When a user types a postal address into your application and you need to drop a pin on a map, calculate distance to a store, or sort delivery routes by proximity, you need coordinates. The Google Geocoding API converts addresses into latitude and longitude pairs (forward geocoding) and the reverse — coordinates back into a readable street address.
It sits behind a vast number of food delivery apps, real estate platforms, ride sharing services, and any product where physical location matters. The service runs on the same address database that powers Google Maps and Google Search.
Why pick Google over open alternatives?
If a postal address exists anywhere in Google's index, you will find it here, including partial inputs like "Eiffel Tower" or "Pizza Hut Brooklyn near 5th Ave." The accuracy across dense urban centers, rural villages, and emerging markets is where Google still pulls ahead of Mapbox, HERE, and OpenCage.
For a logistics startup serving routes across India, Southeast Asia, or Sub-Saharan Africa, the address coverage gap can decide whether the project ships at all. Google's data depth in those regions is hard to match.
When this fits a project
Use cases that fit cleanly include checkout address validation in e-commerce, store locators on a brand website, dispatching technicians to a service address, and onboarding flows for fintech products that require KYC address verification.
The API also handles place name disambiguation — passing "Springfield" returns multiple matches with country and state context so the user can pick the one they meant. Reverse geocoding helps with sharing locations from GPS coordinates back into readable form.
When to skip it
Three situations come up often. If you are building a hobby project or a developer tool that needs to convert an address only occasionally, the free tier of OpenCage (2,500 requests per day) or Nominatim (run it yourself) saves you setting up billing.
If your application primarily serves one country with strong open address data — France with the Base Adresse Nationale, or the UK with Ordnance Survey — local government APIs are free and often more accurate.
If you are doing batch geocoding of millions of records once for analytics, look at HERE Geocoding or Mapbox bulk pricing rather than per-request Google calls.
How to get started
Setup takes about ten minutes. Create a Google Cloud project, enable the Geocoding API in the API library, generate an API key, and restrict that key to your application's domains or IP addresses.
Key restriction matters — unrestricted keys get scraped within hours and you pay the bill. The first request is a simple HTTPS GET to maps.googleapis.com/maps/api/geocode/json with an address parameter.
The response includes formatted_address, geometry.location.lat, geometry.location.lng, and address_components broken down by street, locality, postal code, and country. For reverse geocoding, you pass latlng instead of address.
Pricing in 2026
Google offers 10,000 free requests per month under its monthly $200 credit applied across Maps Platform products. After that, geocoding costs $5 per 1,000 requests.
A small e-commerce site with 5,000 monthly checkouts stays free. A logistics platform doing route optimization for 200 drivers running 50 stops each across 25 working days hits 250,000 monthly requests — roughly $1,250 per month after the credit.
The pricing escalates fast. Track your usage in the Cloud Console from day one and set budget alerts before traffic ramps up. Public pricing reference: cloud.google.com/maps-platform/pricing.
Alternatives worth knowing
- Mapbox Geocoding — 100,000 free requests per month, then $0.75 per 1,000. Significantly cheaper at scale and close enough on accuracy for most North American and European use cases.
- OpenCage — Free tier 2,500 per day, paid plans from $50 per month. Powered by OpenStreetMap. Great for budget-conscious projects.
- HERE Geocoding — 1,000 free per day, strong on logistics and trucking with vehicle-specific routing data.
- Pelias — Open source if you want to self-host with engineers and an OpenStreetMap data pipeline.
- Nominatim — Free OpenStreetMap-based geocoder you can run yourself or use through public instances (with usage limits).
Production details that matter
The API enforces a soft rate limit of 50 queries per second per project, which is high enough for most applications but worth knowing if you process bursts.
Caching results aggressively makes a real budget difference — most addresses do not change, so a Redis cache keyed on the normalized address string can cut your monthly bill by 60 to 80 percent.
Google's terms of service prohibit storing geocoded results for more than 30 days unless you display them on a Google Map. This trips up developers building data warehouses — read the terms before you architect long term storage.
The API handles autocomplete-friendly queries too. Partial addresses like "123 Main" return ranked candidates which is what you want for a typeahead checkout field. Pair it with the Places Autocomplete API for a complete address entry experience.