Google Maps Fresh Reviews Scraper

Apify Actor that extracts only the newest reviews since the last run for a given Google Maps business, sorted chronologically. Built for reputation monitoring — call it on a schedule and it returns just the delta each time.

How it works

1. Opens the place URL, opens the Reviews panel, sets sort order to Newest.
2. Scrolls and reads review cards one batch at a time.
3. Stops as soon as it reaches a review it already saw on a previous run (tracked per-business in Apify's Key-Value Store) — it does not re-scrape the whole review history every time.
4. Pushes only the new reviews to the Dataset, and updates the "last seen" cutoff for next time.

⚠️ What has and hasn't been tested

This was built and unit-tested in a sandboxed environment with no network access to Google (only npm/GitHub/PyPI-type registries are reachable there). So:

• ✅ Syntax-checked, dependencies install cleanly, and the custom logic (relative-date parsing, CID/URL key extraction) is covered by test/utils.test.js — run node test/utils.test.js yourself to see it pass.
• ❌ The actual browser-scraping flow (opening Maps, clicking Sort, reading review cards) has NOT been run against a live page. It's written against Google's currently-documented DOM structure, but this is the part most likely to need a tweak on first real run.

Do not skip the first test run — deploy to Apify, run it against one business you control or don't mind testing on, and check the Dataset output before you point it at real customers.

Deploying to Apify

1. Push this folder to a GitHub repo.
2. In the Apify Console: Actors → Create new → Link Git repository, point it at your repo.
3. Apify will build the Docker image automatically (it reads .actor/actor.json and the Dockerfile).
4. Set input, e.g.:

   {
     "startUrls": [{ "url": "https://www.google.com/maps/place/.../@lat,lng,17z/data=...!4m5!3m4!1s0x...:0x...!8m2!3d...!4d..." }],
     "maxReviews": 100
   }
   

   Use the full Maps URL for the business (right-click the pin → "What's here?" or just copy the URL from your browser address bar when viewing the place).
5. Run it. Check the Dataset tab for output, and the Key-Value Store tab (fresh-reviews-state) to see the per-business cutoff being tracked.
6. Run it again a day or two later — on a business with no new reviews, the Dataset should come back empty. That's the actual "fresh reviews only" behavior working.

When Google changes their DOM (selector maintenance)

Google's class names are obfuscated and do drift over time. If a run starts returning empty/null fields where it used to return data, here's the fastest way to fix it:

1. Open the business's Maps page in a normal Chrome browser, open Reviews.
2. Right-click a review's author name → Inspect.
3. Find the new class name and swap it into src/main.js — the relevant lines are clearly commented:

   const authorName = await card.locator('div.d4r55')...       // author name
   const ratingLabel = await card.locator('span[role="img"]')... // star rating
   const relativeDate = await card.locator('span.rsqaWe')...    // "3 weeks ago"
   const reviewText = await card.locator('span.wiI7pd')...      // review body
   const ownerResponse = await card.locator('div.CDe7pd div.wiI7pd')... // owner reply
   

4. Push the fix, redeploy. Consider adding a scheduled daily smoke-test run on one known business so you catch breakage same-day.

Files

.actor/actor.json         Apify actor manifest
.actor/input_schema.json  Input form shown in Apify Console
Dockerfile                Apify Playwright+Chrome base image
package.json              Dependencies (apify, crawlee, playwright)
src/main.js               Main scraping logic
src/utils.js               Date parsing + place-key extraction (unit-tested)
src/webhook.js             Low-rating alert webhook sender (unit-tested)
test/utils.test.js        Standalone tests for the logic above
test/webhook.test.js      Tests the webhook sender against a real local HTTP server

Low-rating alert webhook

Set webhookUrl in the input to get a POST request whenever new reviews at or below lowRatingThreshold (default 3★) show up. This is the actual reputation-monitoring differentiator -- without it, this is "just" a scraper.

• Slack: paste a Slack Incoming Webhook URL directly into webhookUrl. The payload includes a text field Slack reads automatically.
• Zapier / Make.com / your own endpoint: point webhookUrl at your trigger URL. The full payload (place info, threshold, and the complete review objects) is included as JSON alongside text.
• Only fires when there's at least one matching review in that run, and a failed webhook delivery is logged as a warning -- it never fails the scrape itself.

Suggested next features (not built yet)

• Owner-response-missing flag (reviews with no reply after N days).
• Scheduled runs via Apify's built-in Scheduler (no code needed -- just point a Schedule at this actor with your business list as input).