Local Business Leads - places by category & area (no Maps API)

Get local business leads (restaurants, dentists, plumbers, salons, hotels, ...) for any city, radius or bounding box from the Overture Maps Places open dataset. No Google Maps API key, no scraping, no proxies, no browser: the actor reads the public Overture dataset directly, so runs are fast and cheap.

Each lead has: name, category, address, coordinates, website, business phone, social links, brand, confidence score and data sources.

Use cases

• Build prospecting lists for B2B sales ("dentists with a website in Berlin").
• Local SEO / market mapping: count competitors per category and area.
• Enrich CRMs or maps with open, redistributable place data.

Input

Field	Meaning
city	City name, geocoded with Open-Meteo (default Hanoi).
radiusKm	Half-width of the search square (default 2). Keep <= 15 for fast runs.
lat, lng	Alternative to city.
bbox	[minLng, minLat, maxLng, maxLat] - overrides everything else.
categories	Overture category ids. A parent id also matches children (restaurant includes pizza_restaurant). Empty = all.
keywords	Name must contain any of these words.
requireWebsite, requirePhone	Keep only places that have them.
minConfidence	0-1; 0.7+ filters most dead/odd entries.
maxResults	Best confidence first (default 20).

{ "city": "Berlin", "radiusKm": 5, "categories": ["dental_clinic"], "requireWebsite": true, "minConfidence": 0.7, "maxResults": 100 }

Common category ids

Use	Id	Use	Id
Restaurants	restaurant	Dentists	dental_clinic (alias dentist)
Cafes	cafe, coffee_shop	Plumbers	plumbing (alias plumber)
Bars	bar	Electricians	electrician
Hotels	hotel	Lawyers	attorney_or_law_firm (alias lawyer)
Hair salons	hair_salon	Real estate	real_estate_agent
Beauty	beauty_salon	Doctors	doctors_office
Gyms	gym	Pharmacies	pharmacy
Grocery	grocery_store	Car repair	automotive_repair
Accountants/tax	tax_service	Architects	architect

The full list is the Overture taxonomy.

Output example

{
  "id": "815af368-a492-478a-8f03-b63e4cbfd6cd",
  "name": "Cloud Nine Restaurant Hanoi",
  "category": "restaurant",
  "alternateCategories": [],
  "address": {"freeform": "27 P. Hang Be", "locality": "Quan Hoan Kiem", "region": null, "postcode": "100000", "country": "VN"},
  "lat": 21.033088, "lng": 105.853867,
  "website": "http://cloudninerestaurants.com/",
  "phone": "+842439290011",
  "socials": ["https://www.facebook.com/1985610115067306"],
  "brand": null,
  "confidence": 0.998,
  "sources": ["Overture", "meta"],
  "overtureRelease": "2026-10-21.0",
  "attribution": "Overture Maps Foundation, Places theme (CDLA-Permissive-2.0; ...)"
}

The latest Overture release is detected automatically on every run.

Pricing

Pay per event: $1.50 per 1,000 places (place event, $0.0015). Google-Maps-based lead scrapers typically charge about $5 per 1,000.

Data quality (measured on release 2026-10-21.0)

Query	Places	with website	with phone	with street address
Hanoi, restaurants, 2 km	2,000 (cap)	39%	89%	97%
Berlin, dental clinics, 2 km	268	74%	96%	100%
Austin, plumbing, 5 km	26	77%	92%	96%
Coverage is best in cities and weaker in small towns. No reviews, opening hours or emails. Data can be months old; check confidence.

Limitations

• Overture is an aggregate of open datasets (Meta, Microsoft, Foursquare, OSM-derived, ...). It is not Google Maps: no ratings or reviews.
• Phone numbers are the business numbers as published in the dataset. You are responsible for lawful use (GDPR, anti-spam rules); sole traders' numbers may be personal data in the EU.

License and attribution

Overture Maps Places is published under CDLA-Permissive-2.0 (individual source datasets carry their own permissive licenses, e.g. Meta CC0, Foursquare Apache-2.0). Attribution is required: "Data from Overture Maps Foundation, Places theme, https://overturemaps.org/attribution". It is included in every output item. This actor is not affiliated with Overture Maps Foundation.

FAQ

Why 0 results? Category ids must match the Overture taxonomy (dental_clinic, not dentist - aliases exist for a few). Widen the radius or lower minConfidence. Memory? Runs in 512 MB (peak RSS about 0.5 GB); 1 GB is the default. Speed? 15-35 s per query, mostly network reads of the public dataset.