Top of funnel
Programmatic with the view-through window closed.
Display, CTV, and audio bought with the inventory identified, fraud and low-viewability impressions filtered out before they count, and a view-through window short enough that it cannot claim credit for whatever happened anyway.
Programmatic is sold on a number that is easy to inflate: impressions served to inventory a buyer often cannot even name, credited with a conversion any time one happens inside a view-through window long enough to claim credit for almost anything that follows. A display ad served into a low-quality app full of bot traffic and a completed CTV ad actually watched by an attentive household can show up as the identical line item in a standard report, at the identical price.
This is not the same discipline as retargeting, even though the two overlap in one specific place. Retargeting is an audience definition — who gets shown an ad, based on something they already did. This page is about the inventory itself: where an ad actually runs, and whether that placement is real and fairly represented before any audience targeting gets layered on top. A retargeting campaign can be bought programmatically, but the two questions get answered separately.
How the work differs
- Inventory disclosure
- Every placement identified by domain, app, or CTV network before it counts toward reported spend.
- Fraud filtering
- Traffic screened against third-party invalid-traffic detection before an impression counts as served.
- Viewability standard
- An impression counted only once it clears the industry-standard threshold for pixels in view and minimum duration.
- View-through window discipline
- The view-through attribution window fixed at a short, disclosed length and reviewed on the same schedule as the campaign.
- CTV completion tracking
- Ad completion rate and household frequency tracked directly for every CTV placement, refreshed weekly.
- Channel-level reporting
- Display, CTV, and audio reported as separate line items, each against its own conversion definition.
How it is measured
A conversion counts when it happens inside a short, disclosed view-through window — measured in hours, not the seven-, fourteen-, or thirty-day defaults some platforms ship with — and matches through the same attribution layer used across every channel on this site. Fraud and viewability are filtered before spend is counted at all: invalid traffic identified by third-party verification, and an impression that never actually became viewable, are both removed from the base a campaign gets judged against.
What this channel cannot claim credit for: a conversion attributed only because someone was served an ad inside a window long enough to catch almost anything that happened afterward, whether the ad had anything to do with it or not. Shortening that window is the single highest-impact change available in this channel, and the reported return changes with it, usually downward.
For CTV specifically, where a click rarely exists to measure in the first place, a geographic or matched-market holdout works the same way it does for any other channel: markets where the CTV campaign runs are compared against comparable markets where it does not, and the gap in bookings between the two is the campaign's real effect, independent of whatever the view-through window would otherwise claim.
What we need from you
Raw, placement-level log access from the demand-side platform, since inventory quality cannot be verified from a summary report that only shows category-level spend. Agreement on a view-through window length before the campaign launches, so it is not quietly extended later once a shorter one produces a less flattering number. And whatever CTV or audio creative already exists, since production timelines for those formats run longer than a standard display banner.
Who this is for
Built for operators running enough top-of-funnel volume to justify buying across display, CTV, and audio rather than a single format, where a current programmatic report already looks strong and nobody has checked what inventory that number is actually attached to — multi-location healthcare groups running regional awareness campaigns across several formats at once see this pattern often. It reports alongside paid social prospecting as one of nine top-of-funnel programs, and the creative running inside these campaigns is concepted and tested through creative ads strategy before it ever reaches a placement. Client results show what a finished placement-level audit looks like next to the platform's own summary report.
Every placement is identified by domain, app, or CTV network before it counts toward reported spend, rather than left inside a bundled category like "premium display" that hides where an ad actually ran. A buyer who cannot name the sites and apps their programmatic budget touched has no way to know whether that budget bought real attention or filled inventory nobody would choose deliberately.
Traffic is screened against third-party invalid-traffic detection before an impression counts as served, removing bot and non-human traffic from the base a campaign is judged against. Viewability is enforced separately: an impression only counts once it clears the industry-standard threshold for pixels in view and duration, so an ad technically loaded but never actually visible on screen is not treated the same as one a person could plausibly have seen.
More measurable than its reputation suggests, though not in the same way as a clickable ad. Completion rate and household frequency are trackable directly per placement, and while a CTV ad rarely gets clicked, its effect can still be tested with a geographic or matched-market holdout the same way any other channel's incrementality gets tested. What is not reliable is crediting a CTV impression with a conversion through a long view-through window and calling that measurement.
Enough that placement-level reporting and fraud filtering produce a large enough sample to actually mean something, which varies by how competitive the specific inventory is. Below a certain spend level, the overhead of proper verification and reporting starts to outweigh what a business can realistically learn from the campaign, and a different top-of-funnel channel usually makes more sense until volume grows.
Retargeting defines who sees an ad, based on something they already did — visited a page, started a form. Programmatic, as covered here, is about where that ad actually runs and whether the inventory buying it is real and fairly represented. A retargeting campaign can be bought through a programmatic platform, but the audience question and the inventory question get answered separately, and conflating them is how a bad inventory buy hides behind a warm, high-converting audience.
See what your programmatic spend is actually buying.
A placement-level audit of one active campaign, showing exactly where the inventory ran and what the current view-through window is claiming credit for.