Programmatic Measurement and Viewability

The Viewability Standard

The IAB viewability standard for display: 50% of pixels in view for at least one second. For video: 50% of pixels in view for at least two seconds. These standards were introduced because a significant proportion of programmatic impressions — sometimes 40-60% in open exchange environments — were served to ad slots that were never actually in the user's viewport.

Viewability is a floor, not a ceiling. An ad that is 51% visible for exactly one second meets the standard but delivers minimal impact. Attention metrics (time in view, heat-mapping, scrolling behaviour) provide a more nuanced picture of actual exposure.

Invalid Traffic (IVT)

Invalid traffic is any impression, click, or action generated by non-human sources (bots, automated scripts) or by humans acting in bad faith (click farms, fraudulent apps). The Media Rating Council (MRC) classifies IVT as GIVT (general invalid traffic, detectable by standard means) and SIVT (sophisticated invalid traffic, requiring advanced detection).

IVT rates vary significantly by channel: display has higher IVT rates than search, open exchange has higher rates than PMPs, certain categories (finance, pharmaceuticals) attract more fraud because CPMs are higher. Independent verification from IAS, DoubleVerify, or MOAT provides the most reliable IVT measurement.

Brand Safety and Suitability

Brand safety is the baseline: ensuring ads do not appear next to content that would be unambiguously harmful to the brand (terrorism, hate speech, illegal content). Brand suitability is more nuanced: ensuring ads appear in contexts appropriate for the brand's specific positioning and audience.

The Global Alliance for Brand Safety (GARM) Brand Safety Floor and Suitability Framework provides a standardised categorisation of content risk levels. Advertisers define their suitability parameters; verification vendors score content in real time and block or report ad placements that fall outside defined parameters.

Why Vendor Numbers Disagree, and Which to Trust

Two verification vendors measuring the same campaign will report different viewability, and the discrepancy is a measurement artefact rather than a fault.

The causes are mundane and worth knowing. Tag placement differs — a measurement wrapper placed at a different point in the ad's load sees a different set of impressions. Measurability rates differ, because some environments block measurement entirely and each vendor handles unmeasured impressions differently in the denominator. Filtering differs, since each vendor removes invalid traffic on its own rules before reporting.

The practical answer is not to find the correct vendor. It is to designate one as the system of record for billing, write that into the contract, and use the others directionally. A campaign measured by three vendors with no designated source of truth will spend more time reconciling than optimising.

And always read measurability alongside viewability. A 70% viewability rate on 50% measured inventory is a much weaker claim than the headline suggests, and reporting that omits the measurability figure is omitting the part that qualifies it.

Invalid Traffic Beyond the Obvious

Invalid traffic splits into a category everyone filters and a category most buyers never see.

General invalid traffic is the routine layer — known data-centre ranges, declared crawlers, obvious non-human patterns. It is filtered as a matter of course and is largely a solved problem.

Sophisticated invalid traffic is the one that matters: hijacked residential devices, ads stacked invisibly on top of one another, pixel-stuffed frames, falsified app identifiers and domain spoofing. It is detected statistically and after the fact, which means some of it is paid for before it is found.

What actually reduces exposure, in order of effect: buy through fewer, shorter supply paths, since most spoofing enters through resold inventory; check the seller authorisation files for the domains you buy and treat unauthorised sellers as a reason to exclude; prefer certified counterparties in the supply chain; and make make-goods contractual rather than relying on goodwill after the fact.

Treat a suspiciously cheap, high-performing placement as a finding to investigate rather than a result to scale. The inventory that looks too good is the inventory to audit first.

Sources

What each claim on this page rests on. Entries are typed so you can see which are primary.