Viewability was introduced to solve a real problem — advertisers paying for impressions served below the fold, in background tabs, or never rendered at all.
It solved that problem and it has been widely mistaken for a measure of whether anybody looked, which it is not and was never designed to be.
What the standard actually says
The common industry definition requires a proportion of the ad's pixels to be in the viewable area of the browser for a minimum duration.
For display, that is typically half the pixels for one second. For video, typically half the pixels for two seconds.
Read that plainly. Half of the unit, on screen, for one second, with no requirement that a human was present, looking at the screen, or aware of the advertisement.
An ad scrolled past at speed can satisfy it. An ad at the edge of a screen while somebody reads the middle satisfies it. An ad on a tab that was open while the user made tea satisfies it if the tab was in focus.
Why the industry adopted it anyway
Because it is measurable, consistent and verifiable, and because the alternative was worse.
Before viewability standards, a meaningful share of purchased impressions were never rendered in a viewable position at all. Establishing a floor removed the most egregious waste.
It also gave buyers something to hold sellers to contractually, which had commercial value regardless of what it measured.
The problem is not the metric. It is that a floor became a target, and hitting the floor became a proxy for effectiveness.
The optimisation trap
What happens when viewability becomes the goal.
Buyers optimise toward high viewability rates. Sellers respond by creating inventory that achieves them.
Sticky units that remain on screen while the user scrolls achieve very high viewability. So do units in positions where they are unavoidable, and units on pages designed so that content loads slowly around them.
None of that produces attention. Some of it produces irritation, which is worse than being ignored.
The pages that score best on viewability are frequently the ones a reader is most annoyed by, which should have been a clue.
What attention measurement attempts
The response that has developed over recent years, with real progress and real limits.
Panel-based eye tracking, where a recruited panel is observed with cameras and gaze is measured directly. This is genuine measurement and it is a small sample extrapolated to a population.
Behavioural modelling, using signals like scroll speed, dwell, interaction and viewport position to predict attention, calibrated against panel data.
And a growing body of research linking measured attention to brand outcomes, which is the part that makes the metric commercially interesting rather than merely more accurate.
The findings that recur: attention varies enormously by format and placement, far more than viewability does; some very high viewability inventory receives almost no attention; and there is a threshold of exposure duration below which measurable brand effect is minimal.
The practical caution
Attention metrics are a genuine improvement and they are not a solved problem.
Methodologies differ substantially between vendors, and the same inventory can score very differently depending on whose model is used.
There is no agreed standard, which means attention scores are not comparable across providers and cannot be contracted against the way viewability can.
And there is a commercial incentive for vendors to produce numbers that justify their own existence, which is a reason for scepticism rather than dismissal.
What I would actually do
Treat viewability as a hygiene floor. Set a minimum, enforce it, and stop optimising toward it once it is met.
Use attention data directionally to compare formats and placements rather than as a currency.
Pay more attention to the things that correlate with attention and are easier to assess — format size, position in content, page ad density, and whether the surrounding content is worth reading.
That last one is the most useful and the least measured. Attention to advertising is downstream of attention to content, and inventory on pages nobody is reading properly is not going to receive attention regardless of how it scores on any technical metric.
Which brings the argument back to the same place as most arguments in this industry. The quality of the environment is doing more work than any measurement layered on top of it.
Contracting on it
A practical note for anybody negotiating.
Viewability guarantees are standard and worth having, and the detail matters more than the headline percentage.
Which measurement vendor, since results differ between them. Whether the guarantee is measured in-view or of measurable impressions, which excludes those that could not be assessed. And what the remedy is, which is usually make-good inventory rather than money.
A ninety percent guarantee measured against only the impressions that happened to be measurable is a considerably weaker commitment than it appears.