Proximity, Not Proof
Every pitch for geofencing marketing follows the same arc. The vendor draws a circle on a map, names the radius, and explains that anyone inside it will see your ad. The circle feels precise. The precision feels like control. What the pitch rarely addresses is the question that actually matters: what happens after the person sees the ad, and how will anyone know?
That question does not have a tidy answer. This article does not pretend otherwise. What it has is one campaign we can describe, a set of observations from it, and a framework for deciding whether the tactic is worth running at all. The observations are specific and the framework is honest about what it cannot reach.
The short version: geofencing buys proximity. Proximity is not proof, and it is not free.
What geofencing is sold as, and the question nobody answers
Geofencing marketing is location-based advertising that triggers delivery when a device enters a defined geographic boundary. In practice, most campaigns on social platforms are not true geofences in the technical sense. They are radius or zip-code targeting: the platform delivers to devices it believes are in or near a set of locations, based on the signals users have shared with the app.
The distinction matters less than the vendors suggest. Whether the boundary is drawn by radius or by polygon, the same question applies: what does being inside it actually predict?
What the category sells
The standard pitch has three parts, and all three are true in a limited sense:
Precision: you reach people near the store, not people across the country.
Intent: people near a store are more likely to be considering a category purchase than people who are not.
Efficiency: you do not waste budget on audiences who cannot act on the message.
None of these claims are false. None of them answer the question a marketing lead should be asking before signing off on the spend.
The question nobody answers
The question is this: if someone sees your ad and walks into the store, how will you know?
Not in aggregate, not as an estimate, not as a model output. How will the specific chain of events, from impression to threshold, be visible to your team in a way that holds up to scrutiny?
Every vendor page we reviewed for this article explains how to draw the fence. Salesforce, GroundTruth, Mailchimp, Adjust, Placer.ai, Azira and SafeGraph each have a guide to the mechanics. Not one of them opens with the measurement gap. They close with it, briefly, if at all.
That gap is where this article starts.
What tight targeting actually costs
We call this the Proximity Premium, and it is Octonano's term for a trade that nobody in the category prices honestly.
On one campaign we can describe, spend was essentially flat month over month. The account ran paid social on Meta, targeted to zip codes derived from a set of store locations, supporting a capsule collection over roughly four to six weeks. When we tightened delivery to those store catchments, here is what we observed:
Direction | What moved |
Down | Reach, by roughly a sixth |
Down | Impressions, by roughly an eighth |
Up | Cost per thousand, by roughly a ninth |
Up | Landing page views, by roughly a sixth |
Up | Leads, by roughly a quarter |
Up | Sessions, by roughly an eighth |
The trade in one sentence: we reached fewer people, paid more to reach each one, and more of them acted.
What we think explains it, and why we are calling it a hypothesis
The observation is from one account in one month, and it is real. The explanation is our reading of that data, not an independently verified finding.
Our hypothesis: when budget is split across a set of store-catchment audiences rather than delivered broadly, the platform has a smaller pool of users to optimize against in each set. With fewer signals to learn from in each geography, delivery becomes less efficient, and cost per thousand rises. The action rate rises because the remaining audience is more contextually relevant.
We have not found a neutral published source that confirms this mechanism for Meta specifically. The logic is consistent with how delivery concentration works in algorithmic ad systems, but we are describing what we believe happened, not what we can prove happened. Any vendor who tells you the premium is a known, predictable quantity is overstating the evidence.
What the premium is not
It is not a mistake. It is not a sign the campaign was set up badly. It is a structural consequence of choosing precision over reach, and it can still be the right call. A smaller, more relevant audience that acts at a higher rate can outperform a broad audience that mostly ignores the message.
The problem is not the trade. The problem is that the trade is never named in the pitch.
What it moved
The most honest answer to "does geofencing work?" is: it depends entirely on what you are measuring.
On the campaign we can describe, something real happened on the demand side. Where-to-buy and stockist-page behavior rose by more than half in the same month as the store-catchment campaign, against a shop page running several times larger. The brand's own read at the time was specific: people were no longer only browsing the collection. They were looking for where to find it.
That is a meaningful signal. It is also the ceiling of what we can claim.
Store intent is the measurable outcome
Location-targeted advertising, when it works, moves people from awareness to active consideration of where to purchase. That is not a small thing. For a brand with physical distribution, converting a browser into someone who is actively seeking a stockist is a real commercial step.
The campaigns we have run around brands' own stores, around openings, and around seasonal events in New York have consistently produced this kind of directional movement. People engage differently when the message is tied to a place they can actually reach.
Store intent is as far as the evidence goes. What happens at the till is a separate question, and it requires a separate answer, one that has to be built before the campaign launches, not after.
Where it stopped: the retailer's floor
This is the part of the article that matters most, and it is the part the category never publishes.
We have never been able to connect a location-targeted campaign to a sale at a retailer we do not own. Not once, including on the campaign built around that retailer's own zip codes. Clicks were visible. Click-through was visible. The till was not.
The stated cause is not a technical mystery. There was no point-of-sale connection between our campaign and the retailer's system. Codes and similar mechanisms that would have created a traceable link were proposed and not activated. The gap was a decision, taken before the campaign ran, by people who did not expect to be asked the question afterward.
Why the chain breaks for wholesale brands
For a brand selling through someone else's floor, the measurement problem is structural. The retailer owns the transaction data. The brand owns the ad data. Unless one party shares with the other, those two datasets never meet.
Most wholesale relationships contain no such arrangement. It is not that retailers are unwilling in principle. It is that the question rarely gets asked at the right moment, which is before the campaign brief is written, not after the results are due.
There is a second layer to this. A geo-targeted push around a brand's own store ran alongside offline activity in one period, and the measurable response traced back to email rather than to the geo-targeted ads specifically. We are using this because it is the strongest counter-evidence we have, and it is ours. It does not mean location targeting failed. It means that when multiple channels run together, attribution without a closing mechanism goes to the channel that already has one.
The lesson is not that the tactic is wrong. The lesson is that the tactic is unmeasurable by design if the measurement architecture is not agreed before it runs. Everyone involved should know that going in, not discover it when the results deck is due.
Three conditions worth checking first
Before any location-targeted campaign runs, three conditions need to be true. If all three are not, the Proximity Premium is being paid for nothing. This is Octonano's model, and we call it the Threshold Test.
Condition 1: Is the product actually there?
Not in the chain. In those specific doors, in the right sizes, this week. The clearest result we have seen from store-catchment targeting came where the collection was genuinely stocked in the targeted markets. The signal dissipated where it was not.
Targeting a zip code where the product is unavailable does not create demand. It creates frustration. Confirm the inventory position before the campaign brief is written, not after the first week of delivery.
Condition 2: Is there a reason to go now?
An opening, a capsule, an event, an offer, something ending. Proximity is a context, never a reason. An ad that says "we are near you" without answering "and here is why today matters" is asking someone to make a trip on no particular grounds.
The openings and activations that give someone a reason to turn up are the events that make location targeting coherent. Without a moment, the fence is just geography.
Condition 3: Does anything travel with the person?
A code, a reservation, a dedicated page, a retailer who reports back. Without one of these, the campaign is unmeasurable by design, and everyone should agree on that before it runs, not after.
This is the Closing Mechanism Test applied to one tactic. For a fuller treatment of what a brand can see between a post and a purchase, and why that question has to be answered before the campaign goes live, see our earlier piece on what a brand can see between a post and a purchase.
If all three conditions are true, the Proximity Premium is a reasonable price to pay for a tighter, more relevant audience. If any one of them is not, tighten the fence and you are still paying more for a campaign that cannot close.
Location is the context, not the idea
The misconception the whole category sells is that targeting precision is the hard part. It is not. Drawing the fence is trivial and cheap. The creative is where the campaign is won or lost, and it is the part that gets the least attention in a geofencing brief.
Location tells the platform where to deliver. It tells the person nothing.
What actually worked in the creative
Across the accounts we have run, the clearest pattern in the creative is this: location is the context, not the creative idea. The ad still has to give someone a reason to act. Knowing they are near a store does not supply that reason.
What worked was product clarity. The item legible at a glance. The occasion obvious. In one monthly creative review on one account, the strongest single ad was a close-up product detail, at well over twice the click-through of the variant it was tested against. That is one test, on one account, in one review period. It is not a universal rule. But the direction is consistent with what we see across location-targeted work: when the product is unmistakable and the occasion is clear, proximity becomes useful. When the creative is vague, proximity is just narrower waste.
The added value problem
What is needed alongside the targeting is a physical moment people traveled for: a promotion, a gift, something available there and not elsewhere.
Without that added value, a location-targeted ad is asking someone to make an extra physical effort for no particular reward. The fence is tight. The ask is not answered.
The consent problem
Location-based advertising depends on access to location signals, and the conditions under which those signals are collected have changed materially in recent years. A brand buying a geofencing campaign is downstream of decisions that were made between a user, a device, an app, and an ad system, and most of those decisions are no longer as permissive as they were when the category was built.
What changed with mobile permissions
Both Apple and Google have tightened the conditions under which apps can access precise location data. Users are now prompted to choose between precise and approximate location, and apps that request background location access face additional scrutiny and, in many cases, rejection from the app stores. The Electronic Frontier Foundation has documented how ad SDKs have historically passed location data into ad systems by default, without meaningful user awareness of what "agreeing to share location" actually meant in practice. Google's own location controls documentation distinguishes between what a user permits at the app level and what flows into advertising systems, and the gap between the two has narrowed.
The practical consequence for advertisers is that the pool of devices with precise, consented location signals attached to their ad profiles is smaller than it was. Vendors do not typically disclose what share of their targeting is based on precise versus approximate location, or how recently the location data was collected.
What to ask before you buy
This is not a legal opinion and it is not jurisdiction-specific compliance advice. It is a set of questions a buyer should put to any platform or vendor before committing spend:
What location signals are you using, and how were they collected?
What share of the target audience has granted precise location access, as opposed to approximate?
How recent is the location data being used for targeting?
What consent framework governs the collection, and has it been reviewed against current app-store requirements?
A vendor who cannot answer these questions clearly is not necessarily acting in bad faith. The data supply chain for location advertising is genuinely complex. But a buyer who does not ask them is accepting a targeting claim without knowing what it is built on.
The consent question is not a reason to avoid the tactic. It is a reason to buy it with your eyes open.
Where this argument is wrong
An article that makes a case without naming the cases where the case does not hold is not analysis. It is advocacy. Here is where the argument in this piece is incomplete or wrong.
Attribution is possible. Just not in this situation.
Campaign-level footfall attribution is credible where one party controls the measurement stack, the locations, or a closed loop. Some operators publish exactly that. Out-of-home advertising, for example, has produced cases where exposure data and footfall data were matched through a controlled methodology, with a single party owning both sides of the measurement. Those results are self-reported and the methodology is not always fully disclosed, but the underlying setup is different in kind from a brand advertising around doors it does not own.
The argument in this article is not that store-visit attribution is impossible. It is that it is unavailable in the specific situation most wholesale brands are in: running location-targeted ads around a retailer's locations, with no point-of-sale connection and no data coming back from that retailer. What would make it available is named in condition 3 of the Threshold Test: a code, a reservation, a dedicated page, or a retailer who reports back. The gap is not technical. It is a data-sharing arrangement that has to be agreed before the campaign launches.
The tactic has no value for brands without physical distribution
For a brand that sells exclusively online, or through a single owned channel, there is no floor to drive to and no proximity premium worth paying. The argument for location targeting depends entirely on there being a physical moment worth amplifying. Without it, the fence is just a smaller audience at a higher cost.
Precision can be a reason to spend badly
A narrow audience feels rigorous. It looks defensible in a brief. But precision applied to the wrong conditions is not discipline. It is a smaller version of the same mistake. A tightly drawn fence around a store where the product is not stocked, or around a week with no activation, is not better targeting. It is a more expensive way to reach the wrong moment.
The email case, again
The geo-targeted push that traced back to email rather than to location ads is worth sitting with. It does not prove that location targeting failed. It proves that in a multi-channel environment, the channel with the measurement infrastructure gets the credit. If email has a click trail and location targeting does not, email wins the attribution report regardless of what actually drove the behavior. That is a measurement problem, not a media problem. But the practical consequence is the same: the channel without a closing mechanism is invisible in the results.
Where to start
The decision to run a location-targeted campaign is not a media decision. It is a measurement decision made first, a creative decision made second, and a media decision made third. Most briefs reverse that order.
Before the fence is drawn, the question worth answering is this: when this campaign ends, what will we be able to say it did? Not what we hope it did. Not what the platform's reporting will show. What will we be able to say with confidence, given the data we will actually have access to?
If that question does not have a clear answer, the campaign is not ready to run.
We offer a review of exactly that question: whether a planned location campaign can answer what it will be judged on, and what would need to be in place before it launches. If you have a campaign in planning and want to work through the Threshold Test before the brief is signed off, get in touch.
Frequently Asked Questions:
How much does geofencing marketing cost, and what drives the price?
Cost varies by platform, targeting method, audience size, and campaign duration. On social platforms like Meta, tighter geographic targeting reduces the available audience, which typically raises cost per thousand impressions relative to broader delivery. On specialist location-targeting platforms, pricing is usually quoted per thousand impressions or as a managed-service fee, and it rises as the fence gets smaller and the audience more specific. The less inventory the platform has to optimize against in a given geography, the more it tends to charge for each impression it does deliver. Beyond media cost, there are production costs for creative, and often agency or platform fees on top. A campaign targeting a handful of store catchments is not cheap relative to its reach, and the cost-per-person-reached is almost always higher than a broad campaign at equivalent spend. Budget accordingly.
Is geofencing advertising legal, and what consent does it require?
Location-based advertising is legal in the United States, but the legal position is not uniform. Several states, including California, Virginia, and Colorado, have enacted privacy laws that impose specific obligations on the collection and use of precise geolocation data, and the definitions of what counts as "precise" vary between them. Federal law does not set a single national standard. What a brand is actually accepting when it buys a third-party location audience is worth understanding clearly: app-level permission, the prompt a user taps to allow location access, is not the same as consent for downstream sharing with ad-tech systems. The Electronic Frontier Foundation has documented how ad SDKs have historically moved location data into advertising pipelines well beyond what a user understood they were agreeing to. A brand buying a third-party location segment is purchasing signals that passed through that chain. The questions worth putting to any vendor before committing spend are what consent framework governs their data collection, whether it has been reviewed against current state-law requirements, and what their liability position is if the underlying consent does not hold.
How large should a geofence around a store actually be?
There is no universal answer, and anyone who gives you one without knowing the store, the category, and the customer is guessing. The relevant question is not how large the fence should be, but how far a customer in this category will realistically travel to visit this type of store. A destination retailer in a city draws from a wider radius than a convenience format. A seasonal pop-up in a dense neighborhood may need a tighter fence than a flagship in a suburban mall. Start with the realistic travel catchment for the store type, not with a default radius. Then consider whether the platform can actually deliver to that geography with sufficient scale. A fence that is technically precise but contains too few devices to optimize against will raise cost and reduce delivery quality.
Can a brand target customers around a competitor's store?
Yes, technically. Most location-targeting platforms allow a brand to draw a fence around any address, including a competitor's location. The strategic logic is that someone near a competitor's store is in the category and actively shopping. The practical limits are worth naming. First, the person is already at or near a competitor, which means the switching cost is real and the ad has to work harder to change behavior in the moment. Second, the brand still has to pass the same conditions that apply to any location campaign: the product has to be available nearby, the creative has to give a reason to act, and there has to be some way to know if it worked. Competitor targeting does not change the measurement problem. It adds a behavioral hurdle on top of it.
What is the difference between geofencing, geotargeting, and radius targeting?
Geotargeting is the broadest term: delivering ads to users in a defined geographic area, which can be as large as a country or as small as a city. Radius targeting draws a circle of a specified distance around a point and delivers to devices the platform believes are within it. Geofencing, in the strict sense, creates a virtual boundary that triggers an action when a device crosses it, typically in real time. In practice, most campaigns described as geofencing on social platforms are radius or zip-code targeting: the platform delivers based on where it believes users are or have recently been, not on a real-time boundary crossing. The distinction matters because the precision implied by "geofencing" is often not what is actually being delivered. Ask the platform which method it uses and what signals it relies on before assuming the fence is as tight as the name suggests.
Why can advertising platforms not confirm that someone visited a store after seeing an ad?
The core problem is that the ad delivery system and the store's point-of-sale system are entirely separate. A platform like Meta knows when it delivered an impression and to which device. It does not know whether that device then entered a store, and it does not have access to the retailer's transaction records. Store-visit estimates that platforms do report are modeled outputs: they compare the behavior of exposed users against a control group using location signals, and they estimate the difference. The methodology has known limits. It requires a sufficient sample of users who have location tracking enabled, it depends on the accuracy of the location signals available, and it produces a probability, not a confirmed visit. For a brand advertising around a retailer's doors with no point-of-sale connection and no data-sharing arrangement with that retailer, the chain between impression and transaction is invisible by construction.
Does location-targeted advertising work for a brand that sells only through other retailers?
It can move store intent, which is meaningful but limited. A brand in this position can use location targeting to raise awareness among people near a stockist, drive them to a where-to-buy page, and increase the probability that they seek out the product in store. What it cannot do, without a closing mechanism agreed with the retailer in advance, is connect that intent to a transaction. The retailer owns the point of sale. Unless the retailer shares data back, or a code or reservation system is in place before the campaign launches, the brand will be able to see digital engagement and nothing beyond it. For wholesale brands, the honest case for location targeting is that it supports consideration and stockist discovery. The honest limit is that it cannot prove conversion without the retailer's cooperation.
How long does a location-targeted campaign need to run before the results mean anything?
Long enough for the platform to optimize delivery, and long enough for the audience to act. On social platforms, algorithmic delivery typically needs time to move past the learning phase, during which cost and performance are less stable. For store-catchment targeting, where the available audience in each geography is smaller than a broad campaign, that learning period can take longer. A campaign that runs for only a few days is unlikely to produce stable data. A campaign of roughly four to six weeks gives the platform time to optimize and gives the audience time to encounter the message more than once. Beyond duration, the results only mean something if the measurement architecture was in place before the campaign started. A longer campaign with no closing mechanism produces more data about the same unanswerable question.
What should a brand agree with a retailer before advertising around their stores?
Three things, and they need to be agreed before the campaign brief is written, not after the results are due. First, whether the product will be in stock in the targeted locations during the campaign period, and who is responsible for confirming that. Second, whether any mechanism exists to connect the campaign to the retailer's floor: a dedicated code, a landing page the retailer will honor, a promotion exclusive to those doors. Third, whether the retailer will share any sales or footfall data for the campaign period, and in what form. Most wholesale relationships contain no standing arrangement on any of these points. Raising them after the campaign has run produces an argument about numbers neither party can verify. Raising them before produces either a measurable campaign or an honest conversation about whether the campaign is worth running at all.
Should location-targeted advertising be run in-house or by an agency?
The answer depends on what the brand needs to manage and what it needs to learn. Running in-house gives the brand direct access to the platform data, faster iteration on creative, and no markup on media. It requires someone with hands-on experience of algorithmic delivery, audience segmentation, and creative testing, because the decisions that determine whether the campaign works are made inside the platform, not in the brief. An agency that specializes in this tactic brings platform experience, an existing read on what creative performs in the category, and, if the tactic requires more precise location targeting than social platforms provide, access to programmatic partners and the data infrastructure that goes with them. The risk with either route is the same: buying the tactic before the measurement question is answered. That decision is not a platform decision or an agency decision. It is a strategic one, and it has to be made before the campaign is briefed.
