Facebook's video metrics screw-up, explained

As far as screw-ups go, Facebook inflating average video watch time on its platform is a classic face plant — embarrassing but not big enough for publishers and advertisers to reconsider or pull back their investments on the platform.

About a month ago, Facebook started notifying advertisers that a calculating error had led to the platform overstating average viewing time for videos on its platform, according to The Wall Street Journal. Instead of dividing the total time spent watching a video by the total number of people who watched that video, Facebook’s metric reflected the total time spent watching divided by the number of views the video had generated. With Facebook counting views at three seconds, that meant anyone who had seen just a glimpse of the video was not getting represented in the metric. In fact, Facebook told advertisers that its metric was off by 60 to 80 percent, according to The Journal.

If that sounds like a huge deal, that’s because to an extent it is. Here’s why:

OK, so why is this such a big deal?
Facebook is dominating the digital ad market right now. Some estimates have 85 percent of new digital ad spending going to Facebook and Google. With great power comes great expectations. Any changes that Facebook makes can have a seismic effect on how marketers operate. By miscounting average watch time, Facebook essentially said that people aren’t watching as much video as it originally claimed. For Facebook, which has made video a top priority of its business going forward, this is bad. For advertisers that might have made high-level budgeting and spending decisions based on the average total watch time metric, it’s alarming.

But you said it’s a problem only “to an extent.”
Multiple agencies including GroupM and Horizon Media claim the screw-up hasn’t affected pricing or audience deliveries for campaigns. This is primarily because agencies don’t use average watch time to buy video on Facebook; instead, they use performance metrics like impressions, 10-second views and completed views — none of which were affected by Facebook’s error. GroupM also has used Moat to verify its Facebook video campaigns and in a statement said it’s “had a clear sense of the short duration times since implementing” Moat.

“Of course time spent is important, but it’s important in terms of completion rates or costs per completed views as opposed to a metric of average time spent,” said Anita Walsh, director of social strategy and marketplace buying at Horizon Media. “True paid results are not affected by this.”

Still, that doesn’t mean advertisers are pleased with the news. GroupM called it “careless and unfortunate,” while Walsh said “something like this shouldn’t take two years to uncover.”

If it didn’t affect paid campaigns, why does this matter then?
Facebook is mostly a closed network. Because of its size and influence, Facebook has largely ignored calls by the publishing and advertising communities to accept third-party measurement and verification within their walled gardens. It allows some. For instance, Facebook allows advertisers to verify campaigns using Moat, Nielsen and Integral Ad Science. But that’s not enough. Publishers and advertisers want universal third-party measurement, which they could not only trust but also use to compare performance across multiple platforms.

“It’s not an easy answer to solve. Standardization means you treat all of the platforms the same, and that’s a disservice to the platforms and the unique experience each has built around video — on TV the experience is all linear and the same,” said Jason Kirk, chief business officer at data firm Zefr. “But there will be a faster move of budgets from TV to digital if they do figure it out.”

How likely that is to happen is ultimately up to Facebook, Google and the other walled gardens in digital media — so don’t hold your breath.

“The holy grail would be to have all publishers and platforms held to the same measurement because no publisher should grade their own homework,” said Walsh. “I hope this pushes the discussion forward, I’m not sure that it will.”

This seems to be all about marketers — what about publishers?
Publishers aren’t left completely unscathed. Three publishing executives, all speaking anonymously because getting on the wrong side of Facebook is dangerous to your health as a publisher, said that their companies don’t price custom video for Facebook based on average watch time — views and engagements are what marketers are interested in. But considering that many publishers are devoting resources to Facebook video and live video — because Facebook wants video — to learn that people aren’t watching as much video as Facebook claimed is a cause for concern.

Facebook convinced publishers to distribute directly on its platform “under the guise that people will remain engaged with the content, which meant we’d still be able to monetize,” said one publishing exec. “They took that engagement away from us and it’s not existing on their platform, and that’s freaking publishers out. They’re going to have to make sure they don’t go back on the promises they made a while ago.”

There seems to be some giddiness in the reaction.
Bingo. Call it schadenfreude. Facebook seems to go from strength to strength, so it’s only natural that publishers, marketers and agencies jump at an opportunity to point out that Facebook too can screw up. It’s also a bit of a bargaining chip. The mess-up allows marketers and publishers to put pressure on Facebook for better — and unbiased — measurement. The good news for Facebook: This will probably be forgotten in a week.