How Does Geofence Advertising Work?

There are three common ways to use it. Geofencing in the narrow sense reaches devices inside the fence now. Geo-conquesting draws fences around competitors’ locations to reach their customers. Geo-retargeting builds an audience from devices seen inside a fence, such as visitors to a trade show, and shows them ads afterwards, wherever they are.

How does a platform know where a phone is?

It reads location signals attached to the ad request: GPS from apps the user has allowed to use location, Wi-Fi and cell data, and the device’s IP address. Their accuracy varies widely.

Precise location permission: Android describes it as usually within about 50 meters (160 feet), sometimes better.
Approximate permission: accurate only to an area of about three square kilometers, so that phone can’t be placed inside a building-sized fence at all.
IP address: the Media Rating Council’s location measurement guidelines warn that IP-based location can differ significantly from where a device actually is, and say it should be avoided for matching a device to a place.
Height: fences are flat. The same guidelines note they can’t tell one floor of a multi-story building from another.

So the fence on the map is usually drawn more precisely than the data can honor. A fence around one shop on a busy street will also catch people walking past and the café next door, and will miss some visitors whose phones only share an approximate location.

What can Google and Meta actually do?

Less than many pitches suggest: neither offers building-sized fences.

Google Ads: radius targeting requires at least 1 km, there are no custom shapes, and Google warns that ads aimed at a very small radius “might only show intermittently or not at all”. Google’s own programmatic platform, Display & Video 360, has the same 1 km floor.
Meta (Facebook and Instagram): targets a radius around a dropped pin. Its location setting now covers people living in or recently in an area, so it can’t be limited to the devices that are inside the fence right now. Check the current minimum radius in Ads Manager before planning around it.
Programmatic platforms (DSPs): this is where tight fences live. Platforms such as Simpli.fi and GroundTruth let you draw custom shapes that follow a building outline, a car park or a venue.

If an agency promises to fence a single building, ask which platform it means. If the answer is Google or Meta, the fence will be far larger than described.

Where is geofencing restricted?

Around healthcare facilities in at least five states, and increasingly in the trade in precise location data itself. This is a summary for planning, not legal advice; check with counsel before running campaigns near sensitive locations.

Near healthcare facilities

Washington: no geofence within 2,000 feet of an in-person healthcare provider to identify or track people seeking care, collect their health data, or send them health-related ads (RCW 19.373.080). Consumers can sue.
New York: no geofence within 1,850 feet of someone else’s healthcare facility to deliver ads, build profiles or infer anyone’s health status (General Business Law § 394-g).
Connecticut: no geofence within 1,750 feet of a mental health or reproductive or sexual health facility for tracking, data collection or notifications.
Nevada: no geofence within 1,750 feet of a medical facility or other in-person health provider for tracking, data collection or health-related ads, even with consent.
California: since 1 January 2026, AB 45 bans geofencing healthcare facilities to track or identify people, collect their information, or send them targeted ads.

These rules don’t only affect health businesses. A fence around a competitor, a shopping street or a neighborhood can easily take in a clinic, so check what is inside every fence before launch.

The location data itself

Sensitive data: precise geolocation is treated as sensitive data under every comprehensive state privacy law in Foley & Lardner’s 2026 comparison chart, which covers 20 states, generally with stricter consent rules.
Sale bans: five states now ban selling precise location data outright, whatever the consumer agreed to: Maryland, Oregon, Virginia, Connecticut and New Jersey, most taking effect in 2026 (Venable’s mid-2026 update).
FTC orders: the Federal Trade Commission has barred several location-data brokers, including Gravy Analytics and Mobilewalla, from selling data about visits to sensitive locations: medical facilities, places of worship, schools and childcare, shelters, union offices and military sites among them. The Mobilewalla order also stops it collecting location data from ad auctions for any other purpose, which is where much low-cost geofencing data comes from.

How do you measure geofence results?

Mostly with modeled estimates, so treat visit counts as directional and insist on a comparison group.

Google’s store visits are anonymous, aggregated figures extrapolated from users who opted into location history. Smaller advertisers may not pass Google’s privacy thresholds and get no visit data at all.
Programmatic platforms usually draw a second “conversion zone” around your own premises and count served devices that later enter it.
A control group decides whether any of it means anything: devices that matched the fence but weren’t shown the ad. Without one, a campaign aimed at people already near your store takes credit for visits that would have happened anyway.
The definition of a visit matters too: how long a device has to stay, and how passers-by are filtered out.

Tie visit figures back to sales wherever you can. Marketing analytics is how a modeled visit becomes a number the finance team will accept.

What should you ask a geofencing agency?

Ask what the data is, where it comes from, how small the fence really is, and how they will prove the ads caused anything.

Which platform will run the campaign, and what is its smallest workable fence?
Where does the location data come from, app GPS with consent or IP and auction data, and roughly what share is each?
How do you keep fences away from healthcare and other sensitive locations, and which state laws have you checked?
How is a visit defined, and will results be measured against a control group?
What is the all-in cost: price per thousand impressions, platform fees, management fee, minimum spend and minimum term?
Can I see the fences and the reporting in my own account?

Published geofencing prices come almost entirely from vendors and resellers rather than independent benchmarks, so compare quotes on the full cost, not the headline rate.

Questions

What is the difference between geofencing and geotargeting?

Geotargeting reaches people by area, such as a city, a postcode or a radius, usually including where they live. Geofencing draws a tighter boundary around a specific place and aims at who is, or was, physically there.

What is geo-conquesting?

Drawing fences around competitors’ locations to reach their customers with your offer. It is generally allowed, but if the competitor is a healthcare provider, the state rules above can apply.

Is geofence advertising legal?

Generally yes, with growing limits: bans near healthcare facilities in Washington, New York, Connecticut, Nevada and California, stricter treatment of precise location under state privacy laws, and outright bans on selling it in five states. For regulated categories, check with counsel.

How small can a geofence be?

On Google Ads the minimum is a 1 km radius. Programmatic platforms can follow a building outline, but phone location is often only accurate to around 50 meters at best, so very small fences catch neighbors and miss some visitors.

How much does geofence advertising cost?

It is mostly bought per thousand impressions, sometimes per visit. Published figures come from vendors rather than independent sources: GroundTruth, a vendor, quoted $3.50 to $15 per thousand impressions for mobile and desktop in February 2024. Ask for the all-in cost, including fees and minimums.

How does geofence advertising work?