Tales From the Trenches: Month One
What's Working, What's Not, and Why the Bots Are Still Ruining Everything
Warning: I’ll often use “X” and “Twitter” interchangeably in my writing - including all derivative terms. Generally, I’ll use the Twitterised forms because, well, they are just better; Elon’s decision to kill off “Twitter”, “Tweets” and “Tweeting” baffles me to this day.
A Month of Trials, Tribulations, and Tweeting
As you might remember from last time, I’ve recently committed to dive head first into the choppy swamps waters of Twitter. I expected to hate it but, in truth, it has actually been quite a good laugh.
I was (and still am) very critical of many aspects of social media and I’m very unsure about whether history will look back on this “innovation” as a net-positive or not. All that being said, I don’t think social media is going away and so, in my view, it’s something we should spend time trying to improve, contain and iterate on.
Zooming in on Twitter specifically, I think that despite its myriad flaws, it seems to be the best way of ingesting realtime information on almost every topic; there’s a good chance that the majority of experts and public figures in your field of interest are on Twitter.
Obviously, there are HUGE issues with misinformation, echo-chambering and the whole “freedom of speech vs freedom of reach” dynamic but, for me, I believe the quality of information and discussion is still the best that is currently available.
Previously, my only real foray into social media was Facebook - this is almost the exact opposite experience to that of Twitter - it’s largely an alternative way to communicate (less effectively) with people that you already know. I very rarely learnt anything more useful than that my second-cousin now lives in Twickenham or that I’ve forgotten my auntie’s birthday (again). Hmmm, I guess it’s also good for sharing pictures.
So, in summary, although it’s a very flawed place, I’ve enjoyed the first month of this Twitter/X experiment and will continue to engage with the community there.
Progress Review
First off, I think having the metrics-based approach to finding out what works and what doesn’t has been really interesting and, without that, I’d be quite disillusioned already.
Whilst the numbers are useful, it’s obviously early days and drawing too many conclusions at this point is unwise. Let’s take a look though …
That don’t impress me much (oh, oh, ooh)
I possibly have the brains but I don’t yet have the touch - here is a snapshot from the first month (January).
I don’t want this piece to be a forensic investigation into every datapoint but this graph tells some believable stories:
Video content is watched, liked and shared a LOT
The peaks around Jan 5th and Jan 16th were both following the release of a small tech-focused animation (one about programming, one about AI)
X has claimed they want to become a video-first platform and I believe they’re encouraging this by boosting the reach of video content. This is good for small accounts who don’t have many followers (like me).
Until you have a large following, text-posting (not replying) is barely worth it
The impressions I get from posting under my own account are tiny and have barely received any engagement at all. Basically, until you become identified as a high quality voice - you just get drowned out by everyone else.
The plateaus around Jan 7-13 and Jan 22-28 were periods where I wasn’t able to create visual content and just made sure I stuck to my daily schedule by putting something out in text.
Posting images is slightly better but still seems to garner little attention for small accounts like mine. Memes generate poor results too.
Making relevant/funny/intelligent replies to large accounts can bring massive results. Almost every other peak on that graph is due to a large-account reply.
If you can reply quickly, reply quickly - I now set up notifications for large accounts that are relevant to my audience.
If there is something major happening in your space (e.g. DeepSeek v3 fallout) then consider preparing some words/ideas in advance that may be relevant. You can then roll these out if/when the right person also posts on the issue.
If you can be witty, be witty.
Unless you want to be hounded by bots, trolls and scammers - stay clear of extremely controversial topics or mega accounts (Musk, AOC etc)
Everything is relative
Whilst my account is still absolutely minuscule, I think I’m getting somewhere - I am starting to work out what works, what doesn’t and what else I should be trying. Here are some month-over-month metrics:
Now obviously, most of these 1000%+ gains are base-effects explained by the fact that I’m starting from such a low level. Regardless, I’m taking it as a positive for now, and over more months, I should be able to see a more informative pattern emerge.
You’ll notice the engagement rate (engagements divided by impressions) has actually dropped. Simply speaking, this means people are liking less of what they see than before (not good) but, after looking into this more, I believe it is due to two things:
Base Effects - I only made a handful of posts in December but received a few likes on them. I don’t think more than 10% of those engagements were real (see 2)
Bots - I had a massive percentage of bot-followers in December. They add ‘likes’ to everything and pollute the signal. I don’t care about vanity metrics or likes from bots. That’s not why I’m here.
Ah, yes! Bots! Let’s talk about them again
As I mentioned last time, at the start of this quest, I spent a while blocking any follower that I expected to be a bot - sadly, it turned out to be the majority of my following!
Although I wasn’t foolish enough to think that would be the end of it, I wasn’t sure how quickly these bots would replenish themselves - turns out, pretty damn fast.
I acquire new bot followers every single day.
I receive likes from bots every single day.
I occasionally receive full-blown (ai generated) replies from bots.
Right now, I’m able to stay on top of this - I block and report any account that I’m confident is a bot. It’s tedious but manageable and there are some tools that make it simpler. Hopefully this will become a higher priority for the X engineering team at some point - until then, it’s just one of those things.
Experimentation
One thing I am having some fun with is the experimentation - what works, what doesn’t etc. For example, I’m trying videos with background music and videos without music, posts with different capitalisations, posts with line-breaks, posts with emojis, posts at different times of day etc … I don’t have enough data to draw conclusions yet but I think that will come.
There’s a limit to how far I’m willing to go with this; I don’t want to completely sacrifice the content at the expense of engagement - that’s a road to nowhere and, as someone who detests clickbait, I really don’t want to become part of the problem.
There’s a lot of research on diminishing attention spans and why it seems to be a growing problem - much of that is attributed to the sheer volume of information that people are exposed to nowadays BUT much of it is also attributed to the format and delivery of that content too. It’s hard to say whether this is inherently bad or not - perhaps it’s a requirement of our inevitable transition to Homo Technologicus - but I want to exercise some level of responsibility and caution on this point.
I remember, many years ago, I started digging around on the topic of A/B testing for some e-commerce related stuff; I stumbled across this comment on a podcast that, until now, hadn’t quite landed properly:
What we know is that if you A/B test a website enough times, it will turn into a porn site. Because when you A/B test, you will end up with clickbait. You will end up with prurient images that people sort of click on in the short run because… ‘ehh.’ That is no way to do the work that you care about - Seth Godin
After seeing the other side of these metrics, I can see how it could become very easy to optimise for engagement, degrade yourself, your content and your online-persona in the name of clicks; I want to get that on the record here - almost like a reminder to myself to be alert to the risk and to hold myself a little more accountable. If this sounds like virtue-signalling, then sorry, I guess it kinda is.
Next month
I have genuinely enjoyed putting together the animations and want to double down on a sort of animated-infographic approach. Finding the data, charting the data, and animating the charts has been really fun and has given me an excuse to play around with some technologies I haven’t done much with before (manim and D3). Having content based on data also serves as a jumping off point for my own commentary and interpretation too - eventually this could lead to some interesting discussions with others.
I’ll continue to engage with large relevant accounts where it makes sense and, although it seems largely a waste of time, I’ll continue to regularly post on my own account for now.
This month I increased my genuine user following by 60% and (with a fair-wind) I think I can repeat this growth in February too.
Thanks for reading.
Automation Consultancy ⚙️🚀
Codefield helps businesses build and implement automated solutions for an extremely wide range of challenges.
Regardless of sector or size, we find that there are many, many places where leveraging Automation and AI can help businesses reduce (or even eliminate) the time spent on repetitive and/or tedious tasks, allowing them to focus on the more valuable strategic work that takes the business forward.
If you think there are areas of your business that could be automated or streamlined, then feel free to reach out for a friendly (and in no way pushy) chat at automation@codefield.io