If you missed it, in the first part of this post, I asserted that if you write something near and dear to your heart — something you really want to be seen and perhaps even trigger some discussion around — and then post it in the weekend, you’re potentially in for some heartbreak. This time around, I come armed with data! In a reply to Magi, I mentioned that I might just be curious enough to pull together the data set to show what I was talking about re: a lower degree of engagement with posts published over the weekend.

Turns out:

  1. Yes, I was curious enough
  2. But also ‘Yes’, in that it was every bit as painful to pull together as I thought it might be.

Getting the Data

I started looking at the WordPress export options. I found right away I could export the full set of post content as an XML, so we were off to a good start.

Unfortunately, while this table contained a lot of metadata and references, it did not contain either a count of comments per post or the count of likes.

I could possibly have eventually worked around the comment count limitation, as it did include the raw comments themselves as well as the comment IDs, but pulling in this data expanded each post’s data set to the number of rows required, and I decided to forgo this approach.

If I had found a way to export Views and the Like count, I might’ve continued faffing around with the comment data contained in the XML — but View and Like data is held outside of my server, on WordPress’ own servers, being part of the capability that JetPack offers. Unfortunately, there isn’t any native ability to export this data from JetPack either. Possibly there are other plugins that will let you do this, but a very cursory Google kept resulting in the answer of, ‘but why would you even want to do this?’

Grr.

So I knew I was in for a manual process.

From the XML, I ended up creating a table in Excel referencing only the post title, URL and published date. After some fiddling with the date format, I managed to get Excel to understand the WordPress output and filtered it down to look at 2020’s posts.

If you’re wondering why just 2020, it’s because I didn’t even work out how to enable likes properly until late into 2019. I was also still finding a footing as a blogger then. As for cutting off 2021? Well… three things. 1) I just don’t want to manually trawl through that many posts for the data, 2) 2021 posts haven’t necessarily settled in engagement stats yet, and 3) I believed that a whole year of results should be enough to paint the story. ;)

Next stop: The all-time post stats page, where I CTRL-F’d for titles, got their view count and then opened the posts into a new tab to obtain both the comment and like counts.

152 rows of post data later, the 2020 data set was complete.

The Results

Distribution of posts published by day, across all of 2020

Initially, I went with just displaying raw numbers and honestly, that approach told the story I expected it to. Strong drop-offs were visible in the weekend for all engagement metrics. But then I took a quick look at the distribution of posts published against days of the week and saw that this wouldn’t be a very fair way to tell the story.

While I have a reasonably even distribution of posts across the weekdays, clearly, even in 2020, I had some inkling of the issues with weekend posting and kept it down, relatively speaking.

So for my results to be accurate, I would have to factor this in; otherwise, I might as well go see about running as a politician.

I landed on representing everything in percentages1, alongside the post-distribution. Each day’s metrics become relative to itself. All else being equal, if a day represented, for example, 12% of the posts made in 2020 — it should also represent 12% of the year’s engagement metrics.

Days where the engagement metrics represent a higher percentage than the posts that account for them can be considered as overperforming. Inversely, days where their engagement metrics fall below the percentage of posts are underperforming.

Of course, it all has to balance out to 100% in the end, so if a day does under- or over-perform, the next question is, where did those stats come from or go?

This one graph represents altogether too much time

Important note; the data set is orientated around when the posts were published. It makes no remark on when the engagement happened. e.g., likes on posts published in the weekend may have trickled in the days following publication.

Sunday appears to be the worst day to publish something you would like engagement on

It’s down on all measures, but in particular comments.

If you can hold your post for just one more day and schedule it to go live on Monday instead, you’re going to very likely have a much better chance of spawning chatter. So if it’s a topic near and dear to your heart, or if you want to get the opinions of others, hold!

Saturday isn’t nearly as bad as Sunday, but it does hold the dubious honour of being the only other day of the week to underperform on all metrics. So again, if you’re at all able — hold your post and schedule it for another day.

Mondays and Fridays are particularly strong days to publish

This makes intuitive sense to me.

Mondays sometimes need a bit of an easing back into the workweek. By Friday, the checking out process has begun and browsing is on the rise!

Any surprises?

I was a little surprised that Saturday and Sunday weren’t even further behind, truth be told. Although it’s probably worth noting, I have not removed consideration for the Humble Choice posts which I predominantly publish over the weekend and garner views year-round, if not comments or likes so much.

So I would advise you to potentially treat the Views metric for both Saturday and Sunday as somewhat suspect, in that they might be overstating reality vs. a more regular post. Just more reason to consider holding your post for a scheduled release on Monday. :)

Of course, if this doesn’t suit you — ignore the lot of it!

I’m not as zealous about this as two posts on back-to-back days might suggest, I swear. I just like playing around with data and testing my perceptions with it where possible. It isn’t something I do very often here on the blog, but every once in a while… *gestures vaguely at post* …this happens.

My conclusions, in this case, remain unchanged — if you are specifically after engagement with a particular post, don’t post it over the weekend.

Write it whenever you want — or can! — but schedule it for a day it can do the most good for you.

Unless, you know, you don’t want to. Then as you were! As long as you’re having fun, it really doesn’t matter. :)


This was posted during Blaugust 2021, the annual blogging event hosted by Belghast. Blaugust is an event aiming to welcome new blogger blood into the fold and revitalise those who’ve been at it a little longer.

The Blaugust Discord is still available to join in, year-round!

Footnotes

  1. As measured against the overall totals, i.e., just under 21% of all views in 2020 were from posts published on a Friday.