Facebook’s inflated metrics shake the foundations of digital advertising

Imagine, if you would, that you work for a company that sells bananas. 

Each month your boss gives you targets for selling bananas. In the last month he sets you the target of selling 100 cases of bananas. But you only sell six. What do you do? If you come clean he will fire you for being rubbish at selling bananas. That is your job after all. 

But here’s the thing. The boss never checks to see how many cases of bananas you actually sold. So, you know, you could just tell him you sold a bit more than you actually did.

What would Mark Zuckerberg do? Hard to say, but we now know what Facebook in Australia would do. They would say they sold 100 cases of bananas.  And they would collect their salary and do the same thing next month too.

The only problem would come when someone who does not work for you starts to count how many boxes of bananas you sold. 

In October 2016, following a host of stories about how Facebook have been marking their own homework, ratings company Nielsen altered how they count a video “view” on Facebook. According to Facebook, if you watch a video for more than three seconds on the platform, that counts as a “view.” But Facebook didn’t count anyone who didn’t watch or watched for less than three seconds, massively inflating average viewing time.

Nielsen started counting everyone. Australian video streams dropped by 94 percent over the course of a single month.

Nielsen measured total Facebook streams of 8.6bn in July 2016, 12.5bn in August and 9.9bn in September. But in the three months after changing the measurement method, numbers plummeted to 560m in October and November and 580m in December. 

Facebook CEO Sheryl’s Sandberg said that viewing times – the ones Facebook had been misrepresenting – didn’t matter anyway. Viewing times counted for less than “sales effect”, she said:

“What matters the most is the A/B test that these people saw ads on Facebook and Instagram, these people didn’t, and here’s the sales lift.”

Except of course, maybe they didn’t see the ads in the first place, and instead did what everyone does, which is thumb past them as fast as they possibly can. Could it happen here? Oh please. It already is. That’s why P&G just pulled their advertising off Facebook and other digital networks until they could get third party verification for what must be dodgy numbers.

Other critics have been even more vocal. The Ad Contrarian said:

“Despite the overwhelming evidence that all the data we are given about online advertising eventually turn out to be lies and nonsense, marketers and agencies just can’t seem to get it through their skulls that they’re being screwed blind.”

So what does this mean for publishers? I think that if you are posting content to Facebook in return for traffic you are trapped in a losing game. 

As P&G’s actions have pointed out, the entire bedrock of digital advertising, which in the end rests on the idea that people see the ad you tell everyone they are seeing, is in question. (In the first half of 2016, Facebook took 43 percent of all digital advertising growth in the US and Google took 60 percent. Everyone else lost 3 percent)

With the ads views in question, publishers’ ad-driven business model must also be in question. 

So publishers need to stop thinking Facebook is their pal (Disintermediation klaxon! Facebook have just hired an MTV producer to start making original video content itself) and get smart. Take a leaf out of the book of the music business or cable TV in the US: bind together and licence all of your premium content to the highest bidder. 

Then, as one observer has noted “let all the other social media platforms be a sewer of cat pics and fake news.” ESPN does not let Time Warner crawl its content for free. Baidu isn’t allowed to crawl Alibaba’s content.

Publishers need to get smart. Anything else is just bananas. 

By |2017-02-10T00:01:00+00:00February 10th, 2017|Analysis|Comments Off on Facebook’s inflated metrics shake the foundations of digital advertising

About the Author: