What Is the Best Geocoding API?

There is no single best geocoding API. The right one depends on whether you need to store results, how much you geocode at once, which countries you operate in, and what your privacy obligations are.

That framing matters because most comparisons stop at accuracy and price per request. Those are rarely what breaks a project. The things that break projects are storage restrictions discovered at scale, batch limits that turn a one-hour job into a three-day job, display terms that lock your results to one vendor's map, and data you cannot inspect or correct.

This page walks through the dimensions worth evaluating, and what the Stadia Maps Geocoding API does on each.

Key Takeaways

• Storage rights are easy to overlook and expensive to discover late. Per Stadia Maps' published analysis, Mapbox, AWS, and ESRI often charge 6x to 8x the standard query price to store a result.
• Batch capacity varies widely. Google lacks a dedicated bulk geocoding API, Mapbox limits requests to 1,000 queries, and Stadia Maps processes 5,000 per request.
• Data provenance is inspectable on some providers and opaque on others. Stadia Maps documents five open sources and returns fix-it URLs so records can be corrected.
• Display restrictions matter. Google generally prevents displaying its geocoding results on another provider's map.
• Stadia Maps geocoding data refreshes at least monthly, draws on a dataset of over one billion locations globally, and does not track end users.

What Should You Evaluate in a Geocoding API?

Six dimensions, roughly in order of how often they cause problems after launch.

1. Storage and caching rights. Can you write a returned coordinate to your database and keep it? This is a licensing question, not a technical one, and it is where cost models break. On Stadia Maps, temporary storage is allowed on all plans and permanent storage requires Standard or above with no surcharge. The Terms of Service contain the full terms.

2. Batch capacity. If you need to geocode a hundred thousand addresses, the per-request limit determines whether that is a script that runs over lunch or a pipeline you have to babysit.

3. Data sources and transparency. Where do the addresses come from, how often are they refreshed, and can you correct an error you find?

4. Display and portability terms. Some providers restrict where their results can be shown. That constrains your architecture later, not now.

5. International accuracy. Address formats, scripts, and data density vary enormously by country. A provider that is excellent in North America may be weak where you are expanding next.

6. Privacy posture. Whether the provider collects end-user data, and whether that is compatible with your compliance obligations.

Geocoding API Comparison: Storage, Batch, and Portability

The competitor figures come from Stadia Maps' published analysis in The Hidden Cost of Search, which places AWS and ESRI in the same 6x to 8x storage band as Mapbox. The Stadia Maps storage terms are in Section 8 of the Terms of Service, which requires an active Standard, Professional, or Enterprise subscription to permanently store geocoding results.

Provider terms change, and each vendor's are their own. Verify against current documentation before deciding.

One more term worth checking that is harder to tabulate: display rights. The same analysis notes that Google often prevents displaying its geocoding results on a map from another provider. Storage rights and display rights are separate clauses, and either one can create lock-in.

Where Does Geocoding Data Come From?

A geocoder is only as good as the data underneath it, and most providers treat that data as a trade secret. The Stadia Maps sources documentation lists all five, and every result carries attribution back to its origin:

• OpenStreetMap for global roads, addresses, and points of interest
• OpenAddresses for authoritative government address data. Stadia Maps is a corporate sponsor of the project
• Who's On First for the administrative place hierarchy
• GeoNames for populated places and natural features
• Foursquare Open Source Places for business and venue data

Two consequences of building on open data, both from The Open Data Superpower:

You can fix errors. Stadia Maps includes "fix-it" URLs for many records directly in API responses, pointing to datasets that accept contributions. Reporting a bad address to a proprietary provider is, in that post's words, "shouting into a void."

Updates are continuous rather than periodic. Legacy providers rely on data captured during periodic drive-bys, which can remain unverified for years. Stadia Maps refreshes geocoding data at least monthly.

Which Geocoding Endpoint Do You Actually Need?

"Geocoding API" covers several distinct operations. Picking the wrong one produces bad results or unnecessary cost.

• Forward geocoding turns a complete address or place name into coordinates. Best accuracy on complete input, and it uses approximation to locate addresses that are not mapped directly.
• Autocomplete search is tuned for speed on partial input. Use it while the user is typing, then hand off to forward geocoding when they finish. See What Is Address Autocomplete?
• Structured geocoding takes pre-parsed components (house_number, street, city, region, postalcode, country). Removes string-parsing guesswork, and improves international accuracy.
• Reverse geocoding turns coordinates into a place. See What Is Reverse Geocoding?
• Bulk geocoding handles up to 5,000 queries in one request for pipeline work.

On the Stadia Maps credit system, forward, structured, and reverse geocoding each cost 20 credits per request, and autocomplete v2 costs 1 credit per request. Bulk geocoding is billed per query inside the batch at the same rate as an individual request, so batching costs no more than calling the endpoints directly.

How Do You Test a Geocoding API?

The fastest way to compare providers is to run your own hardest addresses through each. Here is a forward geocoding request against Stadia Maps:

GET https://api.stadiamaps.com/geocoding/v1/search?text=P%C3%B5hja+pst+27

The response is a GeoJSON FeatureCollection. Each feature carries coordinates, parsed address components, the matched layer, a confidence score, and source attribution. That last part matters for evaluation: you can see which dataset produced the answer.

For runnable examples in TypeScript, Python, Kotlin, Swift, PHP, or cURL, see the forward geocoding documentation. An EU endpoint is available at api-eu.stadiamaps.com to keep traffic within the EU.

Build your test set carefully. Do not evaluate on well-formed addresses in major US cities, where every provider performs well. Test the cases that will actually hurt: apartment and unit numbers, addresses in the countries you are expanding into, non-Latin scripts, new construction, rural routes, and deliberately misspelled input.

How Good Is International Geocoding?

It varies by provider and by country, and any vendor claiming uniform global accuracy is overselling.

What Stadia Maps documents on this:

• Multilingual search across all primary data layers, including administrative areas and points of interest. Searching a Korean city by its English name is the kind of case this is built to handle.
• Localized address templates developed through an open-source collaboration, so results respect how addresses are actually written in each country.
• A stated limitation. Per The Open Data Superpower, Stadia Maps is "continuing to refine partial matches in specific East Asian scripts."

That last item is the kind of thing worth looking for in any vendor's documentation. A provider willing to name where it is still improving is easier to plan around than one that claims everything works.

Read the Determining Result Quality documentation before you build. Confidence scores and layer types are how you detect a weak match in production rather than shipping it to a user.

When Is Stadia Maps the Right Choice?

You are geocoding once and storing the result. Permanent storage is included on Standard and above with no surcharge and no per-user re-validation. If your data is stable and access is repeated, this is a significant cost difference between providers.

You are processing addresses in bulk. 5,000 queries per request, with no extra credits or fees for batch mode.

You need to inspect or correct the data. Five documented sources, per-result attribution, and fix-it URLs in responses.

You have privacy or residency requirements. No end-user tracking, query-based rather than session-based search, and an EU endpoint for keeping traffic in-region. See What Is a GDPR-Compliant Mapping API?

You need to self-host. The geocoding engine is available for on-premises deployment for teams with strict security or latency requirements.

Where you should look elsewhere: if your application depends on a proprietary business listing dataset that only one vendor has, no amount of open data will replace it. Evaluate on your own hardest addresses before committing either way.

Where to Go Next

From the Blog

More on geocoding from the Stadia Maps blog:

• The Hidden Cost of Search: Why Your Geocoding Bill Is Higher Than It Should Be
• The Open Data Superpower: Why Global Search Is Moving Beyond Proprietary Silos
• Precision Meets Privacy: Elevating the Consumer Search Experience
• 75 Million More Addresses: Expanding Geocoding Precision
• Bulk Geocoding API Is Now Available

Geocoding & Search

• What Is Geocoding? for how geocoding works and where the data comes from
• What Is Reverse Geocoding? for the coordinates-to-place direction
• What Is Address Autocomplete? for search-as-you-type
• Geocoding & Search overview docs for endpoint selection
• Terms of Service for the full storage and display terms
• Sources documentation for the full data provenance
• Determining Result Quality for confidence and layer signals
• Geocoding product page and pricing
• Google Maps alternatives and Mapbox alternatives for migration detail