Ad viewability “should not be a KPI”, according to study

A large study of almost 10,000 people found that ad viewability affects ad recall, but goes on to state that it should not be a KPI.

The research, compiled by IPG Media Lab, Integral Ad Science and Cadreon, sought to explore whether ‘viewable’ ads are actually effective.

The viewability problem is considerable, with more than half of all ads not being seen by humans. But how effective are those ads that are seen?

Also, how do ads perform relative to the industry standards?

How is ‘viewable’ defined?

According to the Media Ratings Council (MRC) a banner ad is considered viewable if it is at least 50% in view for at least one second.

A rich media ad only needs to be 30% in view for the same duration, whereas a video ad must be 50% in view for a minimum of two consecutive seconds.

IPG Media Lab said it did not set out to recommend changes to the MRC standards, nor to develop a new standard.

However it found that ads that exceed the MRC standard have 16% recall as opposed to ads that meet the standard, which have just 2% recall.

The research

A total of 9,876 people took part, and were shown one of 189 different ad scenarios on web pages that “matched their typical consumption habits”. In addition, eyetracking was employed for a subset of the panel.

The participants were then invited to complete a post-exposure survey to see if the brand metrics were affected.

The findings

No surprises: viewability is highly related to ad effectiveness. However, the duration is way more important than how much of the ad is in view.

However, while those ads that met the MRC standard attracted the attention of around half of the participants, they performed no better than ads that didn’t meet the standard.


The research states that, “The MRC standard isn’t a magical threshold for ad effectiveness.”

It also found that ads that didn’t meet the standard sometimes outperform those that did, for example, partial ads that are in view for a longer period of time.

What about ads that are above the MRC standard? It turns out that they don’t always work either, “specifically when time in view is low”.

“Time in view is king”

The study shows that ads that are seen for at least seven seconds increase ad recall by 17%.


By contrast, even ads that are fully viewable only increase ad recall by 11%. Time in view matters a lot more than percent in view.


The research also shows that all ad formats benefit from increased viewability, with video ads particularly more effective if seen for a longer period.


Large format ads “have the most to gain from increased viewability”, while video ads that surpass the MRC standard were seen by 96% of participants.

Johnny, remember me

So then, you’ve made the effort with the creative and spent the money to place the ads, but will people actually remember them?

At the MRC standard for banner ads (1 second/50% in view) there is a 19% chance of ad recall.

However, if your ad is fully in view for seven seconds, that number doubles to 38%.


For video ads, the MRC standard (two seconds/50% in view) only results in 10% ad recall, but if fully in view for seven seconds ad recall triples to 30%.

Unfortunately for the UX purists, autosound results in 275% uplift in effectiveness, from 8% ad recall to 22%. ‘Audio on’ is unsurprisingly more effective for ads not in view, though the study did not record the amount of people who had a negative response to audio (I’d wager that not all ad recall positively affects propensity to buy).

What was surprising was that there are “no additional gains in effectiveness” when ads were contextually matched to content.