What is measurable depends on the format you bought and on the outcome you care about. Static and digital units both attribute, but on different timelines and at different volumes, and a few inventory types cannot be attributed at all. Check availability before you buy rather than after.
Static units
Your creative is up continuously, so any device in the viewshed during the flight counts. Nothing has to be reconciled against a schedule, which is why static delivers weekly and can be attributed from the first week.
Static creative is generally needed about two weeks before launch, against two to three business days for digital.
Digital units
A digital screen rotates. Slots are roughly eight to twelve seconds, and a shared unit may divide screen time eight ways. To know which devices were present while your creative was actually on screen, exposure has to be filtered against the operator's play logs, matched at second-level timestamps.
That filtering is why the numbers arrive later, usually near campaign end, and why they are lower. On a unit shared eight ways, share of voice divides by eight. Lower exposure volume with stronger conversion rates is the normal digital pattern, not a problem with the campaign.
Play logs, and the units that cannot be attributed
Play logs come by direct API from three large US vendors and by manual process from everyone else. Vendors who do not collect play logs cannot be attributed at all, and those units are flagged as non-attributable on the map at planning time.
It is not unusual for one unit in a small digital set to have no vendor play log data and to be dropped from the study, so it is worth confirming coverage across the whole buy while you are still planning it.
Attribution by inventory type
Roadside digital and static billboards: full device-based attribution, with exposed devices matched to web visits, app installs, and store visits. The viewshed and flight methodology applies cleanly.
Place-based digital such as venues, gyms, and elevators: device-based attribution works, venue context adds audience precision, and verification is playout-based.
Transit: attribution works for station media. In-motion media such as train exteriors and bus wraps is a mobile unit, and mobile units cannot be attributed regardless of format, so market-level lift methods carry the weight there.
Airports and multi-level indoor locations: device signal in terminals is often measurable, but some multi-level airport and transit locations cannot be attributed at all. Traveler audiences also convert away from their home markets, which the attribution design has to account for.
Attribution by outcome
App downloads: exposed devices matched through app measurement partners.
Web visits and purchases: pixel-based matching. See the pixel tracking article for what you have to install.
Foot traffic: location-based visit measurement against your own locations. No pixel is needed.
Offline purchases: possible through transaction data partnerships or matched first-party data. It is the heaviest lift of the four and is scoped case by case.
What this means for planning
If you need daily attribution during the flight, weight the buy toward static. If you need creative-level reporting on a rotation, you need a vendor who supplies play logs. If a format you want cannot be attributed, causal lift, search lift, halo effect, and brand lift all still run on it.
What scale attribution needs
Every method has a minimum reliable scale. Too little spend means too few exposed devices, which means a noisy read. Web attribution is the most demanding on this point, because the exposed set is built market by market rather than across the campaign as a whole.
We will tell you whether a planned campaign is large enough to measure what you want to measure before you buy the study.
Third-party validation
For stakeholders who want independent numbers, established third-party partners can run or validate studies: Happydemics, Lucid, and Dynata for brand lift; Measured, Paramark, TransUnion, Attributy, Keen, and Evidnt for marketing mix modeling and omnichannel measurement; and AppsFlyer for app measurement.
Common misconceptions
Myth: Attribution only works on digital units, or only through certain providers.
Reality: It runs on all unit types. The exceptions are mobile units, some multi-level indoor locations, and vendors without play logs.
Myth: A lower exposure count on digital means something went wrong.
Reality: Shared screen time divides your share. Lower volume with higher conversion rates is expected.
Myth: A format that cannot be attributed cannot be measured.
Reality: Only the attribution family depends on the unit carrying attribution data. The lift studies do not.