In shortMeta geo targeting is a radius around a point, not a boundary — so a campaign briefed on one city routinely delivers into neighbouring districts and states, and because the account only reports the regions it delivered to, the ones nobody targeted look exactly like the ones somebody did.
Meta geo targeting is not a boundary. It is a point and a radius, and the radius does not stop at an administrative line because administrative lines mean nothing to it. Target a city and you have drawn a circle whose edge falls wherever the circle falls.
For a coffee shop that is fine. For anything where the catchment is the whole product — a property in one suburb, a clinic people drive to, a showroom — the edge of that circle is where the budget goes to die.
Why the report will not tell you
This is the part that makes it expensive. Ads Manager will happily break delivery down by region, and the breakdown is accurate. What it does not do is mark which of those regions you asked for.
So a row reading “Uttar Pradesh — 18% of spend” looks exactly like a row reading “South West Delhi — 22% of spend”. One was the brief and one was the radius spilling over, and nothing on the screen distinguishes them. You would have to remember what was targeted and read the list against it, for every campaign, every week.
The account reports where the money went. It does not report where the money was supposed to go, so the two can differ for months without a single number looking wrong.
What it takes to catch it
One fact that is easy to throw away: which regions the targeting actually resolved to. The targeting spec holds numeric city keys and a radius; the delivery report holds region names. They are not the same vocabulary, so nothing compares them unless the resolved names were captured at launch and kept.
With that list stored, the check is trivial: every region with material spend either appears in it or does not. Without it, no amount of looking at the delivery breakdown will tell you, because the breakdown was never the problem.
Where the bad leads were coming from
Out-of-area delivery does not usually show up as a delivery problem. It shows up as a lead quality problem, one step downstream, and gets blamed on the creative or the form.
The mechanism is ordinary. People forty kilometres outside the catchment are cheaper to reach, because everyone else targeting properly is not bidding for them. They fill in the form. They are not going to visit the site. So cost per lead improves, the sales team starts saying the leads are rubbish, and both things are true at once.
That is the same shape as a cost per lead that falls for the wrong reason, and geography is one of the commonest causes.
What to do about it
- Read the region breakdown against what was actually targeted, not against what looks plausible. A familiar-sounding state name is not evidence it was in the brief.
- Tighten the radius before excluding regions. Excluding is whack-a-mole; the radius is the cause.
- Where the catchment is genuinely small, pin to the smallest geography Meta will accept rather than a city plus radius.
- Check whether the out-of-area share tracks with the lead quality complaints. It usually does, and that is the argument that gets the change approved.