Ads Geo

Location targeting

Available levels

Verified against the live API on 2026-09-04 for Germany:

postal_code appears in no documentation but works. Postcode-level targeting is therefore possible.

dma does not exist for Germany. It never appeared in any query and looks US-specific.

Radius targeting does not exist. Model a catchment area as a list of postcodes. In practice that is more precise, because it follows municipal boundaries rather than a circle.

With no location set, the campaign runs across every available area.

Three pitfalls

Umlauts are mandatory.

The search_geo tool rewrites ae/oe/ue automatically and tries both forms. Calling the API directly requires handling this yourself.

English exonyms find nothing German. Munich returns two results, none with country_code: DE. Same for Cologne.

Districts and localities do not exist. A German Landkreis returns no results, and neither do sub-municipal localities — those are covered by the parent town's postcode.

Duplicate names. Birkenfeld exists three times: 55765, 75217 and 97834. Always select by postcode, never by name.

Procedure

1. search_geo("Karlsruhe", country="DE")  -> collect IDs
2. Verify the postcodes match the intended place
3. create_campaign(location_ids=[...])

The resulting targeting object:

{"targeting": {"locations": {"include": [{"id": "10020757"}]}}}

Building a catchment area

Three tiers, from tight to wide. Pick by whether the service needs physical proximity.

Tier 1 — immediate surroundings. The postcodes of the home town and its neighbours. Suitable when someone has to travel to the customer.

Tier 2 — economic area. The nearest cities, each as a set of postcodes. A mid-sized German city typically has 7 to 15 postcode entries; large cities have 35 to 99.

Tier 3 — region or country. One region ID, or the country. Suitable when the service is delivered remotely.

A tight area is not automatically better. On a young platform with thin volume, narrow targeting can mean almost nothing gets delivered. If the service can be delivered remotely, there is little reason to restrict — unless the copy deliberately plays on local proximity.

Sensible approach: start wide, then narrow based on results. Reporting with segment="country" shows where delivery lands. There is no breakdown by postcode.

Local proximity in the copy

Because delivery can only be steered coarsely, proximity mostly works through wording:

• "Based in ", "On site in ", "Serving since "

This excludes nobody, but speaks to the people who care about proximity — and unlike narrow targeting, it costs no reach.