Relative Impact shows which of your posts performed best.

It also shows how much better or worse posts performed, so you know if a post was a slight improvement, or a runaway success.

It's our custom algorithm that takes PTAT (see What is "People Talking About This"?) and adds in consumptions and a few other metrics.

Consumptions are important to include because we discovered Facebook appears to factor consumptions into EdgeRank, meaning posts that get clicked a lot get seen a lot.  

There's a few other metrics we thought were important to include, and we're still tweaking it, so if you find posts that you think have an unreasonably high or low Relative Impact, let us know.

Relative Impact is not a fixed number--it is dynamically recalculated based on the group of posts you're examining. This lets you see how much better or worse a post performed compared to the other posts you're looking at.

So if you're looking at the last week of posts, and then expand to looking at the last month of posts, the Relative Impact will be recalculated to show how a post compared to all the posts that month, not just that week.



The backstory:

We get asked all the time, "How do I figure out which of my posts performed best?"

The problem with ranking by any of the engagement metrics such consumptions/likes/comments/shares is that some types of posts generate a lot of consumptions, others are magnets for likes, and still others mostly result in comments.

You can check a post's PTAT, but that has two flaws:

1) PTAT doesn't include consumptions, and as we discovered, Facebook appears to factor consumptions into EdgeRank, meaning posts that get clicked a lot get seen a lot.  

2) PTAT weights all story types the same--a like matters just as much as a share, even though they have very different meanings to users and to EdgeRank. (In North America, people are much more likely to like something than to share it. In South America, it's the opposite--people tend to share things rather than like them.) 

Since Facebook shows what they think are better performing posts to more people, we thought about trying to piggy-back on their hard work of predicting what content is successful by sorting posts solely by Reach  But post reach includes a trending component, meaning that if your previous few posts performed above average, Facebook bets that your next post will perform well and shows it to more people, so even if the post is a dud it will still have high reach. 

So we've written our own algorithm that factors in PTAT, Consumptions, Reach, and a few other metrics.