What is Geo-targeting?

Geo-targeting looks like a simple question — which places should this campaign reach — and the consequential decision is a different one hidden beside it.

Every major platform offers a choice about what a location means. One option restricts delivery to people in or regularly in the place. Another includes people anywhere who have shown interest in it: searched for it, looked at content about it, or otherwise signalled a connection.

The second is frequently the default, and it is right for some businesses and disastrous for others.

A hotel wants people interested in its city, wherever they are. A plumber wants people in the suburb, and does not want to advertise to somebody researching the area from another continent. An account that never opened the setting has accepted whichever default the platform chose, and in the second case the waste is substantial and invisible in any report that only shows performance by targeted location.

Several signals combine, and they differ in precision.

Device location, where the person has granted permission, is the most accurate.

Network address places a device by where its connection reaches the internet, which is usually close and sometimes not. In dense urban areas an address can resolve some distance from the actual location, and on mobile networks it can resolve to wherever the carrier's infrastructure sits.

Behavioural signals — the places somebody interacts with, searches about, or regularly appears near — inform the platform's view of where a person is based rather than where they are now.

The practical consequence is a scale of reliability. Country and region targeting is dependable. City targeting is good. Radius targeting around a point is approximate, and a small radius is approximate enough that a business drawing from a few streets should not rely on it alone.

Business constraints. A service licensed in one jurisdiction, a product that cannot be shipped somewhere, a delivery area with a hard edge. These belong in the targeting because they are facts, and they should be stated as exclusions as well as inclusions where the consequence of reaching the wrong place is more than wasted spend.

Market separation. Where markets differ in language, price, competitive position or customer value, separating them lets each carry its own budget, creative and target. This is one of the few structural splits that usually justifies itself.

Concentration. A business whose customers cluster geographically can concentrate spend where the density is, which is a different decision from excluding everywhere else.

Chasing performance patterns. Location-level performance reports fragment an account's conversions across many regions, and most of the resulting cells are too thin to distinguish from noise. An account that prunes regions on a handful of conversions each is making decisions on random variation, and the pruning is hard to reverse because the excluded regions stop generating the data that would justify their return.

Duplicating what bidding already does. Under an automated strategy, location is one of the signals in the per-auction prediction. Layering manual adjustments on top applies a coarse multiplier to a fine estimate.

Micro-radius targeting. Below the reliability of the underlying signals, a tight radius promises a precision the data cannot deliver and quietly excludes people who are genuinely nearby.

Not targeting a place and excluding it are not the same.

An omission leaves the location outside whatever was included. An exclusion overrides inclusions that would otherwise reach it — a region inside a targeted country, a city inside a targeted state.

Where the consequence of appearing somewhere is regulatory or operational rather than merely wasteful, the explicit exclusion is the setting that refuses. It also survives changes to the inclusions, which an omission does not: broadening a country later silently re-admits every place that was merely not listed.

Two reports are worth reading regularly and they answer different questions.

Performance by targeted location says how the places you chose are doing. It is the one most accounts look at and it is the less informative of the two.

Performance by location of presence, where the platform distinguishes it, says where the people actually were. For an account using the interest-based setting, the gap between these two is where the surprises live, and it is the only place the cost of the default becomes visible.

Open the presence setting and choose it deliberately rather than accepting the default.

State hard business constraints as explicit exclusions, not as omissions.

Separate markets that genuinely differ; do not separate markets that merely have different names.

Leave fine-grained location optimisation to the bid strategy where one is running, and reserve manual adjustments for facts somebody can state in a sentence.

And check the actual-location report at least once, because a large share of geo-targeting problems are invisible in every other view of the account.

Where markets genuinely differ, a geographic split is worth more than a targeting setting, and it is worth doing thoroughly.

Different markets frequently need different currency and pricing in the copy, different landing pages, different offers, different seasonality and different competitive positioning. A single campaign targeting several of them with one set of ads is serving all of them adequately and none of them well.

The counterweight is conversion volume. Splitting a campaign into five geographic children divides the events the bid strategy learns from by five, and five campaigns each below the threshold learn worse than one above it.

The resolution is usually to split where the differences are real and to resist splitting where they are cosmetic. Two cities in the same country with the same offer, the same page and the same language are one market with two names.

Location interacts with scheduling in a way that catches accounts out.

Platforms report and schedule in the account's configured time zone, not the viewer's. A campaign targeting several zones with a daily schedule applies that schedule uniformly, so a nine-to-five restriction is nine-to-five in one place and something else everywhere else.

For an account selling across a wide country or internationally, this is worth checking before any hourly analysis is believed and before any schedule is set. The fix is either to split by zone where the schedule genuinely matters, or to stop scheduling and let the bid strategy handle the variation, which it does at a finer grain anyway.