select all

Facebook’s Black Box Is Bad for Business

Photo: mattjeacock/Getty Images/iStockphoto

If you want to be a large and important media company in 2016, the odds are very good that you’re leaning very hard on video. BuzzFeed, for instance, reorganized recently to make sure every department at the company is cranking them out. And if you’re leaning into video, the odds are high that Facebook is a sizable part of your strategy, maybe even the primary focus. That’s because it prioritizes video native to Facebook in the News Feed, the portal through which an enormous proportion of web users experience, well, the web.

There is a lot of money riding on Facebook’s video platform, so it would be really funny if, for instance, they had miscalculated a key metric that publishers and advertisers use to understand their audiences. That’d be what we in the digital media industry call a “big oopsie-doopsie”!

Anyway, according to the The Wall Street Journal, “[a]d buying agency Publicis Media was told by Facebook that the earlier counting method likely overestimated average time spent watching videos by between 60% and 80%.” Hoo boy.

The problem was that, in calculating the average time spent watching a video, Facebook ignored views under three seconds. This artificially inflated the average by removing the shortest watch times from the equation. This went on for two years, over the course of which advertisers paid Facebook money to display ads, and determined how much money to pay Facebook for ad placement based on metrics supplied by the company.

For most of the 20th century, advertisements were priced and purchased based on statistics and metrics like circulation and ratings, provided by third parties, like the Alliance for Audited Media and Nielsen. Those figures were mostly bullshit, and most people involved in buying and selling ads knew it, but they were agreed-upon bullshit that helped provide, at the very least, relative sizes for various magazines and newspapers.

In the platform era, though, you’re getting metrics directly from the same company that you’re paying to advertise on. If Facebook says that low viewership numbers are countered by lengthy average view times, well, who else are you going to turn to? It’s not that Facebook is blatantly dishonest — it’s in the company’s best interests to be clear about its numbers — it’s that if it commits an enormous, two-year mistake, who has the ability to call them out on it?

This almost certainly isn’t a case of Facebook distorting the data it provides in order to lure advertisers or entice digital publishers. For one thing, there are more subtle ways Facebook could juice its numbers — like, say, defining a “view” as only three seconds of watching (which it does). For another, the media-wide shift to video was never dependent on only a single metric. Advertisers want video (look at TV); people watch video (look at YouTube!); Facebook was going to push its video on the News Feed no matter what. There’s no other social network that advertisers can go to that can compete with Facebook on sheer scale. All this really seems to demonstrate is that amidst its eagerness to join the video train, Facebook made a mistake that served to amplify an already emphatic case for video.

A new metric, “Average Watch Time,” will replace the old, misleading one, which was known as “Average Duration of Video Viewed.” In theory, this whole dustup should make advertisers and publishers more wary of relying on Facebook for generating traffic and revenue. Even better, it should make Facebook recognize that opaque business practices will only hurt them in the end, so they should make their metrics and accounting practices (not to mention their various sorting algorithms) as transparent as possible. But in practice, Facebook already runs the whole damn game, so it means nothing.

Facebook Miscalculated an Important Metric for Two Years