Sample address datasets

Ready-made, multi-country test files generated with a fixed seed. Use them as fixtures, as demo data, or to check how your system handles non-Latin scripts, missing postcodes and unusual layouts. Version 2026.10.1, built 2026-10-26.

FileRowsSizeSHA-256
World sample addresses (CSV)
10 synthetic addresses for every country with an address generator (levels A, B and C).
world-addresses-2026.10.1.csv
2,430337 KB29f322e9c041b4d05aefa6b7b582c4b9b525762912a8387c3eaa6d11084df1e5
World sample addresses (JSON)
The same records as a JSON array.
world-addresses-2026.10.1.json
2,430867 KBecdacad68a8e75844b81c72c2359d00cbbb69620ecf0381c6ce1e4ffe797e3ae
Dataset-matched addresses only (CSV)
Level A countries only: every locality, region and postcode is from one dataset record.
dataset-matched-addresses-2026.10.1.csv
1,210167 KB61a47e3b84515e7369fc5610a3df8a2d2bb984e5a49c0610a1d6fb5645986ae0
World sample profiles (JSON)
10 full synthetic profiles per level A and B country, with every field.
world-profiles-2026.10.1.json
1,9101.7 MB8e2561b75936f34290e15390bb924322cb255b45174a7246681f2db502c85322

What the files contain

  • CSV files are UTF-8 with a byte order mark, comma-separated, quoted per RFC 4180, with CRLF line endings.
  • Missing values are empty in CSV and null in JSON. For example, postal_code is empty where a country has no postcodes or no open postcode mapping; postal_status says which.
  • Local-script columns (street_line, city, region) sit next to *_latin columns.
  • quality is the coverage level (A, B or C), explained on the data sources page.

Need other fields, more rows or a different seed? Use the test data generator.

Licence and attribution

The files are synthetic test data, released under CC BY 4.0. They contain place names and postcodes derived from GeoNames (CC BY 4.0) and Japan Post, so credit “Contains GeoNames data” when you redistribute them.

No row describes a real person or a verified address. Do not use the files to send mail, or to stand in for real people outside testing.