Facebook new Insights are great, with a few things missing, but with fresh new stats such as “when your fans are online” whichÂ sounded great. Indeed, as we’ve shown in a previous post, timing is of the highest importance. We’ve decided to look at this new metric from a data perspective.
Our conclusion is clear: timing with most fans online is a low performing strategy. Indeed, if every user are online at the same time, they will share much more content, leading to an increased competition between stories in the news feeds… and much less chances for brands’ posts to be visible.
Our data perspective
Before digging into the stats, let us set the records straight: “When your fans are online” is not about your own posts metrics: it just says that somehow your fans are connected on Facebook.
Here is the study we ran: we took a sample of 5K pages of varying size. For each, we looked at optimal day and hour, as well as how much fans onlineÂ these optimum have.
In the end the results are disappointing: there are as many fans online on Monday, as well as on Tuesday, as well as… well every day of the week is all the same. So it’s not so useful.Â We’ve found the optimal day brings a whooping… 1.6% extra fans.
Every day has just the same amount of fans online (example for 9 random pages)
And interestingly, for 52% of the Pages, that “optimal-by-not-much” day is Thursday… that’s a lot of competition for an hypothetical presence in users’ news feed.
Note that the same apply for all Pages, whatever their size: the “optimal” day only brings 2% more fans, even for Pages with fewer than 1K fans.
On a side note, an interesting number came out of this study: on average, for any Page, 84% of their fans are online every day. It’s much higher than Facebook’s reported stats of 61%: reason is that for the 39% of Facebook users who don’t connect on Facebook every day, most of them are probably fans of very few, if any, Pages.
The hourly stats do provide some insights: Facebook doesn’t have a uniform presence of users over hours of day, but the problem is that most Pages have just the same peak: for nearly 30% of Pages 9-10pm is the “optimal” hour.
And that’s not of that much use neither: if all Pages have fans online at the same time… these users also get content from all of their friends, online at that same time (and content from competing Pages who also know they’re online).
Timing does impact posts’ performance
What matters is not when your fans are online but when they have news feed space for your post. To look at this we’ve measured the reach of posts depending on the time and day for each of these same 5K Pages.
And our conclusion is that timing does impact a page’s post performance. It’s not just 1.6% difference between days: it’s 30%. And by the way, it’s not Thursday more than half of the time: Every day has potential.
Note that Sunday is the best day for many Pages when in term of “When your fans are online-ness” it was the worst.
And from the hourly perspective, it’s not 9pm for everyone: there are plenty peaks available for users’ attention and engagement.
Note the 9-10pm time slot, where most fans are online, is far from being the best.
Agreed, this post’s headline “Why you shouldn’t post when your fans are online” is a bit provocative, but we do mean that you shouldn’t use the maximum “fans-online-ness” to schedule your posts.
We certainly won’t pretend that posts on Sundays and at 2PM are the best neither: Every page and brand are different!
Nitro, machine learning for timing optimization (and more)
You might claim that other solutions do dig into each brandâ€™s performance already, based on prior results.
But what if the brand publishes only on Friday? What if the brand posts funny videos every Monday and the content that matter most on Wednesday?
You can’t predict the best timing by only looking at raw stats. Topics, previous posts performance and a lot more variables impact the performance of a post. Hours and days are only part of the reasons behind the results gained by a Post.
In Wiselytics, we factor out more than 60 variables to isolate the true impact of timing. Then, and only then, our algorithms estimate the real impact of the day and hour.Â And since Nitro combines this modeling with proprietary data, it also estimates the potential of time slots you’ve never tested before.
If you’re interested, please join our beta.