It is widely accepted that, of all the ads that are served in Ireland only around 54% of these are ever “viewed” by a person. In comparison with the UK this does fare slightly better coming in at around 46% however, a long way behind Germany at an estimated 71%. So yes, there is a long way that we can go in improving this metric although it is important first to ensure that we understand the how viewability is measured and the issues that arise before we decide whether it is a metric that is worth optimising towards.
Viewability is on the face of it a very simple metric – Of the number of ads that are served, what amount of them are actually viewed. To answer this we must first know what is counted as a viewable impression?
The standard definitions are all very well but how do we start to measure these metrics accurately and reliably? What are the issues that arise?
Is it the right metric? – The above IAB standardisation is a great as a starting definition of viewability however, it does raise the question about where this metric came from? It does seem rather arbitrary and do we really notice half an ad if it is on the page for 1 second?
As the main focus of your attention is likely to be on the article or content that is being scrolled through, it does seem a bit of a flaky measurement. According to FLE Marsh persuasion from a banner ad only starts from 5 Seconds, so should the measure of viewability be even more stringent?
What about different sizes? – The size and orientation of any banner affects viewability. For example, vertically aligned ad sizes such as the 160 x 600 can increase viewability by more than 10% than the horizontal formats such as the 728 x 90.
The same goes with the area for a creative, the larger the area the higher the viewability. So should we just be running a giant 300 x 600?
Discrepancies – The technical challenge that arises in accurately measuring whether an ad is on the screen in front of a user is a lot more technically demanding that it initially feels. The technology has to be able to cope with a huge variety of screen sizes, zoomed in/out browsers, multiple tabs and multiple browsers to name but a few.
These technical challenges can lead to very high discrepancies between different measurement tools between 30 – 40%. With such a variation how can we be sure that the measurements that we are making are even accurate or even reliable?
Does it actually make a difference? – This is the big question. Knowing an impression is viewable does not necessarily mean that it is effective. If it is unengaging, irrelevant in the wrong context and untargeted it does not matter that the ad is viewable, it is still going to have no impact on the user as if it was never seen.
If a user does not see an ad why should the brand pay for them? I personally think that this is a very good question. In the US where viewability measurement is ahead of Ireland, there are a few campaigns that are being traded with certain goals and targets set.
For example, a viewability target of 70% could be set. If at the end of the campaign only 65% of ads are viewed, either a rebate is given or a further 5% viewable impression delivery is necessary to complete the deal.
However, in the market we must be very careful not to jump onto such opportunities without rational thinking. Jumping onto this too early without standardised measurements, large discrepancies and without consensus on reconciliation guidelines we will breed confusion, scepticism and mistrust in the digital industry.
So at the moment the un-viewable impression is a tax on the display industry. To get the viewable we must also pay for the un-viewable. It cannot be denied that impressions and therefore budget is being wasted. There is no denying also that there are significant challenges ensuring that there are accurate and precise measurements that the industry can stand behind.
The viewability of impressions is something that does need to be addressed and is not something that we should bury our heads in the sand about. We should strive to increase viewability and to push for value for clients but at the moment are we not measuring the length of a piece of string with an inaccurate ruler?