Skip to content

Alternatives

Where else to get boundaries, honestly

Geocode AI isn't the right answer for everyone. Here's what each common source is good at, and when you should use it instead.

Boundary data sources compared
Geocode AIGADM / geoBoundariesCensus TIGERNatural Earth / MapboxGoogle stylingRaw OSM
API: search, get, reverseMapbox: tilesStyling onlyPublic servers, light use
Global coverageUS only
Cities, districts, postcodesVaries by countryMapbox yes, Natural Earth no
A polygon you can storeNatural Earth yes, Mapbox licensed
CostFree tier, then per unitFree (GADM: non-commercial)FreeNatural Earth free, Mapbox via salesGoogle Maps pricingFree, policy limits
Describe an area in words

Natural Earth and Mapbox share a column for space; their differences are in the cards below.

Each option in one card

  • GADM

    Downloadable dataset

    Good at: Consistent administrative levels for every country, popular in research.

    Limits: Static files, updated every few years. Free for academic and non-commercial use; commercial use needs permission.

    Pick it for research and non-commercial analysis where a fixed snapshot is fine.

  • geoBoundaries

    Downloadable dataset

    Good at: Open (mostly CC BY), documented sources per country, several admin levels.

    Limits: Files, not an API for search, reverse lookups or point checks. Levels and coverage vary by country.

    Pick it when you want an open, citable dataset to load into your own GIS.

  • US Census TIGER/Line

    Government dataset (US)

    Good at: Authoritative US states, counties, places, ZCTAs and more, public domain.

    Limits: United States only, and shapefiles you host and query yourself.

    Pick it for US-only work that must match Census geography exactly.

  • Natural Earth

    Downloadable dataset

    Good at: Clean, public-domain countries and first-level regions for small-scale maps.

    Limits: Generalized for world and continent maps; no cities, counties or districts.

    Pick it for world maps and dashboards at country or state level.

  • Mapbox Boundaries

    Commercial tilesets

    Good at: Global boundaries as vector tiles for styling and data joins on Mapbox maps.

    Limits: A licensed product for use with Mapbox maps, sold through sales rather than per call.

    Pick it if you are already on Mapbox and need styled choropleths at scale.

  • Google Maps boundaries

    Map styling

    Good at: Data-driven styling lets you colour Google's own boundary features on a Google map.

    Limits: Styling only: you don't get the polygon to store, query or export.

    Pick it to highlight regions on a Google map in your UI.

  • OpenStreetMap directly

    Open data, public services

    Good at: The most detailed open boundaries, from countries to neighbourhoods, updated continuously.

    Limits: You assemble relations into polygons, cache, and stay within the public services' usage policies.

    Pick it for one-off files, or if you run your own Nominatim and Overpass.

When Geocode AI fits

You need boundaries at request time, worldwide, from countries down to districts and postcodes: by name, by point, with the hierarchy, as files or as point checks. Or your areas are written in words and you need them as polygons.