From plaintext to structured address
Real-world address data is semi-structured: street names and house numbers frequently appear together, postcodes are sometimes inline with the city, and the country may be missing entirely. The Address Parser turns those plain-text strings into clean, structured fields.
The solution combines trained neural networks, address databases and term dictionaries, alongside open-source tools and open data, to parse and geocode addresses across countries and formats.
Address parsing integrates with Search Cluster for indexing, record matching and geospatial search. The geocoder produces latitude/longitude coordinates for location context.
What gets pulled out
Postcode & place
Postal codes and place names, including alternative spellings and transliterations.
Country
Country names and ISO codes, recognised in multiple languages.
State / county
Federal states, counties, cantons, regions — depending on country structure.
Street & building
Street name, house number, and building identifier as separate semantic types.
Apartment / floor
Apartment, floor, staircase — recognised even when written together with the street.
Delivery notes
PO Box and other delivery instructions extracted into dedicated fields.
Geocoordinates
Latitude / longitude derived from the parsed address for location-based features.
What the components unlock
Address matching
Feed structured components into Address Matcher for high-quality record reconciliation.
Form normalisation
Split a single free-text address field into clean street / number / city / country in your CRM.
Geospatial search
Geocoded coordinates power radius-search, mapping and proximity-based features.
Logistics & routing
Validate deliverability and pre-route shipments based on extracted postal data.
Parse addresses in your stack
We're happy to walk you through the API and benchmark our parser on your address data.