Recruitment marketing·Targeting

Geofencing recruiting: when it works, and when it quietly gets expensive

Geofencing recruiting is sold as precision targeting and priced as a premium service. It is genuinely useful for about four hiring problems and a poor trade for most of the rest — and the reason is arithmetic about audience size, not doubt about the technology.

What the word actually covers, the four cases where a fence beats a radius, the frequency maths that decides the cost, and the targeting question worth asking a vendor before signing.

What geofencing recruiting actually is

Geofencing draws a boundary on a map and serves ads to devices inside it. Geotargeting, the term it gets used interchangeably with, serves ads to people whose location profile puts them in an area. The distinction matters commercially: a true geofence is small and event-driven — this building, this shift change, this trade show — while geotargeting is a radius or a postcode list and is what almost every platform actually sells.

In recruiting, three different products get sold under the word.

• Radius targeting on social platforms. Meta, TikTok and similar let you target a radius around a point, typically from about one mile upward. This is geotargeting, it is what most frontline recruitment campaigns use, and it is priced in the normal auction.
• Programmatic mobile display with a drawn polygon. Vendors draw a fence around a competitor’s plant, a hospital campus or a truck stop and serve banner ads to devices seen inside it, often with a retargeting window of 30 days. This is what "geofencing recruiting" usually means when an agency sells it as a distinct service.
• Location-history audience building. Building an audience from devices that visited a place in the past, then advertising to them elsewhere. This is the one most affected by the privacy changes below.

Where it works

Geofencing earns its premium in a narrow set of cases, and they share one feature: the location is a better proxy for the qualification than any interest or demographic signal available.

Credentialed workers who cluster in buildings. Nurses, techs and therapists are physically concentrated in hospitals in a way that no interest targeting reproduces. A fence around three hospital campuses is a genuinely better filter than "interested in nursing".

Drivers at fixed points. Truck stops, terminals, weigh stations and distribution yards concentrate CDL holders. In a role family where our 2026 benchmark data puts the median cost per applicant at $26.86 — the most expensive frontline family we measure — a targeting method that improves qualification has real headroom.

Events with a start and an end. Trade shows, job fairs, graduations from a training program, union halls on meeting nights. The fence has a natural boundary in both space and time, which is the condition geofencing was designed for.

A very tight commute requirement. Where a job genuinely cannot be done by someone more than fifteen minutes away — an early shift with no transit, a site with no parking — a small radius is not a gimmick, it is the requirement.

Where it does not

The failure mode is almost always the same, and it is arithmetic rather than technology.

Small fences make small audiences, and small audiences get expensive fast. This is the part vendors do not put in the deck. Across 1,334 campaign-months in our benchmark data, the median cost per applicant at a monthly frequency under 1.5 was $7.85. Above a frequency of 4.0 it was $26.78 — the same job, the same platform, three and a half times the price, purely because the audience was too small to avoid showing the same person the same ad repeatedly. A fence around one competitor’s plant is, by construction, a tiny audience. It will deliver, and then it will get expensive in about two weeks.

The devices in the fence are not all the people you want. A hospital fence catches patients, visitors, contractors, delivery drivers and the coffee shop across the road. A plant fence catches everyone on the road beside it. Polygon precision is sold as though it were population precision, and it is not.

Banner inventory converts worse than social. Most programmatic geofencing runs on mobile display, where the click is cheap and the applicant is not. Judge any geofencing proposal on cost per applicant, never on impressions, clicks or "devices reached" — and ask for the apply rate. In our data the apply rate explains roughly 70% of the variation in what an applicant costs.

Privacy changes have thinned the underlying data. Since Apple’s App Tracking Transparency prompt and the tightening of location data brokerage, the pool of devices sharing precise background location is a fraction of what it was when geofencing recruiting decks were first written. Vendors quoting reach figures should be asked what share of devices in the fence they can actually observe, and over what window.

The question nobody asks the vendor

Fencing a location is targeting a population, and some locations are proxies for protected characteristics. A fence around a place of worship, an immigration services office, a specific school, a neighbourhood chosen by demographic composition, or a facility associated with a particular community is a targeting decision that a regulator, a plaintiff or a journalist can read as something other than geography — regardless of intent. Ad platforms have already restricted location-based targeting granularity for housing, employment and credit advertising for exactly this reason.

The safe version of the practice fences occupational locations — hospitals, terminals, plants, training centers, trade shows — and documents why each fence was chosen. If a proposed fence needs a paragraph of explanation, do not run it.

What to do instead, most of the time

For the large majority of frontline hiring, a radius on social gets the same result for less money and less risk. A five to fifteen mile radius around the work site, with an audience large enough to hold frequency under about two impressions a month, is the configuration our data supports. The levers that actually move cost sit downstream of targeting anyway: apply rate explains about 70% of the variation in cost per applicant, creative click-through about 30%, and the auction itself around 18%.

Where geofencing is worth buying, buy it as a supplement to a radius campaign rather than as a replacement, cap it at a small share of the budget, run it against a hard cost-per-applicant target, and give it a fixed window. Treated that way it is a useful instrument for a few specific problems. Treated as a strategy, it is a small audience with a premium attached.

Frequently asked questions

Sources

Boostpoint 2026 Social Job Advertising Benchmark: 891 campaigns, 1,334 campaign-months, 15 role families. Frequency bands, apply-rate bands and role-family cost per applicant.