Stats, lies, and video views: Why Facebook must be held to account

data
Mistruths have been a common theme over the last 12 months. From Brexit to the U.S. election, porky pies, or at least major errors on which decisive details have then been based, are rarely out of the headlines.

The majority of cases result in widespread outrage and demands for real explanations. It’s enough to make you wonder why the string of confessions by Facebook hasn’t been met with similar requests for information- so far the party line is ‘chalk it down to a bug’.
Facebook and Google already dominate digital advertising, with 85 cents in every new dollar spent going to one of these two firms. A duopoly, this level of dominance means companies are putting a lot of faith in the processes and methods of two private organisations. But what if that trust is misplaced?
Until a recent correction, since last May Facebook Page Insights have been overstating reach, which basically means brands thought their posts were getting to more of the public than they actually were. If you keep an eye on your own reach numbers you’ll realise how depressing the news is. Repeat visits were counted as new, unique page views, so in some instances the real figures may be a whopping 55% lower than original readings.
Facebook also managed to understate the level of video completions. ‘This may result in roughly a 35% increase in the count of video watches at 100%’, a network-written blog post said. In effect, they were declaring that less people were watching videos in their entirety than actually were, damaging perceived worth. Conversely, it was simultaneously overstating the average length of time people watched clips for, further skewing reality and calling our trust into question.
These are not small numbers by any stretch of the imagination, and the errors don’t end there. Publishers can now allow readers to consume entire articles without leaving Facebook itself, and the number of people finishing reading a whole article in this way has been overstated by 7-8% since August 2015.
This poses a lot of difficult questions for the wider industry. Smoking Gun PR prides itself on the use of data, both in terms of planning and reporting campaigns. If we suddenly declared our methods were not fit for purpose the repercussions would be significant. Clearly, though, client Facebook data has been reliant on the network getting its own Insights right.
We’re not the only ones feeling let down. Perhaps the most significant victim is advertising worth. Zuckerberg’s firm insists no paid-for measurements have been miscalculated, but the perceived value of sponsored and promoted content is based on the overall popularity of Facebook, and in particular its levels of engagement, including organic. Mess with any of the official numbers and the whole thing comes into disrepute.
Google has been investigated time and again over suggestions of bias in its search results. The algorithms at work are like a secret sauce; nobody outside Google can ever know the latest incarnation, if they did the company would risk giving rivals a blueprint of how to deliver equally effective search. The Facebook situation rings further alarm bells about online legitimacy, the likes of which can only be laid to rest properly if those responsible are truly held accountable, as the rest of us would be.
[Tweet “#Facebook measurement fails mean figures are over 50% out in some cases – #socialmediamarketers”]