Digital media problems: one step forward, one (maybe two) back

Just when you think it’s time to crack open the bubbly, another digital car crash roars into view, writes Dominic Mills.

It’s depressing, isn’t it? Digital media’s problems never seem to get any better. Any time there’s progress in one area – say brand safety or viewability – it all seems to go backwards in another.

Every time this happens, belief in the value of digital advertising takes another knock.

This time, according to the latest Q3 report for the UK from Integral Ad Science, an outfit which specialises in measuring digital media quality, the problem is in video.

IAS measures about 5bn impressions a day and, given its status, history and experience working with publishers and brands, one can take this to be a reasonable snapshot of the universe.

First, the good(ish) news: there are improvements, quarter on quarter, in some key areas.

UK Ad fraud in Q3 averaged 9.1%, down just over 3 percentage points from Q2. Hooray, and about time too.

But it’s not the time to crack open the bubbly. Fraud is a cat-and-mouse game. Every time buyers and publishers close down one fraudulent practice, another opens up. This is a problem that will never be put to bed.

But while there is an improvement in brand risk – averaging 9.1% in Q3 versus 13.3% in Q2 – there is a marked decline in viewability, down almost 10 percentage points from Q2 to 45.5%. This, at a time when there is a focus on viewability, is a marked and most unwelcome reverse.

But there’s devil in the detail. First, these figures are averages across the universe, and we all know that averages can mask as much as they reveal (remember the old story about the statistician who drowned in water that was on average 9.73 inches deep).

Second, these averages are across both, on the one hand, network and exchange impressions, and on the other publisher direct impressions (see chart, below). In every case, publishers score better than networks and exchanges, and the scores from the latter – because of the higher proportion of impressions they serve – drag the average down.

The lesson is obvious – it’s always better to buy direct from publishers. The trouble, as we all know, is that it is the publishers who hold the preponderance of quality inventory, and that costs more.

Strange, isn’t it, when the evidence stares them in the face, that buyers still prefer lower cost over quality, not just of audience, but over safety, viewability and fraud. But hey, if – for example – you’re not paid to chase viewability and publishers aren’t rewarded for providing it, why bother?

Still, seeking comfort where we can, at least the figures in the UK are better than those in the US, where they seem to be going in the opposite direction: viewability down 1 percentage point; brand safety risks up 2 percentage points.
[advert position=”left”]
And then there’s video in the UK, which is a car crash every way you care to look at it. Video brand safety is twice as bad as that for display – with 16% of impressions at risk; at 10.1%, video fraud rates are higher than for normal display; and viewability is seriously bad – 67% of impressions don’t match the IAB standard.

You don’t have to be a genius to figure out why this might be: demand is hugely outstripping supply of quality inventory. IAS says that, although video scores are not significantly worse than in Q2, video’s performance reflects the fact that this is a medium in its infancy.

There’s something in this – the explosion in video has really only occurred in the last two years or so – but it is clear that old buying habits die hard. Video is more expensive than display, so the urge to buy cheap means buyers rummage around even more ferociously in the bargain basement – with consequences we can all see.

The quality broadcasters – ITV, C4, Sky, UKTV and so on – don’t have enough inventory or sufficiently wide distribution to match demand. YouTube does, at low CPMs too, but its inventory is aggregated programmatically through the exchange and network platforms – where quality thresholds are lower.

Video is clearly not going to improve any time soon, and the quicker quality publishers – especially the newspapers – can up their game, the better. But we are in for a long haul here.

One of the problems is in the fragmented nature of the industry, whether that’s publishers, exchanges and networks, or third-party suppliers like IAS all beavering away semi-independently to create products that tackle the major issues. But producing industry-wide solutions to industry-wide issues is difficult, not least when it takes an age to even define the problems, let alone agree on the solutions.

But you have to start somewhere, and to this end IAS has just launched a new product, Pub Expert (sounds like something Al Murray is working on), which aims to help publishers and buyers resolve discrepancies in viewability measurement.

As I understand it (see what I mean about difficulties in defining problems), publishers measure viewability using the publisher pixel, which fires up when the ad container is available on a loaded page.

Advertisers and buyers, meanwhile, measure the ad call, which only fires once the creative is loaded in the container (OK, I never said it would be easy to understand). The trouble is that there can be a time difference between the two (and a greater difference with ‘heavy creative’), which leads to discrepancies of milliseconds when you’re measuring the ‘start time’ for viewability.

According to IAS, this discrepancy can affect the viewability metric for as many as 20% of impressions – which can amount to a serious amount of money over time.

But the wider point is that confidence in all the multiple aspects of digital media campaigns can only be improved when there is one number that all parties are confident in. When different sides use different numbers, it’s hard to assess value.

And that’s the bigger story. The system is so jumbled, confused and, at times, at odds with itself, that a common understanding of value is entirely missing.