Escalate now: reviews that mention food safety, injury, security, discrimination or legal threats (for apps: security or privacy problems, data loss, wrong charges, child safety)
• Reply drafts (optional): a polite reply to each review that mentions the specific points the customer raised. The Actor never posts anything. You read the drafts and post the ones you want yourself.

Output is one row per review and one summary row per place or app, plus a Markdown report and CSV files in the key-value store.

Supported scrapers

The source is detected item by item, so you don't need to say which scraper you ran.

Scraper	Report	What is read
Google Maps Reviews Scraper	place report	text, stars, date, place name and ID, owner reply yes/no
Tripadvisor Reviews Scraper	place report	title, text, rating, date, language, place name, ID and address, owner reply yes/no
Booking Reviews Scraper	place report	title, liked and disliked text, score (1–10, mapped to 1–5 stars: 1 → 1, 10 → 5), date, language, hotel ID, property reply yes/no. The hotel name is taken from the hotel's Booking.com link.
App Store Reviews Scraper	app report	title, text, stars, date, app ID, app version. The app name, category and overall rating are looked up in the public App Store listing.
Google Play Reviews Scraper	app report	text, stars, date, language, app ID, app version, developer reply yes/no. The app name is looked up in the public Google Play listing.

TripAdvisor and Booking.com datasets are mapped from the fields those scrapers publish and have been tested less than Google Maps, App Store and Google Play. If a field comes out empty or wrong, open an issue on the Actor page.

App reports are written for a product team: themes are product feedback (crashes, bugs, login and account, missing features, pricing and subscription, ads, UX, performance, customer support), the fix list suggests product, engineering or support actions, and it names the app versions that get the most 1–3 star reviews. If an app name can't be looked up, the app ID is used as the name.

How to use it

1. Run one of the scrapers above for your place(s) or app(s). If the scraper has a "Personal data" or "Scrape reviewer info" option, you can turn it off, because this Actor doesn't use reviewer data.
2. Open this Actor and pick that run's dataset in Reviews dataset.
3. Pick who should get reply drafts, add an optional sign-off, and start the run.

There are two other ways to pass reviews:

• reviewsUrl: a public link to a JSON, JSON Lines or CSV file.
• items: a JSON array pasted into the form.

Exports of the scrapers above work as is. For other files, columns are matched by common names: text/review/content, stars/rating/score (4, 4.2/5, 8/10 and ★★★★ all work), publishedAtDate/date, title/placeName/place (the place name), reviewTitle (the review headline), appId (makes it an app review) and reviewId/id.

Input

Field	Default	Description
datasetId	–	Dataset of a run of one of the supported scrapers, chosen with the picker. Read-only access.
reviewsUrl	–	Instead of a dataset: a public JSON or CSV file, up to 50 MB.
items	sample	Instead of a dataset or file: reviews as a JSON array.
maxReviews	1000	Maximum number of reviews with text analysed (and charged) per run, from 1 to 5,000. The newest reviews of each place or app are taken first.
periodDays	30	Length of the trend window: the last N days are compared with the N days before. The window ends at the newest review of each place or app.
themeTaxonomy	–	Theme names from an earlier run (the taxonomy field of a place or app row). New reviews are grouped into these names, so you can compare runs one to one.
allowNewThemes	true	With themeTaxonomy: whether new themes may be added.
businessContext	–	One or two sentences about the business or app, such as opening hours or a support email. Used in the analysis and in the replies.
replyTo	unanswered	none, negative (3 stars or less, no owner or developer reply yet), unanswered (no reply yet) or all. App Store exports don't show developer replies, so every App Store review counts as unanswered.
replyLanguage	auto	auto replies in the language of each review. Or name a language.
replySignature	–	Sign-off line added to every reply.
defaultPlaceName	My business	Place name used for rows that don't have one. Not used for app reviews.

{
  "datasetId": "<your scraper run's dataset>",
  "replyTo": "negative",
  "replySignature": "— The Harbor Street Cafe team",
  "businessContext": "Independent cafe, open 7:00-16:00. Contact: hello@harborstreet.example"
}

Output

Place row (rowType: "place-summary", trimmed):

{
  "rowType": "place-summary",
  "placeName": "Harbor Street Cafe",
  "reviewsAnalyzed": 10,
  "avgStars": 3.3,
  "currentPeriod": { "from": "2026-09-21", "to": "2026-10-21", "reviews": 7, "avgStars": 3.1, "negativeSharePct": 57.1 },
  "previousPeriod": { "from": "2026-08-22", "to": "2026-09-21", "reviews": 4, "avgStars": 3.5, "negativeSharePct": 50 },
  "summary": "Average rating fell this period, mainly because of long wait times. Customers consistently praise the friendly staff and the baked goods.",
  "fixFirst": [
    { "rank": 1, "theme": "Wait Times", "action": "Add a second barista during peak hours and set up a queue system.", "why": "3 mentions, avg 2.0 stars, new this period.", "mentions": 3, "avgStars": 2, "priorityScore": 9, "trend": "new" }
  ],
  "complaints": [
    { "theme": "Wait Times", "mentions": 3, "sharePct": 30, "avgStars": 2, "priorityScore": 9, "trend": "new",
      "exampleQuotes": ["Queue out the door and only one person at the till. 25 minutes for a latte is too long."] }
  ],
  "praise": [{ "theme": "Staff Friendliness", "mentions": 3, "...": "..." }],
  "escalations": [{ "reviewId": "s09", "publishedAt": "2026-09-02", "stars": 1, "reason": "foreign object in food" }],
  "unansweredNegative": 6,
  "taxonomy": { "issues": ["Wait Times", "Food Quality", "Cleanliness"], "praise": ["Staff Friendliness", "..."] }
}

App row (rowType: "app-summary", reportType: "app", trimmed): the same fields as a place row, with product feedback themes.

{
  "rowType": "app-summary",
  "reportType": "app",
  "source": "app-store",
  "placeName": "Duolingo: Learn Languages",
  "placeId": "570060128",
  "placeCategory": "Education",
  "reviewsAnalyzed": 10,
  "avgStars": 4,
  "fixFirst": [
    { "rank": 1, "theme": "Missing Features", "action": "Have the product team investigate and document user requests for missing features, prioritizing based on user impact.", "mentions": 1, "avgStars": 1, "trend": "n/a" }
  ],
  "taxonomy": { "issues": ["Missing Features", "Customer Support", "Pricing", "AI Errors", "Gamification"], "praise": ["Language Variety", "Learning Effectiveness"] }
}

Review row (rowType: "review", trimmed):

{
  "rowType": "review",
  "source": "google-maps",
  "placeName": "Harbor Street Cafe",
  "reviewId": "s04",
  "publishedAt": "2026-10-09T00:00:00.000Z",
  "stars": 2,
  "excerpt": "Queue out the door and only one person at the till. 25 minutes for a latte is too long.",
  "status": "analyzed",
  "sentiment": "negative",
  "severity": "medium",
  "issueThemes": ["Wait Times"],
  "issues": [{ "label": "long wait", "theme": "Wait Times", "severity": "medium", "quote": "25 minutes for a latte is too long." }],
  "praiseThemes": [],
  "needsEscalation": false,
  "hasOwnerResponse": false,
  "replyDraft": "Thank you for your feedback. We sincerely apologise for the long wait... — The Harbor Street Cafe team",
  "replyFlags": [],
  "charged": true
}

The key-value store also holds:

• REPORT.md: the full report for all places and apps
• fix-list.csv: one row per complaint theme and place or app, with rank, mentions, average stars, priority, trend, action and quotes
• reviews.csv: one row per review, with themes, sentiment and the reply draft
• OUTPUT: run summary with links to the files

How themes are built. Each review is tagged with short topic labels. Labels from all reviews of a place or app are then grouped into a small, stable set of themes for it: at most 12 complaint themes and 8 praise themes. Priority is the number of mentions × star impact, where each mention scores the stars it lost compared with a 5-star visit, with a minimum of 0.5. When a review has no star rating, the issue's severity is used instead. Trend compares the share of reviews that mention a theme in the last period with the period before. It is shown as n/a when either period has fewer than 3 reviews.

Privacy

This tool works at business and app level only. It does not use or output reviewer names, usernames, reviewer IDs, avatars, profile links, photos, reviewer locations or the owner's earlier replies. A review row keeps only the review ID, date, stars, app version, the review title and a 240-character excerpt, which is enough to find the review when you reply. Names found in the text are replaced: in quotes and excerpts, staff names are replaced with [staff member] and other names with [name]; themes, the fix list and summaries refer to staff by role ("the cashier", "a staff member"); reply drafts don't name the reviewer or staff ("our team member"). Your business or app name, address and sign-off are kept as is, even when they contain a first name ("Joe's Pizza"). Name detection is automatic, works best for English reviews and can miss a name, so read quotes and drafts before you share them.

Reply drafts

• Polite and specific, 40–90 words (apps: 30–80 words), written in the review's language unless you choose one.
• Store limits: Google Play developer replies can't be longer than 350 characters, so Google Play drafts are written short, sign-off included. A longer draft is cut at the last full sentence that fits (flag reply_shortened); if too little would be left, it is dropped, flagged reply_dropped_too_long, and not charged. App Store drafts stay far below Apple's 5,970-character limit.
• App replies ask the user to contact support with their device and app version, and never promise a fix date or a new feature.
• Drafts never promise refunds, discounts, vouchers, free items or other compensation, and never admit liability for health or safety claims. If a draft still mentions compensation, it is dropped, flagged reply_dropped_compensation, and not charged.
• A draft with a leftover placeholder such as [date] is flagged has_placeholder.
• The Actor does not post replies. Read every draft before you post it.

Pricing

Pay per event. Reviews or reports that fail are free. Rating-only reviews (no text) are counted in the stats for free.

Event	Price
Review analysed (themes, sentiment, severity)	$0.003
Reply draft written	$0.002
Place or app report (taxonomy, ranking, trend, fix list)	$0.02 per place or app

Example: 200 reviews of one place with replies to 60 unanswered negative reviews cost 200 × $0.003 + 60 × $0.002 + $0.02 = $0.74.

Before it starts, the Actor works out the most the run can cost. If that is above the run's maximum cost, it stops without charging and tells you the amount to set. Available on paid Apify plans.

Limits

• Up to 5,000 reviews with text per run (maxReviews), from up to 100 places and apps. A dataset is read up to its first 20,000 items.
• Each review is analysed up to its first 4,000 characters (title included). businessContext is capped at 600 characters.
• Up to 400 distinct topic labels per place or app go into theme grouping. Rarer labels join the closest theme, or "Other issues" / "Other praise".
• Themes, actions and quotes are AI-generated from the review text. Quotes are checked word for word against the review, and a quote that doesn't match is removed. Small samples (a few reviews per place or app) give rough themes. Treat the ranking as a guide, not a measurement.
• A file link must be downloadable without a login. Private Apify dataset links don't work as reviewsUrl: pick the dataset in Reviews dataset instead.

Good to know

• On the free Apify plan the Actor returns a short preview with an example result and prices instead of running.