Skip to content

Address Geocoder

Try for free

Turn addresses into latitude and longitude with a clean full address, its parts (street, number, city, region, postcode, country) and a match confidence. Reverse mode turns coordinates into the nearest address. US addresses use the US Census Geocoder, other countries OpenStreetMap.

📍 Address Geocoder

Paste addresses and get their latitude and longitude, a clean full address, its parts (house number, street, city, state or region, postcode, country) and a match confidence from 0 to 1. Reverse mode does the opposite: give it coordinates and it returns the nearest address. US addresses are matched with the US Census Geocoder, every other country with OpenStreetMap. You can also point it at the results of another run, such as a lead or store list, and it geocodes the address fields of every row. No API key.

✨ Why use Address Geocoder?

  • 🇺🇸 Fast and accurate in the US: US addresses go to the US Census Geocoder in batches. 1,000 different San Francisco addresses from the city's own records took about 40 seconds; 98.6% landed within 100 m of the city's address points and 99.8% within 250 m (median 22 m).
  • 🌍 The rest of the world too: addresses outside the US are matched against OpenStreetMap. In our tests, 150 of 150 Dutch addresses were placed within 100 m.
  • 🧹 Clean parts, not just a point: house number, street, city, region, postcode, country and ISO country code in separate fields, plus one formatted address line.
  • 🎯 Honest confidence: each result is checked against what you typed. A result in another town, on another street or with another house number gets a low score instead of looking like a perfect match.
  • 🔄 Reverse geocoding: coordinates to the nearest address, with the distance between your point and that address.
  • 🔗 Works on other runs' results: geocode a whole dataset, from one address field or from separate street, city, state, postcode and country fields.

🎯 Use cases

  • 🗺️ Put a lead or store list on a map: add coordinates to every row before importing it into a CRM, BI tool or mapping app.
  • 📦 Delivery and territory planning: get coordinates to measure distances and assign addresses to areas.
  • 🧾 Clean and split addresses: turn messy one-line addresses into separate fields.
  • 📍 Name a GPS point: turn coordinates from a device, photo or app into a street address.

🚀 How to use

  1. Paste addresses into Addresses, one per line, or set Dataset ID to the results of another run and name the fields to read (Address field for one-line addresses, or the street, city, state, postcode and country fields).
  2. For reverse geocoding, set Mode to Coordinates → address and paste latitude, longitude pairs into Coordinates (or name the latitude and longitude fields of a dataset).
  3. Click Start and download the results as JSON, CSV, Excel or HTML, or send them on with an integration.

Input fields

FieldWhat it does
ModeAddress → coordinates (default) or Coordinates → address (reverse)
AddressesOne address per line. Messy text is fine: extra spaces, upper or lower case, apartment or suite numbers, "fifth avenue" for "5th Ave". A Google Maps or Apple Maps link works too; the address and map position in the link are used
CoordinatesReverse mode: one latitude, longitude pair per line. 48.8584 2.2945, (48.8584,2.2945) and 48.8584;2.2945 work too
Dataset IDGeocode the rows of another run's dataset instead of (or as well as) the lists above
Address fieldDataset field with the whole address on one line (dotted paths like location.fullAddress work). In reverse mode, a field with lat, lon in one value
House number, Street, City, State or region, Postcode, Country fieldFor datasets with the address split into fields. Rows whose country is the US go to the US Census Geocoder
Latitude field, Longitude fieldReverse mode: dataset fields with the latitude and the longitude
Default countryAdded to addresses that do not name a country, for example Germany for Marienplatz 1, München. US sends them to the US Census Geocoder. When an address names another country (10 Downing St, London, UK), that one wins
Language of place namesOutside the US: country, region and city names in English (default), German, French or the local language (Deutschland, Санкт-Петербург). US results are always in English
Remove duplicatesOn by default: each address (or coordinate pair) is looked up once, ignoring case, spaces and commas. Skipped duplicates are not charged. Turn it off to get one row per input line

📥 Input example

{
    "addresses": [
        "350 5th Ave, New York, NY 10118",
        "1600 Pennsylvania Ave NW, Washington, DC 20500",
        "10 Downing St, London, UK",
        "Unter den Linden 77, Berlin, Germany",
        "Av. Paulista 1578, São Paulo, Brazil"
    ]
}

To geocode another run's results with the address split into fields:

{
    "datasetId": "YOUR_DATASET_ID",
    "streetField": "street",
    "cityField": "city",
    "regionField": "state",
    "postcodeField": "postalCode",
    "countryField": "countryCode"
}

Reverse mode:

{
    "mode": "reverse",
    "coordinates": ["48.8584, 2.2945", "40.7484, -73.9857"]
}

📤 Output example

{
    "input": "Unter den Linden 77, Berlin, Germany",
    "found": true,
    "latitude": 52.5155757,
    "longitude": 13.3800699,
    "formattedAddress": "Unter den Linden 77, 10117 Berlin, Germany",
    "houseNumber": "77",
    "street": "Unter den Linden",
    "city": "Berlin",
    "region": null,
    "postcode": "10117",
    "country": "Germany",
    "countryCode": "DE",
    "precision": "address",
    "confidence": 0.9,
    "source": "OpenStreetMap (Photon)",
    "error": null
}

A US address:

{
    "input": "350 5th Ave, New York, NY 10118",
    "found": true,
    "latitude": 40.7478508,
    "longitude": -73.9850756,
    "formattedAddress": "350 5th Ave, New York, NY 10118, United States",
    "houseNumber": "350",
    "street": "5th Ave",
    "city": "New York",
    "region": "NY",
    "postcode": "10118",
    "country": "United States",
    "countryCode": "US",
    "precision": "address",
    "confidence": 1,
    "source": "US Census Geocoder",
    "error": null
}

A reverse lookup (48.8584, 2.2945):

{
    "input": "48.8584, 2.2945",
    "found": true,
    "latitude": 48.8584481,
    "longitude": 2.2945757,
    "formattedAddress": "5 Avenue Anatole France, 75007 Paris, Île-de-France, France",
    "houseNumber": "5",
    "street": "Avenue Anatole France",
    "city": "Paris",
    "region": "Île-de-France",
    "postcode": "75007",
    "country": "France",
    "countryCode": "FR",
    "precision": "address",
    "confidence": 0.9,
    "source": "OpenStreetMap (Photon)",
    "error": null,
    "distanceMeters": 8
}

Reverse lookups also get distanceMeters, the distance from your point to the address found. Rows from a dataset get sourceRow, the position of the row the address came from, so you can join the results back.

FieldWhat it is
inputThe address or coordinate pair as you gave it
foundA result was found. Rows with found: false are not charged
latitude, longitudePosition of the result (WGS 84, as in GPS and web maps)
formattedAddressThe whole address on one line, written the way that country writes it
houseNumber, street, city, region, postcode, country, countryCodeThe address parts. region is the state code for US addresses and the state or province elsewhere, where the map has one
precisionHow exact the position is: address (the building), street, neighbourhood, postcode, city, region, country or other
confidence0 to 1. 1 is an exact US Census match. 0.3 or less means the result disagrees with your input (another town, street or house number) and should be checked
sourceUS Census Geocoder or OpenStreetMap (Photon)
distanceMetersReverse mode only: distance from your point to the address found
errorWhy an input could not be looked up (empty, not an address, not a coordinate pair)
sourceRowDataset input only: position of the dataset row the address came from (0 = first row)

⚠️ Good to know

  • You pay only for addresses (or coordinate pairs) with a result. Addresses with no result, inputs with a problem and skipped duplicates are not charged.
  • US streets are abbreviated, USPS style (350 5th Ave, New York, NY), since that is how the US Census Geocoder writes them. US addresses it cannot match (new streets, some private roads) are tried on OpenStreetMap instead. When the Census service is busy, a small run uses OpenStreetMap for its US addresses after a few seconds rather than waiting; the source field shows which one answered.
  • Addresses without a town are guesses. 100 Main St, Springfield exists in dozens of states; such results come back with a low confidence. Add the state or postcode.
  • Outside the US, speed is about one address per second, because the public OpenStreetMap geocoder asks for fair use. An address it matches poorly is asked again in another form (the English city name, or without the house number), which takes a second or two more. Ten thousand US addresses take minutes; ten thousand addresses elsewhere take about three hours. Leave the run timeout long enough.
  • Coverage depends on the map. OpenStreetMap has house numbers for most of Europe and many cities elsewhere, but in some places only the street or the town is known. precision tells you which.
  • US Census points sit on the street line, worked out from the house-number range of the block, not on the building itself. For a large building set back from the road this can be 100 to 200 m off (the White House: 160 m); for ordinary houses it is usually within a few dozen metres.
  • A postcode alone is not always found. UK and most European postcodes work; full Canadian postcodes (M5V 2T6) mostly are not on the open map, so add the street or the city.
  • PO boxes have no position. They are placed at their town and postcode (precision: postcode).
  • Reverse lookups in the US also use OpenStreetMap, since the Census Geocoder only geocodes addresses.

❓ FAQ

Which countries are supported? Every country. US addresses use the US Census Geocoder, the rest OpenStreetMap.

How do I spot bad matches? Filter on confidence. Results of 0.9 and 1 matched your street, town and number. Results of 0.3 and below are in another place than you wrote and are worth a look.

Can I chain it to an earlier run automatically? Yes. Use an integration or webhook on the earlier run that starts this Actor with datasetId set to the first run's defaultDatasetId and the address fields named.

Why are some addresses only matched to the street? The map has the street but not that house number. The point is then on the street, and precision is street.

//From the blog\\

Latest articles

All articles

The internet is full of data.
Start owning yours.

Scrape.ly gives you the infrastructure, the tooling, and the team to extract, process, and operationalize web data at any scale. Setup takes minutes.

No credit card. No setup calls.