The word covers three different things
Vendors use "geofencing" for any location-based targeting, which hides real differences in capability and price.
The first type is simple geographic targeting: show this ad to devices in this county, city, or ZIP code. Every platform does it, it costs nothing extra, and it is what most local campaigns actually need.
The second is radius or polygon targeting: draw a shape on a map — a five-mile ring, or a custom boundary — and target inside it. Slightly more precise, still standard, still not exotic.
The third is the one worth the name: targeting devices that entered a specific small location during a specific window, and reaching them afterwards. That is the capability people are picturing when they hear geofencing, and it is the only one where the price premium is sometimes justified.
What the third type is genuinely good for
A short list, and it is worth knowing because outside it the premium is wasted.
Competitor locations. Reach people who visited a competing dealership, gym, or clinic in the last thirty days. Legal, widely done, and effective in considered-purchase categories.
Events and venues. Trade shows, fairs, stadiums, conferences — reach attendees during and after. Useful when your customers cluster somewhere for a day.
Places that signal a life event. Furniture stores and moving-truck rentals for home-services businesses; hospitals and clinics for certain medical categories.
Outside those cases, it is an expensive way to do what plain geographic targeting does for free.
The setting that wastes the most money
Every major platform has an option controlling whether you reach people physically in a location or people merely interested in it. The default is frequently the second one.
For a local service business this is catastrophic and invisible. You are paying to reach someone in another state who searched about your city once. Delivery looks healthy, clicks arrive, the leads are unusable, and the campaign gets blamed for underperforming.
Set it to physical presence. Then verify by checking the delivery-by-location report a week in, because the setting and the actual delivery are not always the same thing.
Radius sizing, done properly
Two errors, opposite directions. Too tight, and you starve the campaign of volume — platforms need audience size to optimize, and a three-mile ring in a rural county may not have enough people in it to learn from.
Too wide, and you buy attention you cannot serve. If you will not drive forty minutes for a job, do not pay to advertise forty minutes out.
The right radius is your actual service area, drawn by drive time rather than distance, and centered on where your customers are rather than where your office is. Those are different points more often than you would expect — especially for businesses on the edge of a metro.
The privacy question, answered honestly
Location targeting relies on signals from apps and devices, and that supply has become less precise as platforms tightened permissions. Coverage varies by app mix and by market, which means precision claims should be treated as approximate rather than exact.
Anyone selling you device-level certainty is overselling. The honest version: geographic and radius targeting are reliable; visit-based targeting is directional and works better in dense areas with heavy app usage than in rural ones.
Plan accordingly, and do not build a strategy that only works if the precision claims are perfect.
What to actually do
Use plain geographic targeting by county or ZIP for the bulk of your budget. Set physical presence, verify the map, and match the boundary to your real service area.
Test visit-based targeting only for the specific cases above, with a small share of budget and a clear read on cost per booked job.
And spend the money you saved on better creative. Precise targeting of a weak offer is still a weak campaign.