The unit's viewshed defines a geographic area. Mobile ad IDs present in that area while your creative was running are collected, then matched against your conversion events. Nothing is installed on the billboard, and coverage is partial by construction: a typical device match rate is 10 to 15 percent.

A viewshed is the area from which a unit is actually visible: a radius plus a direction, not a circle drawn around a pin. Most are machine generated from the unit's position, size, and orientation.

Some are drawn by hand, and that is a feature rather than a compromise. A 35 acre open air venue needs its boundary mapped by a person so that roads running past it are excluded and traffic driving by is not counted as exposed.

Geofence size tracks the unit type. A highway billboard is typically measured at around 100 meters; street level furniture, where people are walking rather than driving, is closer to 30 to 50 meters.

Dwell time is the other half of the definition: the length of time a device was inside the viewshed. Devices ping frequently enough that it can be calculated directly, as the difference between the earliest and the latest ping recorded while the device was in the viewshed. Different unit types, sizes, and environments call for different dwell times, for the same reason they call for different radii.

Mobile ad IDs from app SDK data, licensed through data aggregators. Consumers granted location permission to the apps in question, and the data is opt in.

AdQuick does not own the device panel. It is licensed, and AdQuick operates and processes it, which is why coverage is a function of who opted in rather than a product decision.

Impression counting rests on industry measurement from Geopath rather than on the device panel. Delivery and exposure are therefore two independent readings that should corroborate one another, which is useful when a stakeholder wants a second source on the same campaign.

A static unit runs your creative continuously, so every device in the viewshed during the flight is a candidate exposure.

A digital unit shares its screen. Slots run roughly eight to twelve seconds, and a shared unit may divide screen time eight ways. Exposure is therefore filtered against the operator's play logs down to the seconds your creative was actually on screen. See the article on attribution by inventory type.

Because the panel covers a fraction of any market, the raw capture is scaled up to estimate real world activity. The model is panel exposed devices multiplied by a market-level factor derived from that DMA's panel share, which typically lands somewhere between 6x and 10x. It is not a fixed multiplier.

Panel captures and up-sampled exposures are kept as separate series in reporting, so you can always see which one you are reading.

This matters most when you compare across channels. Agree which metric family you are comparing before you compare anything: booked impressions, weighted impressions, and up-sampled exposures are three different series, and putting the wrong pair side by side is the most common way a good result gets dismissed.

10 to 15 percent is the typical device match rate on an exposed audience. Not 100, and not close to it.

That is not a defect being worked on. Location data comes from consumers who opted in through apps, licensed through aggregators, so the ceiling is set by opt-in behavior, which no vendor controls. Not everyone carries a smart device, and not everyone who does keeps location services on, so the captured sample represents a portion of the population rather than a census.

Match rates also vary by platform and by market. Signal loss on one mobile platform reduced volume without eliminating it. Small campaigns swing hard, because fewer units and shorter flights mean fewer captured devices and a less stable match rate against them. That is the main argument for running a first test at real scale rather than at minimum viable spend.

It does mean deterministic one-to-one matching misses most of the real world impact of a campaign. A conversion count is a floor, not a total.

It does not mean the measurement is unreliable. A consistent sample supports comparison and modeling, and that is what the lift studies are for. The failure is treating a partial count as a complete one, which is why exposed device counts should never be compared against digital impressions.

For audience matching against other platforms, exposed device lists have matched at materially higher rates, above industry norms in AdQuick's experience.

Exposure is tracked at unit level and aggregates upward, so reporting can answer both how a single billboard did and how a market did. Statistical reliability scales with volume, and unit-level attribution reads on low-traffic units get noisy, which is why small campaigns should lean on market-level measurement.

Exposure assembly, deduplication, and matching run as an automated pipeline rather than as analyst spreadsheets. Study design and interpretation are human; data collection is not.

Quality filtering removes low-precision points, impossible movement patterns, and fabricated signals, and anomaly detection runs across the pipeline as a data-integrity check. Methodology corrects toward representativeness by extrapolating per market from researched device-ownership and opt-in rates, rather than treating raw device counts as complete.

The honest answer depends on which product you are asking about, because attribution is inherently identity-dependent while much of the measurement suite is not.

So the useful framing is not that OOH avoids the identity problem. It is that OOH has a genuine measurement path that does not touch device IDs at all, which cookie-dependent digital largely lacks. If identity durability is a stakeholder concern, weight the measurement design toward the second list. Device IDs can also be backfilled up to six months for past campaigns.

The device ID passback file contains a device ID and a device type. Nothing identifying, no names, no contact information. It is delivered as a CSV drop to a cloud bucket, and it is included rather than billed separately. Exposure data also exports into your own MMM, attribution platform, or warehouse.

Myth: There is a beacon or sensor on the billboard.

Reality: There is not. Location comes from licensed app SDK data, opt in, collected off the device.

Myth: AdQuick owns the device panel.

Reality: The panel is licensed. AdQuick operates and processes the data.

Myth: The measurement starts from your converters and works backwards.

Reality: It starts at the unit. Exposed device IDs are collected first, then watched for a conversion.

Myth: Match rates on exposed audiences are near complete.

Reality: They are typically 10 to 15 percent, and small campaigns vary considerably around that.

Myth: The passback file has real names in it.

Reality: Device ID and device type only.