What Is Geofence Advertising? How It Works, Platforms, and Accuracy

How a geofence works

A geofence is a closed polygon, usually a circle of some radius, defined by latitude and longitude coordinates. A device's position is matched against that polygon in real time. When the position crosses the boundary, an event fires: an ad bid is placed, an in-app push is sent, a mobile-ad-ID is logged for later retargeting, or a location-attribution record is written.

The position itself is resolved by one or more of three signals running on the device or inferred from the network it's attached to:

• GPS. The phone's GNSS chip resolves its position from satellite signals. It is at its best outdoors with a clear view of the sky and degrades where tall buildings reflect signals before they arrive. A peer-reviewed PLOS One field study measured 7 to 13 metres of iPhone horizontal error in a dense urban setting, which is the only accuracy figure on this page with a primary source behind it.
• Wi-Fi positioning. The phone reports nearby Wi-Fi access point identifiers to a lookup service that maps them to known coordinates. It is the indoor and dense-urban workhorse, and it is tighter than cellular. Published accuracy ranges for it come from vendor blogs rather than measurement, so we do not quote one.
• Cellular triangulation. The carrier estimates position from which towers the device is attached to and from timing-advance values. It is the loosest of the three by a wide margin and the fallback of last resort. Again, the metre figures that circulate are vendor claims, not measurement.

iPhone horizontal accuracy, dense urban setting

7 to 13 m

Peer-reviewed field measurement. Vendor-blog accuracy ranges for Wi-Fi and cellular were removed rather than repeated

Smartphone GPS accuracy in an urban environment, PLOS One

Common geofence radii

Radius is a tradeoff between false-positive rate (people captured who weren't actually in the place) and false-negative rate (people missed who were). The right number depends on the use case and on how clean the underlying location signal is. The radii below are configuration conventions, not measured accuracy: the 100 to 150 metre figure is what Android's own developer documentation recommends for reliable geofence triggering, and the rest are common practice rather than published standards.

The major geofence advertising platforms

Geofencing isn't a single product. It's a feature inside a layer of demand-side platforms (DSPs) and location-data specialists. The tradeoffs are entry cost, fence precision, and what happens to the location data after the campaign ends.

Pros and cons of geofence advertising

Operator-honest read on what you actually buy when you buy a geofence campaign.

• Pro: physical-world targeting. A geofence around a stadium, a competitor, or a service-area neighborhood is a more behaviorally-revealing signal than most demographic segments.
• Pro: measurable visit attribution. Platforms like GroundTruth and Foursquare can tie an ad exposure to a later store visit, closing a loop that pure display can't.
• Pro: works for verticals where individual targeting is restricted. Healthcare, pharma, and political advertising can use location instead of identity to reach the right room without targeting a person.
• Con: accuracy degradation. Position drifts in urban canyons, where the peer-reviewed measurement is 7 to 13 metres of iPhone horizontal error. Smaller radii catch fewer real visitors; larger radii catch more non-visitors. There is no setting that fixes both.
• Con: opt-out and permission cliffs. Apple's App Tracking Transparency cut the share of iOS devices carrying a usable advertising identifier, and devices that opt out are largely invisible to attribution. The opt-in percentages that circulate come from vendor panels with shifting bases, so we do not quote them. The best-measured evidence that people actively refuse interruptive advertising is DoubleVerify's 2025 finding that 41% of North American consumers use ad blockers.
• Con: bid-stream latency and intermediary rake. A typical DSP geofence campaign passes through an exchange, an SSP, and a DSP before the bid lands. Every additional hop is a layer of latency and a percentage shave that doesn't reach inventory.
• Con: privacy regulation risk. CCPA and GDPR treat location data tied to an identifier as personal data. Sensitive locations (clinics, schools, government buildings) carry their own per-jurisdiction restrictions and have already produced enforcement actions.

Industries that use geofence advertising the most

Adoption clusters in verticals where physical location is the buying signal: either because the customer has to physically arrive (retail, QSR, auto) or because the customer's room is already legally protected from individual targeting (healthcare). We used to print a per-vertical CPM against each of these. Every one of those figures traced to an agency blog rather than a platform, an industry body or a methodology-published benchmark, so they are gone. Nobody at our sourcing bar publishes a geofencing CPM at all, by vertical or in total.

• Retail and QSR. Competitor conquesting, loyalty re-engagement, aisle-level promos, and store-visit attribution.
• Automotive dealerships. Conquest targeting at competing dealers, service-reminder retargeting, financing offers, usually layered with competitor fences.
• Real estate. Listing-proximity, neighborhood, and high-value area targeting.
• Healthcare and pharma. Location instead of identity: clinics, hospitals, conferences, with compliance tooling around the targeting.
• Local home services. HVAC, roofing, plumbing, landscaping. Targeting residential neighborhoods around recent storm activity, recent installs, or competitor service trucks.
• Events, entertainment, campuses and bases. Burst frequency around venues on high-volume formats.

Privacy compliance: CCPA, GDPR, and the location ladder

Location data, when joined to a device identifier, is personal data under both GDPR and CCPA. The compliance baseline for any operator running geofence campaigns is affirmative opt-in consent for collection, transparent disclosure of why the data is collected, the ability for a user to access or delete their location records, and minimization of retention.

Apple's App Tracking Transparency framework adds a hard ceiling on how many iOS devices are addressable at all. The opt-in percentages quoted around this vary wildly by data source, vertical and prompt design, and none of the published panels clears our sourcing bar, so we quote no figure. The direction is not in dispute: a large share of iOS users decline, and those devices are largely invisible to attribution. On the wider question of whether people want interruptive advertising, the best-measured signal is DoubleVerify's 2025 finding that 41% of North American consumers use ad blockers.

Sensitive-location targeting is its own compliance surface. Geofencing women's reproductive healthcare facilities led to a 2017 Massachusetts attorney general settlement; similar restrictions cover schools, places of worship, and government buildings in multiple jurisdictions. The cost of getting this wrong is regulatory, not just reputational.

How CPVD differs from impression-based geofencing

Standard geofence advertising bills CPM, a cost per thousand impressions inside a fence, at a rate nobody publishes. The advertiser pays for inferred reach: the platform's bid stream said a device carrying a usable identifier was inside the radius. Accuracy drift, tracking opt-out and the intermediary fee stack all sit between the budget and the actual human. On that last one there is a real measurement to lean on: the ANA's Q3 2025 benchmark, on web and mobile excluding connected TV and the last quarter the ANA itemized that cut, puts transaction costs at 26.9% of the ad dollar, media-productivity loss at 26.1% and benchmark-qualified working media at 47.1%. On the ANA's current all-environments basis the figure is 45.1% for Q2 2026, with transaction costs at 27.2%.

Cost Per Verified Delivery (CPVD) replaces the inference layer. WilDi Maps charges from $0.25 per delivery to an opted-in person's phone as they drive through a corridor you own, with the rate set by the account's monthly deposit tier: $0.25 at Enterprise ($3,000 a month and up), $0.32 at Pro ($1,000 and up), $0.40 at Local ($250 and up), $0.50 at Starter. Tunnels and zones cost more than background. Each delivery is GPS-verified at the device, in an app WilDi controls, with no exchange, SSP or DSP in the middle. This is not a claim to buy more impressions per dollar; a delivery is a different unit, and it is neither a lead nor an impression.

For the side-by-side breakdown of CPVD against CPM, CPC, and CPA, see What is Cost Per Verified Delivery? For the specific case of billboard mobile retargeting accuracy, see How accurate is geofencing tied to a billboard?