The media is furious with Facebook in light of the news that the social media giant fudged some key video metrics — and changed an entire industry as a result.
If you’re an advertiser or work with advertisers, you’re likely already more than aware of Facebook’s recent video scandal. But if you can’t bear to keep up with even more bad news in a shockingly bad news cycle, here’s the gist: Facebook fudged a video viewership stat, contributing to what’s now known as the “pivot to video.”
The “pivot to video” refers to publishers’ and advertisers’ 2015 prioritization of video over text-based content, with the belief that video would be the key to any media company’s survival in an increasingly digital world. This pivot led to the hiring and firing of entire teams and a complete reshuffling of many companies’ strategies — but the data that fueled it may have been vastly inflated.
As it turns out, when Facebook was reporting “average duration of video viewed” on its partners’ videos, it was only including any views that lasted three seconds or longer, failing to account for the many, many users who scrolled away after a couple of seconds. Because of this questionable (or downright dishonest) practice, average time data was reportedly inflated by 150 to 900%.
Worse still: a class-action lawsuit a group of advertisers brought against Facebook alleges that the social network was well aware of its error long before it informed any of its partners or advertisers. According to the lawsuit, emails between Facebook employees show intent to “obfuscate the face that [they] screwed up the math.”
For brands and advertisers, whether it was intentional or not, bad data and a lack of transparency are incredibly frustrating. Advertising’s overreliance on vendors like Facebook, LinkedIn, and Google has been a source of concern for years, and this latest scandal further cements a sense of distrust in third parties. While it’s unrealistic to completely separate from digital ad vendors, these three key lessons from the Facebook video scandal can help guide future partnerships with vendors:
1. Take “walled garden” data with a grain of salt.
A “walled garden” refers to a browsing environment where users are restricted from seeing and controlling certain portions of the website, and it’s exactly how Facebook, Google, and Amazon Ads work. If a company doesn’t allow you to integrate your own analytics program, it’s safe to assume that the metrics are at least a little inflated.
While overall trends in the data probably aren’t wrong, the numbers are likely estimates rather than exact reflections of what’s happening with your ads. After all, these vendors have a vested interest in keeping you on board, but not necessarily that same dedication to reporting accurate, real-time numbers — especially if those numbers might prompt you to advertise elsewhere.
2. Validate from as many sources as possible.
In instances where it is possible to integrate your own analytics programs, you should absolutely do so. One measurement can have errors, delays, or inaccuracies; many measurements are far less likely to. That refers not just to different analytics programs’ measurement for the same statistic, but also to all of the statistics a single program offers.
While we’re still waiting for more information to come out in the current Facebook scandal, it’s likely that advertisers who focused primarily on the “average duration of video viewed” statistic were more impacted by the shady calculation practices than those who weighed every available viewer stat.
3. Listen to consumers.
Slate argues that Facebook’s false metrics covered up a single, deep dishonesty: that users wanted to see more video. This was the fundamental idea driving so many restructurings, hirings, and firings, but at the end of the day, it simply wasn’t true. As it turns out, people aren’t very interested in watching editorial news videos or video ads.
A lot of consumers said this — an Onion article even parodied the idea — but many brands took Facebook’s data at face value and continued to produce videos anyway. Data isn’t a crystal ball, telling us things about consumers that they don’t even know about themselves. Start with the consumer, trust the consumer, and build from there.