Facebook engagement data gets more accurate — and bigger!
We’ve observed this week Facebook improved their reporting of engagements for articles that in some cases results in a 10x increase.
CNN got a big viral boost from their coverage of Donald Trump’s bizarre cabinet meeting this week. They earned over 300,000 engagements on Facebook, considerably more than almost everything else they covered this year.
That 300,000 figure is higher than usual as a result of a change in Facebook’s reporting methodology.
Facebook has been making a lot of improvements to their data. They have been exposing improved breakdowns of engagements in their API. And in December they acknowledged some discrepancies in the way their engagement counts were getting reported which they intended to fix.
“We’re improving our methodology for sampling and extrapolating potential audience sizes.”
The latest update appeared earlier this week. The impact was felt instantly, and they explained on their blog that this would be the new normal,
“Going forward, some high traffic publishers will see an increase in their engagement metrics. This fix applies to new URLs that are shared to Facebook and will not change metrics shared before June 9, 2017.”
You can see the effect of the change on Kaleida’s engagement history graph for that CNN article. Just before 1am London time on the 13th June, it suddenly jumped from 9,274 engagements to over 90,000. Later that day, it dropped again to what looks like its original trajectory, recovering back to over 270k engagements a few hours later — presumably because Facebook had to temporarily revert back to their old tracking system.
The reason Facebook described the engagements count figure as an estimate in their documentation in the past is because of this sampling approach to reporting engagements. The API didn’t reflect the actual numbers of engagements but rather a sensible estimate from a sample using statistics.
We’ve observed a similar pattern of increase in engagements this week across all of the publishers we track. Overall, just before 1am on Tuesday morning, the total engagements increased by 77%: the 10x improvement seen by the CNN piece is at the high end of the range.
Publishers everywhere are likely to see a considerable increase in engagement counts for their articles. It’s unclear if it will affect one segment more than another, but have no doubt the shape of the market is going to look different in some unexpected ways.
We’ll do our best to put these changes in context with what’s happening across the wider media ecosystem. Always feel free to pop a question our way: help@kaleida.com.
