Viewability By Design
While the unscrollable site might be a too radical for most publishers, a number have redesigned in recent years to improve engagement, optimize for mobile or develop new ad products. These include The New York Times, Time.com and the Atlantic-owned political magazine National Journal.
Considerations around ad viewability are often part of these efforts. When the Times revamped its site about a year ago, for instance, it relocated its 300x250 banners to increase viewability scores.
Likewise, the National Journal’s redesign in October 2013 enabled it to launch native content partnerships and boosted its banner viewability rates from an average of 50% to 67%, according to the magazine’s publisher and president, Poppy MacDonald.
She attributes this increase to moving banners to the top of the page and introducing a direct-sold custom display unit that takes up the full width of the site. “You can’t miss it,” MacDonald said, “so it takes away that right-rail blindness.”
The redesign, she added, was an opportunity for National Journal to enhance the audience experience by increasing engagement with the magazine’s editorial content as well as its advertiser platform.
“We considered what opportunities we could present advertisers as part of this redesign,” MacDonald said.
Of course, a redesign is intensive and not every publisher can devote the resources to have one completed. Even accommodating viewability without an aesthetic overhaul can be a pain.
One major online business magazine guides clients to programmatically sold units that have more than 60% viewability, determined by an internal audit. Yet the publisher still needs to package up this inventory manually.
Paying For Seeing
“We define viewability as an ad that is viewed by a real person who is in our target audience in an appropriate environment,” said Jennifer Gardner, Unilever North America’s director of media investment and partnerships. “We believe we’ve reached a tipping point in technology when the time is right to level the playing field with other media.” (Unilever works with GroupM agency Mindshare.)
Gardner is right when she says there’s technology that can level the playing field. The problem is there’s too much of it, and it’s all crashing into each other.
Or as Moat’s Goodhart put it: “Ad tech is a connection of connections.” And any one could keep an ad from loading and thus from being viewable.
For instance, an ad bloated with data beacons might load too slowly and the user scrolls past before it renders. Or maybe the user’s browser crashes. Or maybe other objects on the site load before the ad.
Ryan Bonifacino, SVP of digital at jewelry retailer Alex and Ani, believes a brand’s data policy tremendously affects viewability counts. Specifically, he said, a brand must either completely own and centralize its first-party data, or give it to an agency in order to get an accurate measure.
Bonifacino argues that concerns around viewability usually occur at the acquisition level, toward the bottom of the funnel. However ad fraud – impressions triggered by bots instead of real human consumers – typically happen upper to mid-funnel.
“Which means the brand has had the opportunity to quantify and ultimately score the purchase intent of that device/computer after the first touch point,” he explained. “If first-party data is coupled with, say, the brand’s social data and there are mid-funnel activities like lookalike modeling taking place, the risk of fraud is greatly decreased if known connections about that device/computer exist (whether they be from web analytics, campaign analytics or customer analytics).”
Brian Gleason, CEO of Xaxis North America, which is part of GroupM, sees it differently – most of the fraud problems, he said, happen at the bottom of the funnel thanks to the ad industry’s obsession with crediting the last ad impression before a purchase. This basically means viewability can’t work without sophisticated attribution.
“If you look at 100% viewability and do not take into account attribution, you’re setting yourself up to fail,” Gleason said.
So now the scenario is such that viewability, attribution and fraud are all intertwined. Yet they’re all separate measurements.
And even if we overlook attribution – which is more of a consideration for selling product than it is for branding – viewability and fraud are still closely related. Except, again, they’re not always assessed together.
“One of the biggest differences in terms of viewability metrics is how fraud is actually measured,” said Wayne Gatinella, CEO of MRC-certified viewability and anti-fraud vendor DoubleVerify. “The MRC requires fraud be eliminated from viewability reporting metrics.”
The reason, said David Gunzerath, SVP and associate director at the MRC, is that while non-human impressions shouldn’t ever be counted as viewable, not all accredited vendors can detect fraud.
“For the sake of counting consistency, the [viewability] standards require measurers to apply a standard order of processing in their counting approaches,” he explained.
If a vendor can fight bots, it should also reveal how many impressions were removed due to fraud, Gunzerath said. He added that the MRC is “currently working to enhance the basic filtration requirements for served impressions … and also to provide a framework for assessing those companies that specialize in advanced non-human traffic detection.”
Even though the MRC is working to streamline the issue, many advertisers and publishers feel there are too many vendors offering different viewability solutions.
“There are 15 different ad tech companies that are accredited by the MRC now,” said Matt Prohaska, who helped build out The New York Times’ programmatic initiative and who now runs Prohaska Consulting. “It’s nice that they’ve passed that test, but it’s difficult to get a standard currency.”
Because vendors use different rules for counting viewable impressions, publishers must use – and pay for – multiple viewability companies, Prohaska said.
The same can be said about the advertiser side. Joe Barone, managing partner of digital ad ops at GroupM, said that after three years of testing, the unit uses three preferred viewability vendors “that have shown the highest measurability rates and the lowest range of discrepancy.”
It helps, of course, when publisher and advertiser use the same viewability vendor. For instance, both GroupM and Condé Nast use Moat.
GroupM’s terms and conditions call for payment based off viewability counts from its preferred vendors, and Barone said it tries to make “commercially reasonable attempts” if there are any disagreements between publisher and advertiser.
For Moat’s Goodhart, reconciliation requires “granular day-to-day and creative-by-creative data … be exchanged until the differences are understood or an agreement is reached.”
Most publishers, he added, don’t take issue so long as the advertiser’s vendor’s numbers are within 10% of Moat’s. Goodhart pointed to a template by FOX News that uses these guidelines.
But even an unviewable ad, Prohaska argues, isn’t entirely worthless.
“In a programmatic buying world where retargeting is still the dominant play for a buyer, even if something isn’t in view, it’s still being served and an advertiser can still drop a pixel in a non-viewable banner,” he said. “So one could argue that there’s some value in a non-viewable impression.”