Google Maps Lead Generation & Contact Scraper

Build B2B outreach lists from Google Maps. Search any category in any city and get the business name, category, address, phone, website, opening hours and GPS coordinates — then automatically enrich each lead with the email addresses and social accounts found on its own website.

No login, no Google API key, no browser. Pure HTTP.

---

What you get per lead

Field	Notes
name, categories, primaryCategory	Business name and every category Google assigns
phone, phoneE164, phoneLocal	phoneE164 is dial-ready (+61282967351)
website, websiteDomain	As listed on Google
emails	Found on the business website — Google never exposes these
socialProfiles	Instagram, Facebook, LinkedIn, YouTube, TikTok, WhatsApp, Telegram, X
address, addressParts, region	Single-line plus split components
latitude, longitude	Real GPS coordinates
openingHours	Opening hours plus openNow and openingHoursComplete — see the note below
rating, reviewsCount, reviewsUrl	Social-proof signals for prioritising outreach
placeId, featureId, googleMapsUrl	Stable identifiers for de-duplication
timezone	IANA zone — useful for deciding when to call

---

Quick start

{
  "searchQueries": ["dentist", "dental clinic"],
  "areaName": "jakarta",
  "maxResultsPerQuery": 200,
  "enrichFromWebsite": true,
  "requireEmail": false,
  "exportFormats": ["csv", "excel"]
}

Targeting a location

A search query always needs a location, otherwise Google picks one from the exit IP and results are not reproducible. Three ways, in increasing precision:

Input	Use when
areaName: "jakarta"	Quickest. Built-in cities and countries.
centerLatitude + centerLongitude + radiusKm	"Everything within 5 km of this point."
boundingBox: [south, west, north, east]	Exact area, anywhere on earth.

Areas are tiled into multiple search points for coverage, then every result is filtered against your exact area using the lead's own GPS coordinates — so a 3 km radius really means 3 km. Verified: a 0.6 km radius returned 20 leads from 40 found, all within 0.489 km.

tileSizeKm controls thoroughness: smaller tiles find more businesses and cost more requests. Left empty, it is chosen from the size of your area.

Scraping specific businesses

Skip search entirely and pass placeUrls (any Google Maps URL, including maps.app.goo.gl short links) or placeIds (a Place ID, a numeric CID, or a raw feature ID). You can combine this with searching in the same run.

A note on opening hours and review counts

Google serves two different richnesses of record and picks between them by client. The Apify platform usually receives the reduced one, which has no review count and only today's opening hours. Measured on the platform:

With fetchPlaceDetails	Result
off	reviewsCount 0/25, hours 1 day
on	reviewsCount 15/15, reviewsUrl 15/15, descriptions 13/15, hours still 1 day for 14/15

So fetchPlaceDetails: true reliably recovers review counts, review URLs and descriptions at one extra request per lead. The full 7-day schedule often cannot be recovered from a datacenter IP at all — that is Google's behaviour, not a parsing gap, and every lead carries openingHoursComplete so you can tell which schedules are partial. Running from a non-datacenter IP returns the full week; a residential proxy is not a workaround here (see Anti-bot and proxy).

The run logs a warning when it detects reduced results and the option is off.

Narrowing the list

requirePhone, requireWebsite, requireEmail and minRating drop leads that are not worth an outreach slot. requireEmail is applied after enrichment, so it needs enrichFromWebsite switched on.

---

How enrichment works

For each lead with a website: fetch the homepage, extract emails, social handles and tel: numbers. If the homepage has no email, follow one same-domain contact/about link and try there. One hop — this is enrichment, not a crawl.

Two details that separate usable output from noise:

• Share widgets are filtered out. Almost every site embeds a Facebook share button, so a naive scraper reports that every business has a Facebook page called sharer. Known widget paths are rejected.
• Failures are recorded, not thrown. A dead domain, expired certificate or 403 lands in enrichmentStatus (ok, ok_contact_page, no_website, http_403, fetch_failed:…) and the lead still reaches your dataset with its Maps data intact.

Typical yield on a 40-lead run: 24 had a website, 17 gave social accounts, 12 gave an email. Most businesses simply do not publish an email — requireEmail lets you keep only the ones that do.

---

Anti-bot and proxy

Reconnaissance found no WAF on these endpoints: ten TLS fingerprints all returned 200 from a plain connection.

Do not use residential proxy here. This is the opposite of the usual advice, and it was measured rather than assumed:

Egress	Result
No proxy (the default)	works
Apify RESIDENTIAL	fails — Google answers with a consent/CAPTCHA interstitial

Google polices residential pools far harder than datacenter ranges because those IPs are widely abused. With no WAF to defeat, the residential IP buys nothing and costs you the run. The proxy is off by default; enable one only to spread load at high volume, and leave the group list empty so Apify picks a group your plan actually has — naming a group your account lacks (a free plan has no DATACENTER) fails the run before it starts.

The client impersonates Chrome 150 via curl_cffi, warms the Maps cookie jar once per session, retries with exponential backoff, and rotates TLS profile on 403/429/503.

---

Limits worth knowing

1. Results per query are finite. Google stops serving pages well before the true number of businesses in a large area. Coverage comes from tiling (more search points), not from paging one search forever. Lower tileSizeKm to dig deeper into a dense city.
2. Most businesses have no email anywhere. Enrichment can only find what the website publishes.
3. Plus codes are not present in Google's search records and are emitted as null rather than guessed.
4. Category names and opening-hours wording follow language. Set language and region to the market you are targeting.
5. reviewsCount is Google's lifetime total, not a count of anything this actor scraped.

---

Output shape

Records follow the portfolio envelope — _input, _source, _scrapedAt, recordType — with _error / _errorDetail on failures. Split a run by recordType: LEAD or ERROR. An input that fails to resolve produces an ERROR row rather than vanishing, so inputs always reconcile against outputs.

Exports (exportFormats) additionally write leads.csv, leads.xlsx, leads.json and leads.ndjson to the key-value store. CSV and Excel columns are ordered for outreach — name, category, phone, email, website first — and nested values are flattened (socialProfiles.instagram).

---

Development

pip install -r requirements.txt
python test_errors.py     # offline: parsing, geo, enrichment, exporters
python test_local.py      # live end-to-end run

test_local.py reads _input.json when present, otherwise uses its built-in default input. See CRAWLING_METHOD.md for the endpoint reconnaissance and the two parsing traps that produce silently wrong data.