Google Business Profile Review Scraper

This Actor finds recent low-star reviews on Google Business Profiles across a search term and a location, and puts the business's contact details on every one of them. Give it plumber and Austin, TX and you get back each 1-star review left in the last six months — the review text, its date and its author, alongside the phone number, website and postal address of the business it was left on.

What does Google Business Profile Review Scraper do?

The rating filter and the recency window are both yours to set, so the same Actor answers "every 1-star this month", "everything at 2 stars or below this year", or simply "every review these profiles have".

It reads Google's own JSON endpoints in two stages: a search surface that finds the profiles and carries their contact details, then each profile's reviews, sorted worst-first so one request covers the whole low-star history rather than the most recent page.

• One row per review, carrying the full business record, so any single row is enough to act on.
• maxReviewRating and withinDays are filters and budget dials at once — a profile whose reviews all miss them contributes no rows.
• Specific profiles can be monitored directly by URL instead of searched for, via places.
• Up to 50 search terms per run, each searched separately.

No Google account, API key or cookies are used — this reads what a logged-out visitor sees.

What data does it extract?

One row per review. A business with three matching reviews returns three rows, each carrying the full business record so any single row is enough to act on.

Field	Type	Notes
rating	number	The review's stars, 1–5
publishedAt	string	ISO timestamp — an exact date, not "3 months ago"
relativeDate	string	Google's own wording, e.g. "3 days ago"
text	string	The review body, with the reviewer's line breaks intact. null for a rating-only review
authorName	string	The display name the reviewer published with the review
authorProfileUrl	string	The reviewer's public Google Maps contributor page
authorReviewCount	number	How many reviews the author has written overall — a credibility signal. 0 is a real value
language	string	BCP-47 code for the language the review was written in, e.g. en
imageCount	number	Photos attached to the review
reviewUrl	string	Link straight to the review
reviewId	string	Stable, use it to de-duplicate across runs
businessName	string	The business as Google lists it
businessRating	number	The profile's overall star average, e.g. 4.7
phone	string	Formatted phone number, e.g. (512) 309-1487
website	string	The business's own site, as Google links it
address	string	Street address; null for a service-area business, which Google also shows without one
categories	string[]	Google's own labels, e.g. ["Plumber", "Electrician"]
latitude, longitude	number	Exact coordinates, ready to map
matchedReviewCount	number	How many matching reviews this business had in this run — group or prioritise on it
placeId	string	Google's ChIJ… Place ID
dataId	string	The 0x…:0x… id Google keys the profile's reviews on
mapsUrl	string	Link straight to the profile
lowestRatedReviewsRead	number	How much of the profile's low-star history this run saw (see Reliability)
searchTerm, sourceUrl, collectedAt	string	Provenance for every row

Why scrape Google reviews?

• Reputation-management prospecting. A fresh 1-star review is a business in the market for help, and the row already carries the phone number to call.
• Agency churn signals. Watch your own clients' profiles and get the bad review the day it lands, rather than when they forward it.
• Competitor monitoring across a category and city, with the complaint text itself rather than a shifted star average.
• Complaint-theme analysis across a whole market segment, read from the review bodies.
• Franchise and multi-location oversight, one row per piece of bad feedback, grouped on dataId.

How to use it

1. Open the Actor and click Try for free, or add it to your Apify account.
2. Put your terms in Searches, one per line — plumber, hvac. Set Location to the market, e.g. Austin, TX, or leave it empty if your terms already name one.
3. Leave Maximum review rating at 1 for 1-star only, or raise it to 2 to catch 1- and 2-star.
4. Leave Within days at 180 for the last six months, or shorten it to watch a current window.
5. Click Start, then take the rows from the Dataset tab as JSON, CSV or Excel.
6. Read the run's SUMMARY record for any profile that could not be read, with its error.

To monitor a fixed set of businesses rather than a market, put their Google Maps URLs in Places and leave Searches empty. Save the input as a task and schedule it to get each bad review as it lands; rows carry collectedAt, and reviewId is stable, so successive runs de-duplicate cleanly.

Input

{
  "searches": ["plumber", "hvac"],
  "location": "Austin, TX",
  "maxReviewRating": 1,
  "withinDays": 180,
  "maxItems": 1000
}

Field	Description
searches	Search terms, e.g. "plumber". Up to 50, each searched separately.
location	City, state, ZIP or country, appended to every term. Leave empty if your terms already carry a location.
places	Optional. Google Maps place or review URLs to read directly, instead of or as well as a search.
maxReviewRating	Returns any review at or below this rating. 1 finds 1-star only; 5 returns every review. Default 1.
withinDays	How recent a review has to be to be returned. Default 180, roughly six months.
maxProfilesPerSearch	How deep to page each search before filtering. Default and maximum 120.
countryCode, language	The market the search runs in, e.g. US / en, or GB / en.
maxItems	Cap on dataset items. Default 1000. Free plans stop at 250 requests and 250 results — see below.
concurrency	Requests in flight at once. Default 3, maximum 5.

Supply either searches or places — a run with neither is rejected before it starts. A business whose reviews all miss your filter simply contributes no rows.

Input by URL

places accepts a Google Maps place or review URL — the long kind containing !1s0x…:0x… — or that 0x…:0x… id on its own. A maps.app.goo.gl short link and a bare ChIJ… Place ID do not carry that id, and are reported in failures rather than guessed at. A profile supplied this way returns its reviews and whatever name its URL contained; phone, website, address, categories and the coordinates come from the search surface, so those fields are null on a URL-only run.

Output

One row per review. The shape below is illustrative — it shows the fields and their types, not a captured run.

{
  "reviewId": "ChZDSUhNMG9nS0VJQ0FnSUMxbHZHWVh3EAE",
  "rating": 1,
  "publishedAt": "2026-07-28T14:02:11.000Z",
  "relativeDate": "3 weeks ago",
  "text": "Booked a same-day callout and nobody turned up.\nNo call, no refund.",
  "language": "en",
  "imageCount": 2,
  "authorName": "Dana Whitfield",
  "authorProfileUrl": "https://www.google.com/maps/contrib/108273645190273645190",
  "authorReviewCount": 37,
  "reviewUrl": "https://www.google.com/maps/reviews/data=!4m6!14m5!1m4!2m3!1sChZDSUhNMG9nS0VJQ0FnSUMxbHZHWVh3",
  "businessName": "Lone Star Plumbing Co",
  "businessRating": 4.1,
  "phone": "(512) 309-1487",
  "website": "https://lonestarplumbing.example.com/",
  "address": "1200 E 6th St, Austin, TX 78702",
  "categories": ["Plumber", "Drainage service"],
  "latitude": 30.267153,
  "longitude": -97.731339,
  "placeId": "ChIJLwPMoJm1RIYRetVp1EtGm10",
  "dataId": "0x8644b59a80cc032f:0x5d9b464bd469d57a",
  "mapsUrl": "https://www.google.com/maps/place/?q=place_id:ChIJLwPMoJm1RIYRetVp1EtGm10",
  "matchedReviewCount": 3,
  "lowestRatedReviewsRead": 22,
  "searchTerm": "plumber",
  "sourceUrl": "https://www.google.com/search?tbm=map&q=plumber+Austin%2C+TX",
  "collectedAt": "2026-08-21T09:14:03.512Z"
}

The run also writes a SUMMARY record holding the item count, the failure count and every profile that could not be read, with its error.

Reliability

Both halves read Google's own JSON endpoints rather than scraping rendered markup, so every value is exact: ratings are numbers, review dates are absolute timestamps, and the reviewer's own line breaks survive into text.

Reviews are read worst-first, which is why one request per profile is enough. Sorted that way, a single page carries the profile's entire low-star history rather than its most recent 50 reviews — on a busy profile that page held all 16 one-star and all 6 two-star reviews it had ever received, and the next page was already entirely five-star, while reading newest-first would have needed eight pages to cover the same six months. lowestRatedReviewsRead reports how many that page held; a value of 50 means it filled up, so a profile with more than 50 reviews at or below its floor may have others this run did not read.

A search pages up to Google's own ceiling of 120 results per term, and consecutive pages overlap, so the distinct total per term lands nearer 100. Profiles are de-duplicated on dataId before any reviews are fetched, so a profile that appears on two search pages is never fetched twice.

A profile whose reviews cannot be read is recorded in the run's SUMMARY under failures rather than silently contributing nothing — an intercepted response and a profile with genuinely no matching reviews look identical otherwise, and the difference decides whether a list is complete.

There are no retries by design: the String Unblocker owns proxy rotation and anti-bot solver selection, so a retry loop here would only re-roll the same block.

Limitations

Search and review data only: no opening hours, no menus, no photos, no popular-times data and no "about" text. Reviews are read one worst-first page deep, so this is not a full review-history export for a profile with more than 50 low-rated reviews. Profiles supplied by URL in places carry no address, phone, website or coordinates, because those come from the search surface rather than the review endpoint. Google orders a search by its own relevance rather than anything stable, so re-running can surface a business at a different rank; de-duplicate on reviewId.

How much does it cost?

A run that returns 1,000 rows costs $0.83 on a paid plan, or $1.00 with no subscription. You are charged per result row written to the dataset, so a run that returns nothing costs nothing, and maxItems is the budget dial.

Plan	Per 1,000 rows
Free	$1.00
Bronze	$0.83
Silver	$0.58
Gold	$0.42
Platinum	$0.31
Diamond	$0.19

Apify bills its own platform usage for the run on top of this.

One row is one review, so a business with three matching reviews costs three rows. A profile whose reviews all miss your filter contributes none and costs nothing, which makes maxReviewRating and withinDays budget dials as much as filters.

Runs started from an Apify free plan stop at 250 requests and 250 results, and the run reports that it reached the limit. Any paid plan runs the full input and maxItems you set.

The limit exists because this Actor fetches through our own infrastructure, which Apify does not cover for free-plan runs. It binds on requests as well as results, and one ceiling covers both the search and the review stage together, so a large input cannot spend those fetches for rows the run will not return.

Using it with the Apify API

import { ApifyClient } from 'apify-client';

const client = new ApifyClient({ token: 'YOUR_APIFY_TOKEN' });

const run = await client.actor('usestring/google-business-reviews').call({
    searches: ['plumber', 'hvac'],
    location: 'Austin, TX',
    maxReviewRating: 1,
    withinDays: 180,
    maxItems: 1000
});

const { items } = await client.dataset(run.defaultDatasetId).listItems();
const callable = items.filter((review) => review.phone !== null);
console.log(`${items.length} bad reviews, ${callable.length} with a phone number attached`);

The same call works from the Python client, and from any HTTP client through the Apify REST API. The run's SUMMARY record is in the run's default key-value store.

Is it legal to scrape Google?

This Actor reads public Google Business Profiles as a logged-out visitor. It sends no account credentials, no cookies and no API key, and it accesses nothing behind a login or a paywall. The rows it return are reviews their authors published publicly, alongside the contact details a business chose to list so that customers could find them.

It does return the display name a reviewer published with their review, which is personal data in the GDPR sense even though it is public. If you process these rows for people in the EU or UK, that is your lawful basis to establish, and authorName is the field to drop if you do not need it.

Scraping public data is broadly lawful in the US and the EU, but what you then do with it is governed by your own jurisdiction, your contracts and, where personal data is involved, by GDPR and similar regimes. Contacting the businesses you find is subject to your local marketing rules. This is not legal advice. If your use case is unusual, take your own.

Related actors

• Yelp Business Scraper — the same reputation surface on the other major review platform.
• Capterra Reviews Scraper — software reviews, where the complaint text is about a product rather than a visit.
• G2 Reviews Scraper — the B2B software equivalent.
• Realtor.com Listings Scraper — local business data of a different shape, for the same territory-planning work.

Pair this with a CRM import and the phone number on every row is a call list, already filtered to businesses with a fresh complaint against them.

FAQ

How do I find businesses with recent bad reviews? Run this Actor with your search terms and location and leave maxReviewRating at 1 and withinDays at 180. Every row is a 1-star review from the last six months, with the business it was left on attached.

Can I widen it beyond 1-star? Yes. maxReviewRating returns any review at or below the value you set, so 2 catches 1- and 2-star, and 5 returns everything the worst-first page held.

Does it return the review text? Yes, with the reviewer's line breaks intact. A rating-only review with no words returns null text and keeps its rating, date and author.

Why do I get several rows for the same business? Because each row is one review. Group on dataId, or read matchedReviewCount, which tells you how many that business had in this run.

How many businesses does one search cover? Up to 120, which is Google's own ceiling per search term; overlapping pages mean the distinct count is usually nearer 100. Use more search terms, or more specific ones, to cover a market more deeply.

Do I need a Google account, API key or cookies? No. It reads public Google Business Profiles as a logged-out visitor and never signs in.

Can I monitor specific businesses instead of searching? Yes — put their Google Maps URLs in places. See "Input by URL" for which URL forms carry the id this needs.

Can I search several cities in one run? location is one value applied to every term, so run the Actor once per city — or put each city into the search terms themselves and leave location empty.

Feedback

Something wrong in the data, or a field you need that is not here? Open an issue on the Actor's Store page and we will read it. Bug reports that name the search term, the location and the date of the run are the fastest to act on.