Address Matcher
Comparing postal and domicile addresses — across countries, formats and partial inputs.
Compare addresses with awareness of structure
Addresses are complex entities with many attributes — street name, house number, postal code, place name. They can be spelled out in many variations, and with various levels of completeness and explicitness.
The Address Matcher leverages the Address Parser to dissect and understand each address. The system then compares roughly two dozen semantic types using dedicated dictionaries and matching logic.
The two main use cases are search and record matching, and duplicate prevention. The matcher is part of Search Cluster and works alongside Name, Telephone and Email matchers.
All the ways an address can differ
Abbreviations
"Str." vs. "Strasse", "Av." vs. "Avenue", "Apt 4B" vs. "Apartment 4B".
Word order
"Kennedy 12" vs. "12 Kennedy Avenue" — different countries put numbers before or after the street.
Partial completeness
Postcode missing on one side, country missing on the other.
Country codes
"USA", "U.S.A.", "United States" all reduce to the same country.
Place name variants
"Munich" vs. "München", "Vienna" vs. "Wien", "Zurich" vs. "Zürich".
Misspellings
Typos in any field: street, place, postcode. Edit-distance and phonetics catch them.
Postal vs. official
"Wangen-Zürich" vs. "Wangen bei Dübendorf" — postal aliases vs. cadastral names.
Apartment / floor
Stairs, floor, apartment numbers in different formatting conventions.
PO Box vs. street
Same recipient, two delivery channels — the matcher recognises this.
Where Address Matcher fits
Duplicate detection
Find existing customers when a new record arrives, even with different formatting.
CRM data quality
Deduplicate, merge and clean customer master data across regions and systems.
KYC & compliance
Match an applicant's address against sanction lists or PEP databases.
Geospatial search
Combined with geocoding, find records within radius even when textual addresses don't match exactly.
Match addresses in your stack
We're happy to walk you through the API, run a benchmark on your data, and help you choose the right confidence threshold for your use case.