Facebook ads are 'far less viewable' than some advertisers were expecting
The consequence in being more transparent than before, however, is that advertisers are now paying a lot more attention to the accuracy of Facebook's numbers, and those numbers may not be as good as advertisers were expecting.
Speaking on a Nomura conference call about Internet advertising trends earlier this month, Drew Huening, director of Omnicom's digital ad buying trading desk Accuen, said (according to a transcript provided by Nomura): "Facebook ads are far less viewable than people [advertisers] were expecting."
As part of Facebook's improved measurement efforts - which actually pre-date the public measurement snafus- the company has begun allowing a number of additional third-party measurement firms, including Integral Ad Science, to verify its ad viewability and attention metrics.
Viewability is the advertising metric that determines whether an ad served on a webpage or in an app actually had the ability to be seen. The industry standard for viewability, as set out by trade body The Interactive Advertising Bureau (IAB) is that 50% of an ad's pixels should be visible in the browser window for 1 second. For larger-sized ads only 30% of the ad's pixels need to be visible, and for video ads, 50% of the pixels must be visible for a continuous 2 seconds.
Third-party analysis of Facebook ad metrics show "some really interesting numbers ... not all of which look great for Facebook"
Huening said that allowing Integral Ad Science measurement has seen "some really interesting numbers coming out of that, not all of which look great for Facebook, and I think people have been surprised by that."
While caveating that it's still early days, Huening explained that Integral Ad Science's viewability criteria - which requires that a display ad needs to be visible for longer than 1 second - appears not to look kindly on Facebook.
Huening said: "So, while I think Facebook has done an excellent job of building a system in which everything is viewed by human eyes, especially in mobile, people are scrolling very quickly, and so they're not reaching that one-second threshold."
Huening's comments refer to the static ads you see in the Facebook news feed, rather than video, or on the Facebook Audience Network (which lets advertisers buy Facebook ads on third-party sites and apps that are not owned by Facebook), which he expects "will be contributing pretty substantially to [Facebook's] bottom line, a couple percentage points potentially" in its fourth quarter.
He added that he doesn't think this viewability metric will lead to a dent in advertiser spend - ultimately, if advertisers are happy with the results they are getting from ads that have been viewed for just milliseconds, that's all that matters -but he said it "does require recalibration from an advertiser perspective."
Facebook was not immediately available to comment.
Facebook's view is that not all platforms should be viewed the same - even if that makes life easier for ad buyers
Earlier this month, at Business Insider's IGNITION conference in New York, I asked Andrew Bosworth, Facebook's VP of ads and business platform about the company's recent measurement efforts.
(The segment starts at 1:51:57)
I mentioned that every time I ask ad buyers to name the one thing Facebook could improve, it always comes back to measurement. And the key issue seems to be that they can't drop their own tags (pieces of code used to track ads) to target and measure Facebook ads in the same way they do with other online ads across the web. And even though Facebook works with third-party measurement firms, often the data ad buyers receive back is unique-to-Facebook data that can't be compared apples-to-apples to their other online advertising spend.
Bosworth explained that user data is "sacrosanct" to Facebook and advertisers being able to attach tags is the exact type of data leakage that "frankly represents a real risk to the security of the information that users are trusting us with."
He added: "We're not trying to be obstinate, we were not trying to be obstructionist ... it's honestly a commitment we have that, the nature of the platform, the platform can't exist without that commitment that we have given to people."
But on the marketer front, Bosworth says he believes Facebook is offering marketers ample tools to "measure, iterate, and get better on the platform, and we are committed to doing more and more," adding that its move to create a measurement council - made up of marketers, agency partners, and executives from measurement firms who will provide feedback each quarter on what Facebook is doing well and what Facebook can do better - will be a "huge help" on that front.
And as far as ad standards are concerned - such as the definition of a viewable ad impression - Bosworth said he believes they are not always helpful.
Bosworth said:
"Bear in mind these platforms are different. As much as we wish that a YouTube, and a Snapchat, and a Facebook video were all just interchangeable, they're not. They're different platforms, they're different consumer behaviors, and trying to pretend that they're the same platform may be convenient [for media buyers] ... but it's good until it's not. It looks good, it makes life easier, but does it help them provide the best content? Is it going to be the best return on investment? Maybe not. So I think there's a degree to which we need to recognize that different platforms are different, even if some of the pieces look similar. But having said that, I think certainly on things like 'what's an impression', 'who's a real person', I think we can make real progress in this industry."
Facebook's definition of an ad impression: Seen by a real person for more than zero seconds
I asked Bosworth what Facebook defines as an impression.
He responded: "For us an impression is seen by a real person. That means it has to enter the view stream, it has to have been there for more than zero seconds, and has to be a human on the other side of it."
North of zero seconds seems a lot lower than the 1-second IAB industry standard - and could well be why advertisers were surprised, as Accuen's Huening referred to in the Nomura call, at Facebook's viewability metrics.
When pressed on exactly what "more than zero" means, Bosworth responded: "No seriously, you'd think this would be not controversial in the industry we work in. More than zero pixels, more than zero seconds. [A] very good place to start for what's an impression. It is not a universally-held opinion of what's an impression."
He added: "Some people have a standard that's higher than that, a lot of places that people in his room are buying on places that maybe have a standard that's lower than that, and you should look into that."
Here, Bosworth appeared to be referring to the swathes of unchecked online ads marketers buy that are never actually seen by a real person. The ad may be "served" through the website's ad server but it could have appeared at the bottom of a page, or hidden in some other way, meaning a pair of human eyes never had the chance to look at it. Nevertheless, some advertisers still prioritize buying volumes of impressions at cheap prices, rather than quality impressions that are guaranteed to be seen, but come at a higher price point.
So even though Facebook's viewability metrics may not "look great," according to some advertisers, it's important to take into account the wider picture. Accuen's Huening did also concede in the conference call that there is still some value in "ads that aren't even clicked that slowly, over time, increase awareness, consideration, that sort of thing." And marketers have lots of different buying options on Facebook, including "100%-in-view impressions".
Viewability is certainly an important metric for advertisers to consider when measuring the impact of their online ads and the quality of the platforms they are advertising on. It makes zero sense to pay for an ad that wasn't viewed by a human. But ultimately, the most important metrics for advertisers to consider is whether their online ad campaigns - be they on Facebook, or elsewhere on the web - actually had an impact on their businesses.