Geo Targeting

Run an ad with no geographic limit and the impressions spread wide. The trouble is that a large share of them are people you can't sell to right now. With a single store in one city, every click from another one is a straight loss.

Ravi hit this the month the ads ran against an English landing page. Clicks came in above forecast while the share converting to signups stayed low. Splitting the report by location showed half the clicks arriving from countries where payment wasn't supported yet.

This setting assumes that where the money leaks and where the results come from divide along geography. So the audience gets cut by location before anyone starts cutting by disposition.

Start by writing the scope as a sentence. "International" can't be turned into a setting, so write it as the list of countries where payment and support both work.

1. Choose the unit of scope: country, state or province, city, postal code, or a radius in kilometers around a point. Store ads want a radius; a payment-support problem wants countries.
2. Check what determines location. Device location, a guess from the connecting address, the region a user entered in their profile, and place names inside the search query all get used. None is always right — connecting through a corporate network can place someone somewhere else entirely.
3. Use inclusion and exclusion together. Taking the whole country and excluding a few places is easier to maintain than adding thirty places one at a time.
4. Split creative and budget per location. Language, currency, and the hours when ads should run all shift with geography.
5. Break the results out by location. Differences invisible in the overall average surface here.

Also check whether the platform decides on place of residence or current location. Defaults differ, and that is where showing ads to someone who is traveling gets decided. For an audience that travels for work, the gap between those two readings is wide enough to change what the report says.

"Put countries without payment support on the exclusion list. We're paying for clicks and nothing else."

Remove places where conversion is impossible to protect the budget.

"We only have one city switched on, but inquiries are coming from the next province. The location guess can't be that accurate."

Some share of the location estimates is landing wrong.

"Broken out by location, one city converts at twice the rate. Let's move budget."

Results divided along geography, so the allocation should follow.

Pagedeck settled on the regions where both payment and Korean-language support work and ran ads against that list. What Ravi did next was decline to narrow further by age or interest. Stacking narrowing conditions shrinks impressions until there isn't enough data for the platform to learn from. Demographic targeting goes on later, once per-region results are stable, and interest targeting isn't in use yet.