Making #DigitalSense of Viewability - Sadly It's Not Black & White

Sadly it's not as black & white as you'd like it to be.

Digital marketing has had a rough 12 months. A lot of tough questions are rightly being asked around where ads are actually running and who, if anyone, they're really being seen by. As part of that, new 'viewability' standards & ways to measure them have risen from technical obscurity to being key c-suite discussions. Whilst these approaches can certainly form part of a response I'm afraid to say they're not the simple solution that they might at first appear to be - in fact if implemented poorly they can make things much worse for advertisers, publishers AND consumers.

For those who don't know, 'viewability' is essentially a way of measuring whether a digital ad you paid for was actually properly seen by a real person. Low viewability can be caused by fake bot traffic & deliberate fraud, but also less maliciously by an ad loading further down the page than a viewer reads, or because they scroll past an ad too quickly to take it in.

The equivalent in print media would be a way of measuring whether the paper boy dumps a stack of ‘delivered’ magazines in the bin, whether people read far enough through the magazine to see an ad on page 50, and whether people pause for a few seconds on an ad page rather than just flicking on past. Spoiler alert: we of course cannot really measure any of that in traditional media.

The MRC has proposed a popular standard which says that for a video ad to be counted as viewable at least 50% of its pixels must have been on a real person's screen for 2 seconds. That may seem a fairly low bar (and there are arguments for stricter standards) but given the measure is only supposed to check that your ad really was served it makes sense. You might even argue 2 seconds is quite a 'long' time to insist on as viewability then becomes highly influenced by consumers' behaviour and whether they choose to scroll past – longer view time requirements conflate a notion of ad length/effectiveness with mere appearance.

Measuring this rapidly gets very technical, but essentially a range of media owners, ad servers and 3rd party tech companies are developing technology which tries to accurately do so. Perhaps the most famous of these at present is MOAT who provide advertisers with digital tags that then allow them to have a dashboard comparison of the viewability (and other simple delivery measures) across a range of their media.

Most major digital media owners are starting to accept these tags and many big global advertisers are signing up to put this tech on all their ads. Given that there's then a slight fee on every single one of these impressions it's perhaps not surprising that Moat has been snapped up by Oracle for a cool $850 million - they're building a business which could one day take a cut of everyone's advertising. It’s debatable whether all media needs to be tagged eventually, or just enough to have a ‘dip stick’ approach.

After all, who wouldn't want all their advertising to be viewable? Any tool which can help identify where impressions are going to waste, and in turn focus spend on places where it is delivering should be able to radically improve the effectiveness of advertising and well and truly pay for itself. On the open web it's common for viewability rates to be below 50% which is a terrifying fact for senior marketers to hear and many are rightly speaking out and looking to tackle that. Yet whilst this sort of measurement is part of a fix, making media decisions based on blindly looking at viewability scores turns out to be an incredibly bad idea.

As these dashboards start popping up I've seen remarkable examples of their data being analysed without any concept of costs or effectiveness layered on, or without basic checks on how the media was bought and what it was optimised for. So whilst 100% viewability sounds like it should be any sane marketer’s goal it has some unexpected implications.

One of the key challenges is that to get 100% viewability by these measures you essentially have to force consumers to watch ads, and we all know how annoying unskippable ads are. Facebook for instance scores incredibly highly in terms of real people being served ads, but can still end up with a poor viewability score because those users scroll past an ad in less than 2 seconds. They could change their newsfeed so that it locked on every ad and held you there for a couple of seconds, but what advertiser really wants to be responsible for creating that hell?

There are other ways of getting to 100% viewability, in particular if you play with the numbers - Essentially serving your ad to lots of people but in billing terms discounting the ones that don't meet the standard's threshold. This makes a lot of sense and is something many advertisers are starting to investigate, but it has risks of its own primarily caused by how you then get treated in these sites auctions. You also pay a slight premium so potentially end up with very similar costs for your viewable impressions anyway, or in some cases can oddly find the effective costs even higher.

Sticking with Facebook as an example, they offer ways of buying their media which optimise for video views, not just impressions. Bought in this way FB comfortably scores a viewability score in the high 90%s, vastly more than if just bought on impressions. When you buy like this FB's algorithm starts to deliberately chase people it knows are more likely to watch for that little bit longer, which instantly starts to reduce your maximum reach, and of course CPMs are higher in this approach to begin with. Across a wide range of media owners any attempts to force a buy at 80-90% viewability can have serious implications for delivery, the market simply isn't there for it yet.

Surprisingly I have seen data from a couple of publishers which shows that shifting to 'viewable' ways of buying their platforms actually leads to worse overall campaign results in terms of brand & purchase metrics. You lose all the benefits of the natural platform auction and start chasing a specific stat instead. Unfortunately whilst the logic of viewability makes perfect sense there's very little research to suggest that it's actually a measure which will drive better business results if you optimise for it, which is awkward if we're all about to.

Auctions aside there's also the question of whether those 'unviewable' impressions might actually be worth something after all (the short ones, not the fraudulent ones we definitely want rid of). 2 seconds doesn't sound long but if you flick through a newsfeed or website on your phone right now you'll probably realise that you're taking in content and deciding whether to stick with it much faster than that. You might decide to immediately move on to something else, but it will still have been considered.

In a lot of cases then the impact of a deeper understanding of viewability shouldn't be sweeping decisions on where to invest, but more focus on exactly how the content should look in certain settings. To get the most out of what are potentially quite fleeting views you can't rely on a 60 second TV advert with your logo on the end frame, you need to communicate quickly and grab people's attention if you need them to stick around for longer.

It's early days for viewability and there will be further discussions around how factors like sound on/off, or screen coverage should be layered in to follow too. Again on paper these extra factors make perfect sense to add in but a lot of research would be needed to justify their weightings. Undoubtedly keeping a track of all of these measures is important, but their interpretation will continue to require nuance. It is not a black & white cross media digital currency that many in the industry would like to hope it is, and it’s definitely not a magic metric which makes different channels comparable.

There's an alarming amount of data in digital marketing and I'm always wary of how closely we track & monitor it if it doesn't truly correlate with campaign success. As soon as something is measured it's almost impossible not to start optimising for it but there are much bigger questions we really need to be asking about overall platform effectiveness - if you can show a campaign overall drove penetration, brand metrics and notions of physical & mental availability then most of the digital metrics in the middle become insignificant, though the desire to be able to optimise in real time is a tougher nut to crack.

In truth marketers & media agencies have not paid enough attention to exactly where their media is landing and who is seeing it. The current trend to deep dive into viewability should help tackle some of the biggest issues in this field including fraud and some elements of brand safety, but they're very much just a foundation.

At Cannes this year Diageo, who I work for, briefed key media owners & partners on the launch of their Trusted Marketplace. This is an attempt to tackle a lot of the issues in digital, curate a higher quality media supply chain, and begin to have deeper, more trusting partnerships. A minimum standard of viewability and acceptance of MOAT tags were 2 of the principles, but there were 8 more. Arguably more critical was a push for a clear commitment to wider measurement & learning, and an open challenge for partners to talk to their platforms unique strengths and how we should really be thinking about working with them. We'll get there, if we don't get side-lined by blind interpretations of data which only tell part of the story.

I am a marketer who helps global brands make sense of media in a digital world; Follow me on LinkedIn or Twitter. These#DigitalSense posts are my attempt to cut through the hype that surrounds the industry. I have led media, digital, social & partnership strategy on both client & agency side, and am a passionate speaker and industry contributor.