Lies, damn lies and Facebook’s dashboards
FACEBOOK is in the news a lot these days. And for all the wrong reasons — censorship, filter bubbles, fake news and its effect on the US general election. The social network has been taking a PR pounding.
And it’s also having a hard time with the marketers and advertisers that contribute to its massive revenues. These amounted to $7.01bn in Q3 of 2016. But advertisers have been irked by a slew of reporting errors that Facebook has announced this year.
The social network’s latest miscalculations relate to live videos. Facebook forgot that users can log more than one reaction to a live video, and was misallocating a load of extra reactions.
Facebook also admitted it needs to be a bit more accurate about how it estimates audiences when advertisers are booking ads. “We’re improving our methodology for sampling and extrapolating potential audience sizes. In most cases, advertisers should expect to see less than a 10pc change (increase or decrease) in the audience sizes shown in the tool,” said Facebook. Is it cynical to suggest that most advertisers will see a decrease?
These errors aren’t isolated incidents. In September the social media giant admitted that it had been accidentally inflating the average amount of time people were watching videos for the past two years. It was excluding views of under three seconds from the calculations of average view times. The result? Facebook estimated that it had been overstating video view times by 60 to 80pc.
But wait, there’s more. A bug in page insights was uncovered where some repeat visitors to pages were counted twice. Facebook also found that the average time spent on its instant articles had been over-reported by 7pc-8pc since August of 2015. In Facebook’s Analytics for Apps dashboard, referrals were overstated by 6pc on average. And a feature called interest lists was retired when it was found that they could be used to artificially inflate profiles number of followers by around 5pc.
Sure, none of these are huge issues — and there’s no evidence that any of these errors were deliberate. But there are a lot of them. Facebook was very eager to stress that these issues don’t relate to paid activity — only advertising projections, measures of organic reach and audience engagement. But should marketers now be wary of Facebook?
You’re damn right they should. But no more wary than they should be of any channel — especially one that isn’t thoroughly independently verified. Here’s what WPP chief executive Martin Sorrell had to say: “We have also been calling for a long time for media owners like Facebook and Google not to mark their own homework and release data to ComScore to enable independent evaluation. The referee and player cannot be the same person.” It should be noted that WPP owns 20pc of ComScore.
It seems that Facebook is beginning to bow to this industry pressure. In November, it announced that it was “exploring additional third-party reviews to validate the reporting we offer partners”. That sounds a little vague, but will hopefully result in more reliable metrics.
Facebook has been very canny about how it courts masses of digital marketers. And its compelling dashboards — notwithstanding the recent errors — are only one part of the puzzle. It also created a fabulously effective false proxy for success: the like. Are we increasing market share? Doesn’t matter, our page likes have doubled. Are sales up? Who knows, but likes have grown by 110pc. People love us and our product.
But most advertisers have seen past the distraction of likes and are chasing more tangible commercial goals. Some are even voting with their feet.
Earlier this year, Procter & Gamble admitted it would be taking a step back from targeted Facebook advertising, having found serving ads only to specific demographics resulted in a dip in brand awareness and reach. Mars has also stated that reach is more important than targeting. And Coca-Cola’s global chief marketing officer Marcos de Quinto told a beverage industry conference in New York last week that Coke was rethinking its approach to digital.
In this context, where some of the largest global advertisers are rethinking their approach to digital, Facebook needs its reporting to be rock solid. The question is are there any further metrics missteps waiting to be uncovered in a system as vast and complex as Facebook’s?