Geofencing advertising, explained: the fence, the data behind it and the law around it

Geofencing advertising serves ads to phones observed inside a drawn perimeter, using location the device has reported through an app or an ad request. The fence itself is small, a coordinate and a radius. The audience is a set of mobile advertising identifiers seen inside it within a lookback window, and four states now ban fences around health facilities.

Geofencing is sold as a targeting method and is really a data supply chain with a shape drawn on top. The shape is trivial. What matters is where the location observations come from, how long they are kept, which laws now apply to them, and what happens when the thing being fenced is a screen rather than a store. The comparison with geotargeting, proximity and out-of-home by place is on location-based advertising and is not repeated here.

What is geofencing advertising?

Geofencing advertising serves ads to devices that have been observed inside a defined perimeter. GroundTruth, one of the larger vendors, defines it as “the application of a virtual boundary around a location”, drawn as a radius or as a polygon following store walls, car parks or a shopping centre. Two different things share the name. Inside an app you own, the phone’s operating system watches the fence and tells the app when the device crosses it. In advertising, nobody’s phone is watching anything: a location data supplier matches the devices it has already observed inside the shape, and those devices become an audience bought through an ad platform. The first is a feature of the phone. The second is a purchase of other people’s data.

Geofencing marketing vs geofencing advertising

Geofencing marketing is the wider term and includes owned channels: a push notification when a customer with your app walks into your store, a reminder when they leave an airport, an in-app offer inside a stadium. It runs on the operating-system fence and on permission the person granted you. Geofencing advertising is the paid part: reaching people who have never installed anything of yours, by buying audiences built from location data collected elsewhere. The techniques overlap, the data supply does not. Marketing uses your permission; advertising uses somebody else’s.

What is a geofence, technically?

A geofence is a circle. Android’s developer documentation says it plainly: “the latitude, longitude, and radius define a geofence, creating a circular area, or fence, around the location of interest”, and the system raises an event when a device enters it, exits it, or dwells inside it. Apple calls the same thing region monitoring, “a circular area centered on a geographic coordinate”. Both platforms cap it, gate it behind permission, and publish accuracy figures that are worth reading before anyone promises a fence around a single shopfront.

Two lines of the table decide what the phone’s own fence can do for an advertiser. The limits, 100 fences per app on Android and 20 per app on iOS, mean a product promising thousands of fences is not using the device’s fence at all. And the permissions mean the fence only exists for people who said yes: on Android a background location grant on top of fine location, on iOS an Always grant if the app is to be woken while closed. Everything an ad platform does starts from the fact that most of the people it wants to reach never granted either.

How do ad platforms geofence people who never installed the app?

By working backwards from data. The fence becomes a query run against a store of location observations: every device seen inside the polygon during a window is pulled into an audience, keyed by its mobile advertising identifier (MAID), and served ads wherever that identifier turns up next. OUTFRONT’s May 2024 measurement explainer describes the same mechanism from the out-of-home side: a phone passing through a geofenced area pings its device ID with coordinates, and measurement partners assign the MAID to mark the device present without sharing personally identifiable information.

The observations come from four places, and the OAAA’s 2020 guide grades each on accuracy, precision and scale, in its own words:

The two that feed most advertising products are the middle rows. Software development kits inside third-party apps report GPS fixes from people who opted in to those apps, not to the advertiser. Bid-stream data is the location an app attaches to an ad request, which the OAAA rates poor on accuracy and which, the FTC alleged in December 2024, Mobilewalla retained from auctions it did not even win, amassing more than 500 million unique advertising identifiers paired with precise location. Placer.ai describes a panel of “tens of millions of devices” and machine learning to estimate foot traffic from it. Every geofenced audience is a sample, of unknown coverage, of the people who were actually there.

Three mechanics follow from that data model. A lookback window sets how long an observation keeps a device in the audience; Simpli.fi lets advertisers engage devices “for up to 30 days” after they leave a fenced address. Conquesting fences a competitor: in Simpli.fi’s fast-food case study, 4,770 fences were drawn around rival outlets and devices seen there were served ads for up to 30 days. A conversion zone is a second fence around the advertiser’s own locations, so exposed devices that later enter are counted as visits; that study drew 353 of them. Simpli.fi also converts street addresses to fences from plat-line data, up to one million addresses per campaign. None of this touches the phone’s own fence. The perimeter is the least important input.

What does the law say about geofencing?

Two layers, both recent. Federally, the FTC has moved against the data supply rather than the technique. On January 9, 2024 it announced its first settlement with a data broker over sensitive location data: X-Mode Social and its successor Outlogic, which collected precise location through an SDK in other companies’ apps and from other brokers, were prohibited from sharing or selling sensitive location data, required to keep a list of sensitive locations, and ordered to delete data previously collected. On December 3, 2024 the Mobilewalla order added the bid stream: the company is barred from collecting or retaining consumer data from ad auctions for any purpose other than the auction itself, the first time the FTC had alleged that practice was unlawful. The raw material of many geofencing products is exactly what those orders restrict.

At state level, four statutes now make it unlawful to fence health facilities for advertising or tracking. Each was read from the legislature’s own text.

The radii differ by a few hundred feet and the covered facilities differ in scope: Washington and Nevada reach any in-person health care provider, Connecticut names mental health and reproductive or sexual health facilities, New York covers any health care facility other than the advertiser’s own. A national campaign that fences pharmacies, gyms with clinics attached, or a shopping centre with an urgent-care tenant has to be checked against all four. This describes the rules; it is not legal advice.

How is geofencing used for out-of-home?

Out-of-home does not need a geofence to reach anyone; a screen reaches whoever passes it. Geofencing enters out-of-home through measurement, as the mechanism that turns an unlogged medium into an exposed panel. The fence is drawn around the screen’s viewshed, the area from which it can be seen, and devices observed inside it while the campaign runs become the exposed group a brand lift study surveys or an incrementality test compares with a control. The guide to out-of-home advertising covers where that panel sits in the wider buy.

Fencing a screen is not fencing a store

The two fences point in opposite directions. A fence around a store defines the outcome: a device that enters it after exposure is a visit. A fence around a screen defines the exposure: a device inside the viewshed while the ad plays is exposed. A store fence can be drawn from a property boundary; a screen fence has to be drawn from sightlines, and the OAAA’s definition of exposure, presence in the zone while content is viewable, is explicit that it does not require the content to have been seen. A full attribution study uses both fences and a lookback window between them, as the Lamar and Clear Channel studies on the incrementality page do. How the audience estimate behind the screen is built in the first place is on how out-of-home impressions are measured.

Why a moving screen is a moving geofence

A fixed screen’s viewshed is a polygon drawn once and valid for the flight. A screen on a vehicle has a viewshed that travels with it, so the fence is not a shape but a shape per timestamp, and the exposed panel is the union of every device logged inside it along the route. Three things change structurally. The exposure record becomes a route log joined to a location panel, which is why Reveal Mobile lists “moving waypoints” as an input to its exposure computation alongside media plans and ad rolls. Frequency stops being a property of a site and becomes a property of route repetition: the same unit on the same corridor at the same hour builds frequency, a rotated route spreads reach. And the fence inherits the vehicle’s timing, so an exposed panel built from a morning commute and one built from an evening route are different populations under the same impression total. The IAB’s July 2025 guide is direct about the cost: static placements are “more compatible with geofencing and camera-based tools”, while exposure for exterior moving formats “is highly variable depending on vehicle speed, obstructions, and route patterns”. That is the trade. The moving fence can be steered to where the audience is; the record of who it reached is only as good as the route data behind it. Firefly, whose taxi-top and rideshare-top studies are the operator-published examples on the brand lift and incrementality pages, lists geofencing and dayparting as targeting inputs for its digital tops and publishes foot traffic, web and app conversion studies on its measurement page; the GPS log its tops carry is the route data that sentence is about.

What does geofencing advertising cost?

Vendors publish more about geofencing prices than about most media, because the buy is programmatic display priced per thousand impressions or per measured visit. The figures below are the ones this desk could read on the publisher’s own page.

The ranges below are market-level figures compiled from publicly published vendor and industry sources. They are not quotes, and they are not the rates of any single network, including the operator whose campaigns the archive on this site documents, which prices per campaign and does not publish a rate card. Always get a written quote before budgeting.

How it is priced matters more than the numbers. Two models exist: CPM, where the advertiser pays for impressions served to the fenced audience wherever they appear, and cost per visit, where GroundTruth says marketers “only pay for ads that result in a store visit”, which shifts the risk of a thin audience to the vendor. Fence count is usually not a line item; Propellant Media does not charge per fence and instead recommends a fence-to-budget ratio. Minimums are where vendors differ most, from four figures a month at small agencies to five at the larger platforms according to the same page. The undated pages will drift; treat them as the shape of the market rather than its level.

Where to go next

The four location techniques side by side, and when presence means anything, are on location-based advertising. What the exposed panel is used for once it exists is on the brand lift study and incrementality testing pages. Buying the inventory whose fence moves, with time and weather conditions, is on programmatic DOOH.