Last week, I noticed that the links I was posting to my Facebook wall were getting significantly less play than the posts I was sharing from Facebook. For instance, if I shared a post from, say, Mashable, the likes, comments and shares were pretty solid. But when I posted a link to page that wasn’t already being shared by a ton of people, all I was getting was… crickets. Given the state of organic reach on Facebook, (you can read more about it here and here) and my understanding of what types of content publishing Facebook is trying to monetize vs. the way it looks at simple status updates, I dreamed up a quick and easy little experiment to see the extent to which Facebook was throttling down updates (unpaid) linking to original content compared to identical updates with no links to original content.
Using basic A/B testing, I created two almost identical updates. Here’s A:
Here’s B:
Note that A (1 of 2) has a link embedded in the update. In B (2 of 2) the link is in the comments.
I posted both updates within a minute of each other to make sure that the audience size would be the same for both updates, and checked back on them after 30 minutes, then every hour for 5 hours.
The question I had was this: would both updates show up in people’s feeds, and if so, would they be equally visible?
Here is what I found:
1. Update B (2 of 2), the one with no link to original content, had 14x more likes than A (1 of 2), which had the embedded link.
2. Update B (2 of 2), the one with no link to original content, had 44x more comments than A (1 of 2), which had the embedded link.
(The comment count in A (1 of 2) should actually be zero since, by their own admission, the 2 comments posted were a side-effect of the B update, so the 44x might actually be 89x.)
Now… let’s not get ahead of ourselves. This isn’t a “case study” by any stretch of the imagination. For starters, it was a quick little test that produced nothing more than anecdotal data. It’s barely a data biopsy and it isn’t really statistically relevant on its own. Second, other factors might be at play here. For instance, reversing the order in which the updates were posted might have affected the results. (I invite you to test that theory if you want and share your data.) Third, the test only looks at 5 hours. The visibility of both posts could change over time. This is definitely not a study about content lifecycles. Fourth, as Charles Lau astutely pointed out, my own comments in response to reader comments might have also contributed to improving update B’s visibility. (Engagement driving more engagement, and so on. There probably is a cumulative effect at play as well.)
Having said all of that, the difference between interactions for updates A and B is so significant (and consistent with my original observations) that there appears to be something going on with how Facebook decides what type of free content your audience gets to see.
What does it all mean?
1. Organic reach on Facebook isn’t actually dead. BUT… both the type of content shared on Facebook and the manner in which that content is shared, have a massive effect on how many people in your community will see it.
2. Facebook does seem to be manipulating the distribution/visibility of your content based on how it is shared and where it comes from, and I don’t think it has much to do with editorial quality or topic relevance. (Remember that the copy was identical in both updates. The only difference was that one contained a link and the other didn’t.) So the intent here doesn’t seem to have anything to do with connecting an audience with content that most appeals to them (which would actually be pretty cool, if only to move the signal to noise ratio in the right direction). It seems to have more to do with content monetization. Again, Facebook’s shift away from earned media and towards paid media isn’t exactly new.
Admittedly, there could be something else going on, but I just don’t see what it could be. Unless someone has a better theory, it seems that Facebook’s message here might be this: “If you want to use this platform to help you get eyeballs on your content, even eyeballs you have earned, you need to pay for it now.”
Think about that next time someone asks you about ‘earned media.’ Things are a little more complicated than they were in 2011, when I wrote Social Media ROI. (The principles in the book are as true today as they were when I wrote it, and that isn’t going to change anytime soon, but Facebook is far less of an earned media platform today than it once was, and it is important to be aware of that.)
For what it’s worth, I hope this little test will give you some insight into where the content you are posting on Facebook is and isn’t going, and why.
(And now, to figure out ways around this that don’t necessarily involve chucking coinage into Zuck’s fancy toll basket…)
Cheers,
Olivier
PS: Sorry to be the bearer of bad news. I just thought you should know.
* * *
Want to help improve business through your digital programs? Pick up a copy of Social Media ROI – Managing and Measuring Social Media Efforts in your Organization. The #1 Social Business desk reference for executives and digital managers.
(Now also available in German, Italian, Korean, Japanese and Spanish.)
13 Comments
Cool little test. And your results do make a lot of sense. Seems as if Facebook understands the platform is primarily used to share life updates (and I use that term loosely) with “friends”. In which case, they’re more visible. And any posts with links are viewed as promotion, so they’re throttled. Even more-so now that FB is moving toward paid posts. I’ll have to give this a whirl and see what happens.
Yep. Joshua Garity and I also think that Facebook further penalizes content shared by you from your own domains (whether it’s done through a personal FB page or another kind of FB page). That’s also something to look into.
Thank you! I thought there was an update from Facebook itself in August that it would prioritize embedded links over posts with links in picture captions and the like. I started using more embedded links as a result and saw poor engagement for those posts. So, perhaps I misunderstood the August announcement?
Interesting experiment indeed. I echo Charles’ view on engagement driving engagement – starting with your very own (first) comment. Another question in this regard – are the two posts as independent as we may think? I would expect the Facebook newsfeed algorithm, when it sees two posts published within moments of each other, and with almost identical content, to assume that the 2nd was an edited version and probably the one the user intended to really publish (i.e. the user didn’t edit the first post as he ought to have) and therefore give the 2nd more airplay. Only one way to see if this holds true – like you rightly observed, reversing the order in which the updates were posted might have affected the results.
PS: I’ll take you up on testing this and report back with what I see.
PPS: I just noticed on your wall that you’d shared the same link a few hours before posting version A; do you think this might have affected the results: content duplication.
All good points. Let me know what you find out.
(And yes, I had shared it a few hours earlier and the post received almost zero attention, which is what prompted me to do this little test. I was starting to notice that Facebook was throttling down posts that link to domains Facebook knows to be mine.)
First time here and really appreciate the clarity of explanation and your Testing. Sharing with my “master mind” group sand we will all TEST. Again thanks for the information
Hey Olivier, I’m reporting back on the experiment I mentioned. So this is interesting. I reversed the order of the experiment as you suggested – with “1 of 2” referencing the link (a link to a blog post on a recent report we partnered on) in my 1st comment and “2 of 2” drawing the reference in the post itself. If my hypothesis of “given two very similar pieces of content published within moments of each other, the later one will get more airplay” was right, the 2nd post should have done much better but nope. The first did waaaay better – like your 2nd post did. So, it does seem like your initial inference was right. FYI, the 1st one (with link in comment) got 162 likes and 29 comments while the “with a link post (2 of 2)” got a relatively meagre 36 likes and 1 comment (me, thanking my friends for their help). In case you can see it:
1 of 2 is: https://www.facebook.com/lakshmanan.lux.narayan/posts/10152438610270981?pnref=story
2 of 2 is: https://www.facebook.com/lakshmanan.lux.narayan/posts/10152438611505981?pnref=story
Not surprised. I did the same test you did (switched the publishing order of A and B) on a different account and saw the same results. The order was irrelevant. The structure of the post itself was the determining factor.
Your second post cannibalized the reach of your first post. Do the exact same experiment again, but switch A and B, and you’ll get the same results, with the second post getting better reach and engagement.
Oh, and I see your reply to Josh Garity above, but I don’t see Josh’s comment. What’s up with that?
Yep. Regarding methodology, you’re right. But I tried it again in reverse with a different account (and without any prompts this time). Same results.
The time factor might responsible for the change in both post’s reach because it is very crucial for getting your visitors at the right time to your post when they are online, as now on Facebook the news in the users feed ream,in for minimum time and changes frequently.
You’re right but I’ve repeated the test at similar times on similar days and the results are consistent with this test. So… while time can be a huge factor, it isn’t in this case.