Facebook's engagement rate flawed

 

The new engagement rate percentage % from Facebook is great but watch out for a few flaws.

In deriving this % Facebook sums up unique user engagement (we will talk about engagement briefly below) and divides it by the total unique user reach for the post and times it by 100 to get the percentage.  The problem is that there are flaws in its derivation of the engagement figure and even a more serious flaw in its derivation of the reach figure.

Engagement

When it comes to engagement the definition by Facebook is probably different from what you have in mind.

Firstly, as far as I can tell from the Facebook Insights panel interface, it also seems to take into account negative engagement with a post, such as hides and unlikes. Now, as this figure is usually minuscule, we can live with it.

Secondly as of yet, it doesn't take into account replies to comments, likes on replies and likes on comments, video plays etc. See a post by Ben Donkor on this. Now interactions such as video plays are categorised into 'consumption' and not 'engagement' but I think, the engagement figure on insights should include these as well.

What if someone has put a link directly to your 'post' from their Facebook page. Will likes for their post be counted as your likes. No, Facebook doesn't count these likes, as the user who has posted the link may have altered the photo or text accompanying the link. So this is understandable from Facebook.

But anecdotally I would suggest it is fairly common for Facebook managers to post links to a post on someone else's page without altering the content. As the originator of the content and the post you will be missing out on seeing and knowing about those likes. I would say Facebook should accrue those likes to the originator of the post if the post is exactly the same as the one being linked to.

Reach

But my main gripe is on the reach figure. As already mentioned once Facebook sums up the so called 'engaged users', it then divides that figure by the 'unique reach' for the post, to give you an engagement percentage. This % will be used by clients and marketers to judge how engaging a post has been. BUT there is a serious flaw in the way the reach figure is derived.

Have a look at the following snapshot from one of our clients' insights panels:

Now that's a an engagement rate to be proud of

If Facebook is counting the UNIQUE number of engaged users and dividing that by the UNIQUE number of users reached, then how can it be that we have an engagement rate of over 100% for the post above? I have actually come across an engagement rate of 200% for a post!

Is it that the unique reach data is delayed compared to the engaged users figure? No, that's not what's causing this reporting flaw. The reach figure may be slightly delayed, but the engagement rate shown above is from a screen shot after the dust has settled. i.e. after a few days.

How can that be?

Well firstly it's important to know the following:

If you post a link from your Facebook page to a post from another Facebook page, then any likes or reach from your Facebook page will not be apportioned to the post you are linking to. i.e. when a user on your Facebook page sees your post or likes the post on your page, the originator of the post (i.e. the post you're linking to) will not be apportioned with the reach or likes of these users. 

However, if the user clicks on your link post and goes directly to the post you are linking to, and clicks on the like link, then the like will be apportioned to the originating Facebook page. The problem is that the reach which resulted in these likes are not being apportioned.

For example, 20,000 thousand users may see the link post on your page, and say 1,000 may click on the post and like it. Another 50 may like the post directly on your page without clicking on it first. What would happen is that 1,000 likes would be apportioned to the post you have linked to (not the 50) , but the 20,000 reach would not be apportioned. This gives an inflated figure for the engagement rate and significantly underreports the reach of the post. Experience shows that users who like picture posts, are much more likely to click on the picture first and then like it. So by not taking the reach into account but registering the resulting likes, Facebook is significantly skewing the engagement rate figure.

This tactic can actually be used to increase the engagement rate of a post by unscrupulous marketers. Facebook needs to find a way of remedying this. Clients would see a high engagement % as the reach is in effect being under reported. 

The current setup by Facebook functions properly if your posts are seen either via your Facebook page or via Facebook advertising. But if other Facebook pages link organically directly to your posts, then the engagement rate statistic from Facebook is no longer accurate and becomes seriously flawed.