AI-Generated Posts in X Communities: 4 New Followers Per Day, 30 Impressions Average
Why I'm keeping this "failed" experiment running anyway
I posted 147 times in X communities in the past few weeks using an AI content repurposing system I built.
First thing I’ll say: this campaign was mediocre as far as results go.
I don’t know if it’s something I’ll continue. I’m certainly looking for better options.
But it’s ok for certain situations. It’s not hard to run and gets some results.
So despite it being a failed experiment in my terms, some of you may still want to know about it.
The results:
How I built this system
It’s a lot like my Reddit system, which I detailed out here:
Though I have made some particular improvements on the content side.
First thing I did was go through all of the X communities that I wanted to market myself in. I took a wide breath here:
Solopreneur community
Startup communities
Building public communities
Indie SaaS communities
B2B SaaS lead gen
I had seven communities in total.
Then I went through each community and collected content that did above average. If the average post got two or three likes, I was looking for stuff that got 20-50 plus likes and I saved that into a swipe file.
As you’d see in my Reddit campaign post, the system takes my source content and repurposes it to write social media posts based on what’s in that swipe file.
It also aligns the new post with the target audience on the community. It uses a very basic target audience snippet that only has pain points and dream outcomes.
AKA the content creation prompt has 3 variables:
The source content to base the new content on
The example post that performed well on the channel
A target audience snippet with pain points and dream outcomes.
The prompt runs automatically for every example post in my swipe file when I drop a new piece of source content in.
With every new piece of source content, I get around 80 X posts.
The editing challenge
I ran 4 new pieces of source content, which left me with 300+ posts to revise and schedule.
And yes, revising is critical. These posts are not good enough to publish.
I took the time to build out a few scheduling automations to get rid of that problem (this saved me a ton of time, but was complicated to build).
With an hour of editing, I can get out about 20-25 posts.
Not bad at all, but you do have to be pretty consistent about it.
The results
The average post is something like 30 impressions. So not very good as far as posts go.
But like I said I can do 20 per hour. So I end up with 600 impressions per hour of editing work.
That’s not horrible.
Some days I got 10-20 impressions on my website attributed to X. Which really isn’t anything to to write about (even though I am).
As I said in the beginning, these are not results that would typically encourage me to keep going. But there are some benefits.
The silver lining
The core advantage is the new followers.
The campaign does a pretty good job of putting me in front of the right people.
For example, most of my impressions are from the US.
And most of my impressions are from new people. Which would not be the case if I were just posting in my feed.
Given those 2 data points, this is more beneficial in my eyes than:
Being a reply guy (lots of foreign impressions… which is not useful for my business)
Posting on my feed (which wouldn’t get me in front of enough new people)
How I’d optimize this
If I decide this campaign is worth optimizing, here is how I would do it.
Fortunately, I collect the data on post performance based on the prompt that created it.
First, I would go ahead and delete the bottom 20%. Or keep going as is until I have more significant data per prompt and get rid of the bottom half.
That would look like deleting 40 prompts. I’d do some more research and add in 20 new ones.
Every month or so I’d do this process again. Delete the bottom performers, add a few new ones, hopefully getting from 40 impressions per post to 100.
That alone would bring me to 2,000 impressions per hour and is achievable.
Next, I would complain very loudly about postpone.app not having a built it feature to toggle the “also share with followers” feature that X gives you when you post natively.
I know they could easily add this. And it would immediately bump 30-40% campaign performance.
If they don’t comply, I would take the time to figure out a solution for this.
With those two optimizations this could get to 2,600 impressions per hour of work.
To me, that’s good enough. And here’s the thing.
If that toggle is checked, and every post is also being posted to my feed, that creates a flywheel where I’m getting both the benefit of community views and the compounding views of my growing following.
It would dramatically change the long term ROI of this campaign.
Alright that’s all I got for you today!