By Kevin DeLuca
Opportunity Cost of Reading this ThoughtBurner post: $1.92 – about 8.7 minutes
I would be remiss to preach Daily Optimization without analyzing an activity that many people of my generation partake in daily – Facebook posting. I have heard many informal theories from family and friends about how certain factors affect Facebook post performance. “Make sure to post in the morning so that you’ll get more likes! Don’t post a picture too soon after your most recent post! Include a picture (or link) so that more people read your post! Don’t post in the middle of the week!” The list goes on, and people love to make up stories as to when and how you should post to Facebook.
The Internet wasn’t much help in figuring out which stories were right (shocking, I know). Some sites recommended posting in the morning[i], others said post at 5pm[ii], and some suggested the evening was best[iii]. Weekends are bad[iv], or maybe they’re good[v]. And of course you have to think about frequency – some places suggested you should only post about once every two days[vi], while others recommended closer to two posts per day[vii]. Pictures are supposed to get you more likes[viii] and more comments[ix]. Also, much of this advice was created for companies hoping to get more Facebook users engaged with their organization’s page, which I suspect doesn’t apply to everyday Facebook users.
Rather than relying on these ‘studies’, none of which use any basic controls in their analyses, I decided to collect data on my own friends’ Facebook posting patterns for two weeks to see if I could learn anything for myself. Specifically, I wanted to see if I could find out how links, pictures, frequency, time of day, and day of the week affects both likes and comments received by posts. I figured even if the results I find are not generalizable to all Facebook users, at least they would be applicable to my Facebook use, and they would probably be more representative of a normal Facebook user’s experience.
At the time of data collection, I had 759 friends. Out of these, only 642 friends had accounts that 1) were not deactivated and 2) allowed me to see their total number of friends (for control purposes). I collected data on all posts for a two-week period, and during this two-week period only 387 of my friends posted something. But they posted a lot (you know who you are). There were 1,822 posts total, or about 5 posts per person (though a sizable number of friends only posted once). The average number of likes was about 18, and the average number of comments was about 2.5. Forty four percent of posts included a picture, while 34% contained a link. The two tables below summarize these facts.
I recorded basic characteristics of all 1,822 of these posts and generated a few extra control variables. With this dataset, I ran regressions on the data to see if any of the rumored Facebook post optimizing strategies were true. The regressions help us find out what relationships are statistically significant, and give us an estimate of their magnitudes. Below is an easy-to-digest summary of the results, along with an infographic created by the wonderful Megha Arora to help you easily understand and share the results. To see more detailed results, check out the first part of the official ThoughtBurner Daily Optimization Project on Optimizing Facebook Posts. Commentary and daily optimization strategies follow the table and visual.
The biggest effect on likes is that of including a link in your post – posts with links in them had almost 17 likes less on average than posts without links. This is a huge effect considering that the average number of likes on all posts was only about 18. Posts with links had almost 3 fewer comments on them too, which may seem unexpected.
I also didn’t suspect that posts with pictures in them would have fewer comments (about one comment fewer on average than posts without links or pictures). I had also guessed that the effect of pictures on likes would be positive (but, alas, it is insignificant). This could be because pictures don’t actually make posts more likeable on average or because my friends are so good at non-picture posts that they receive the about the same number of likes as picture posts.
The more friends you have, the more likes and comments you have on your posts. This one seems obvious, but it acts as a good control and as a good check. If we had found that people with more friends had lower numbers of likes or comments, then we might be worried that there was some lurking variable contaminating our results. Also, we can be more confident in other significant effects because we know that the total number of friends is being controlled in the analysis. For every 91 more friends someone had, they had 1 more like on their posts on average and for every 1,000 more friends someone had, they had 1 more comment on their posts on average.
Figure 1 below shows the average number of likes someone had plotted by the total number of friends that person had. You can see that the linear trend line is slightly sloping upwards, though it isn’t too steep.
Males in the sample had about 4 less likes on average than females. Looks like my male friends are bad at Facebook posting (sorry boys). Or maybe I’ve found evidence of a gender-based like discrimination on Facebook?! No. The more likely scenario is that my male friends’ posts are correlated with some unrecorded characteristic of posts that negatively impacts the number of likes.
In order to see if posting at certain times of the day affects the number of likes a post receives I first divided the day into five periods: Dawn (2am-6am), Morning (6am-noon), Afternoon (noon-5pm), Evening (5pm-9pm), and Night (9pm-2am). I used these five times of the day because I believe that these five periods represent natural divisions of the day, at least among my group of friends. For example, most of my Facebook friends wake up between 6am and noon, so any time of the day effect that occurs between these hours would be interesting and possibly applicable to most of the friends included in the analysis. The same could be said of the other time periods.
The only time of day effects that were significant were those on likes; posts in the morning, afternoon, or evening all had a significantly higher number of likes compared to posting in the middle of the night. There is slight evidence that “evening” hours are best, because those hours had the largest positive effect. However, I can’t say with confidence that the effect is statistically better than posting in the morning or afternoon. In terms of comments, the time of the day did not have any significant effects.
Figure 2 shows the average number of likes by time of the day. While it looks like the average number of likes is also much higher at night compared to dawn, the difference is not statistically significant after controlling for other factors. Specifically, the percentage of posts with links is about 39% for both the morning and afternoon times of day while the percentage of posts with links at night is close to 32%; since links have a very strong negative effect on the number of likes, this means that morning and afternoon posts have about the same number of average likes as posts at night despite the fact that more of those posts have links. This is a good example of the sort of nuance that I am afraid other analyses of Facebook posting are missing.
The day of the week did not significantly influence the number of likes received by a post. There is no evidence that weekends are better (or worse) than the weekdays. However, posts on Mondays had about 1 comment more on average. Maybe people are trying to distract themselves at work at the beginning of the week, or maybe they comment on things that happened over the weekend. Last, the total number of posts within the two-week period negatively impacted the average number of likes a person’s posts received. For every additional post within two weeks, that person received about 0.5 likes less on average on their posts.
It is important to remember that just because I’ve found these significant relationships doesn’t mean that I’ve found causal relationships. There could be any number of reasons that we observe these effects. The control on males is a good example – it’s probably not that being a male inherently causes your posts to be less likeable. Also, consider the effect of ‘total friends’ on likes: for every 91 more friends someone has, they had 1 more like on their posts on average. This could be because they have more friends to see their posts, but it could also be the case that people with a lot of Facebook friends are just better at creating likeable Facebook posts (higher ‘social intelligence’ or something). Since I didn’t want to impose a subjective ‘quality of post’ measure, we can’t discern between these two possible explanations.
That being said, it doesn’t mean we can’t use this evidence to dispel or give weight to theories about optimal Facebook posting. If someone were to tell me that including links in your posts would increase the number of likes you got, I’d be pretty skeptical given the results of my friend group analysis. Using the information above, it is pretty easy to figure out some general strategies to improve and optimize your Facebook posting.
Again, to see more detailed results, including regression results, more visuals, and more commentary, check out the first part of the official ThoughtBurner Daily Optimization Project on Optimizing Facebook Posts.
Optimal Facebook Posting
My assumption is that you are trying to maximize either average likes or average comments on your posts. Other objectives will require different strategies. First things first – don’t include a link in your post. My theory is that links are too specific in their topic and appeal only to a limited number of people (namely, those who happen to be interested in the link’s particular topic). Only some of your friends will take the time to read the link, and even then only a percentage of them may actually like the post. On the other hand, picture posts or status update posts appeal to more of your friends because, presumably, they all have a general interest in your life (y’all are officially friends on Facebook, after all).
The more friends you have, the more likes you’ll have. This is probably due to some sort of exposure effect – if your posts are all of a similar quality where some percentage of your friends will like them, then increasing the number of friends you have will increase your number of likes even if the quality of your posts doesn’t increase. Maybe you should consider accepting those few friend requests you haven’t responded to yet (you know, the ones that everyone has from those creepy random strangers). In fact, you might even have reason to suspect that the randos will like your stuff more than your average friends. #untappedpotential
Don’t be male. Just kidding – you can be male but just try to post the way my female Facebook friends do. For some reason, the stuff that my male friends are posting is just less likeable than what my female friends are posting. (Come on guys, get it together.) Or maybe both guys and girls are just more prone to liking girls’ posts.
No need to go out of your way to include a picture in your posts if you are trying to maximize average likes. And if you are trying to get more comments, you’re better off just posting a normal status, since picture posts actually had significantly fewer comments than non-picture posts. Also, posting on Mondays might help you score an extra comment, possibly because other people are catching up on your weekend.
Don’t post too often. For every additional post someone posted, that person received 0.5 fewer likes on their posts on average. This could either be because people who post a lot have lower quality posts, or because their friends get annoyed with their frequent posting and stop liking their posts. So make sure you are not annoying and don’t post things you know are low quality.
Last, make sure you post during the day when people are actually awake (as if this wasn’t obvious). To be safe you could make sure to post in the evening (after 5pm and before 9pm), but I can’t guarantee this will be any better than posting in the morning or afternoon.
Remember that my advice is based on data collected over a two-week period. This optimal strategy is probably more applicable to people who post fairly regularly (at least once every two weeks). It could be the case that a better way to maximize average number of likes is to post a super high quality post once a month. This could increase your average number of likes, though it might decrease the total number of likes (which could also, in theory, be your objective).
If you follow the average like-maximizing (optimal) strategy rather than the worst possible like earning strategy, your estimated average likes over a two-week period will be about 23.61 likes higher, assuming the content of your posts doesn’t change. I doubt you’re doing everything wrong, so you can probably think of this as the maximum like effect you can achieve by posting optimally.
I hope this helps clear things up a bit. Depending on how similar you think your group of friends is to mine, the results might be applicable to your own posting. These results also show that the advice you can read online doesn’t apply to everyone, and that to really know your optimal Facebook posting strategy you have to consider your specific group of friends.
A little preview for next time: an analysis of my own personal Facebook posts over all time leads to slightly different results. Like I said, what works best on average doesn’t always work best for an individual. We know in general what might be an optimal Facebook posting strategy, but this isn’t the same as knowing what works best for me (or you) personally. In my next post, I’m going to analyze my own Facebook posts starting from when the like button was introduce in February, 2009 and create my own individualized optimal Facebook posting strategy.
© Kevin DeLuca, Megha Arora and ThoughtBurner, 2015. Unauthorized use and/or duplication of this material without express and written permission from this blog’s author and/or owner is strictly prohibited. Excerpts, links and citations may be used, provided that full and clear credit is given to Kevin DeLuca and ThoughtBurner with appropriate and specific direction to the original content.