Viewable impression tracking and pitfalls.: April 22, 2013 by Maciej Zawadziński Chief Executive Officer
Recently, online advertising seems inevitably moving towards a ‘viewable’ impression model. A model, in which advertisers theoretically shall pay only for advertisements that were actually displayed to a user.
In this post, I am going to explore a few technical aspects of tracking viewable impressions and explain why this model may not solve problems which lie at the origin of it.
Since the early years of online display advertising, the most popular pricing model for selling online ads is banner impression which involves buying and giving prices in CPM (CPM = cost per mille, which means cost per 1000 impressions).
The problem with this model is that many ads are not seen by the user at all because they are displayed below the fold (for example in the middle or bottom area of the page where many of visitors do not scroll to), not to mention the fraud that occurs with long-tail sites that come up with many tricks in order to cheat tracking impressions of many ads that could never be seen on the page.
With more and more inventory available on the market, the increasing expenditure on online advertising and the growing number of ad technologies available, it is natural that advertisers want to find a way to pay only for the inventory that can actually bring value (and obviously ad impressions that are not displayed at all are not something that any advertiser would want to pay for).
Big hope is placed in the so called viewable impression…
The idea behind viewable impressions is pretty straightforward. An advertiser pays only for ads impressions that were actually seen by a user. Sounds simple, but the problem is much more complex when analyzed from the technical perspective.
First of all, what does it mean when we say “an ad was seen”? Usually, this is assumed when at least 50% of the advertisement area was in user’s browser viewport for more than X number of seconds. So far, so good, but it gets interesting when measuring it with various technologies that give different results or when part of impressions are classified neither viewable nor non-viewable due to the limitations of the browser’s capabilities.
There is a good video from TraderTalk TV explaining the technical issues surrounding viewability and giving insight into current solutions that are used by ad technology vendors:
1) The commonly used geometric method (calculating the relative position of the ad on the screen and determining if it’s in user’s viewport) is not cross-browser compatible (i.e. it does not work on Webkit-based browsers) and could be easily cheated.
2) There seems to be a very effective piece of proprietary technology developed by Spider.io which measures the rendering properties of the ad (taking advantage of the discovered fact that rendering frame rates and other parameters of elements actually visible in the browser window differ from those that are not visible).
Other companies on the market that deploy viewable impression technology besides Comscore and Spider.io include: Alenty, DoubleVerify, Moat, Integral Ads (formerly AdSafe), Meetrics, Onscroll, Telemetry, Realvu, Vizu and for sure many others.
Personally, I doubt that any of the above techniques will eventually become a standard as the variety of devices that users use to consume online content is constantly growing and changing (ranging from desktops/laptops, through to smartphones, tablets, game consoles, smart TVs etc.) which will make both techniques unsustainable. Also, proprietary technology is hardly going to be adopted widely by all players on the market.
The main driving force behind pushing the standardization of viewable impression is to improve results of online campaigns.
Of course, besides ad viewability, there are several other factors that impact the effectiveness of ads that are usually taken into account by advertisers:
All of above are nowadays relatively easy to implement with larger or smaller success in online advertising campaigns. However, there are a few other factors that can greatly impact the effects of the campaign, but are rather either omitted or are still in early stages of ad technology implementation:
Similar to financial markets, current CPM prices are reflecting the supply and demand for online ad inventory, including the fact that part of the impressions are non-viewable.
Moving towards CPV should theoretically result in better campaign performance for the same price paid by an advertiser. But that does not fit into a ‘free market’ model…
When the supply of impressions is cut (nobody wants to buy non-viewable ones), we can expect that the CPV, after possibly a few months of destabilization, will get to the level where advertisers pay around the same for similar campaign effects, but with only viewable impressions.
It does not sound that bad, but as the technology isn’t ideal (e.g. some of impressions can be determined as viewable or not), this could lead to even higher prices for viewable impressions and as a result, lower the effectiveness of display campaigns.
Of course, the current model where advertisers pay for ads that may not always be displayed isn’t ideal, but until we have an open standard for viewable ads in place that is cross-device compatible and reliable, I don’t think it will do any good.
Photo credit: x-ray delta one