Geocoding vs Reverse Geocoding: How Addresses Become Coordinates (and Back)

Type "221B Baker Street, London" into a map and a pin appears at 51.5238, -0.1586. Click a random point in a satellite photo and the map names the street. These two translations — address → coordinates and coordinates → address — are called geocoding and reverse geocoding, and they power nearly every location feature you use. They are also messier than they look.

Geocoding: Address → Coordinates

A geocoder parses free-form text into structured components (street number, street, city, postal code, country), matches them against a reference database, and returns coordinates. The interesting part is how good the match is — geocoders report distinct quality levels:

This is why a delivery app sometimes sends drivers to the wrong side of a block: the geocoder silently fell back from rooftop to interpolation. Serious applications always check the returned match level, not just the coordinates.

Reverse Geocoding: Coordinates → Address

The reverse direction searches the database for the nearest known feature to a point. It is inherently fuzzier:

• A point on a building returns that building's address — usually reliable.
• A point in a park, lake, or highway median returns the nearest address, which may be hundreds of meters away and semantically wrong.
• Results are sensitive to tiny coordinate changes near boundaries: moving 10 m can flip the returned district, postal code, or even city.

A subtle consequence: geocoding and reverse geocoding are not inverses. Geocode an address, reverse-geocode the result, and you may get a different address — the nearest database feature to the interpolated point.

Why Providers Disagree

Geocode the same address with Google, OpenStreetMap's Nominatim, and a national postal service, and you will get three slightly different answers. The differences come from:

• Reference data: Google uses licensed and self-collected data; Nominatim uses crowd-sourced OpenStreetMap; postal services use official registries. Coverage and freshness vary by country.
• Parsing rules: "Seoul Station" vs "1 Hangang-daero" vs a Korean-language query can resolve differently depending on the parser's language handling.
• Licensing: this matters more than people expect. Most commercial geocoders prohibit storing results or displaying them on non-affiliated maps. OpenStreetMap/Nominatim data is open (ODbL license) — which is why open tools, including ours, build on it.

Practical Tips

1. Batch jobs: respect rate limits (public Nominatim asks for max 1 request/second) and cache results.
2. Ambiguous input: always pass country or city context when you have it — "Springfield" alone matches dozens of places.
3. Validation: for critical data, geocode, then reverse-geocode the result and compare — large disagreement flags a weak match.
4. Precision honesty: an interpolated match reported with six decimal places is still a ~50 m guess. Don't let decimals imply accuracy the match level doesn't support. (The same trap exists in GPS readings.)

Try Both Directions

Our GPS coordinates finder does both live: search any address or place to geocode it (via OpenStreetMap), or click anywhere on the map to reverse-geocode the point — with the coordinates shown in DD, DMS, and DDM simultaneously. For IP-based location instead of address-based, see IP geolocation.