This month, I got five million social media views.
I achieved this by treating marketing as an engineering problem.
If you had asked me in June, when I started posting and got no views, whether that was realistic, I would have said no.
But every month, my numbers climb because I compulsively experiment and measure my results.
In the past, you could argue that your content was great and you just didn’t have an audience.
Today, you can’t.
I watched someone get 200 million views on their first Threads post with almost no followers.
I got five million views on X and Threads this month with only 18,000 followers.
With the new social media algorithms, you post something and either it’s good and people see it, or it’s not and they don’t. There is no excuse.
This is where engineering comes in. Every post is a test. I change one thing at a time, measure what happens, and keep what works.
For example, I used to write longer posts. But my data showed that my typical post under 80 characters got 2,264 views, while posts between 140 and 280 characters got 606. Short posts performed almost 4x better, so now I make almost everything one line.
I tested topics the same way. In September, about 1 in 8 of my posts were about AI. Those posts averaged 20,600 views, and everything else averaged 4,800.
So I made AI the core of what I write about. In October, more than a third of my posts were about AI, and my best post of the month, “Unpopular opinion: your results with AI largely depend on how smart you are,”, got 499,000 views and 1,800 reposts.
I also used to ask a lot of questions because they get replies. But none of my breakout posts over 100,000 views on X were questions. Strong statements are what break out, so now I make more of them.
On X, 18% of my posts drove 82% of my views and 71% of my new followers. One post, “The AI job tsunami is coming,”, got 648,000 views and 210 new followers on its own.
That’s more followers than I got in all of June and half of July. That’s why I don’t care about flops anymore, and I optimize everything around writing posts that have the best odds of breaking out.
If your content isn’t working, you don’t have an audience problem. You have a testing problem.
At this point, you may be wondering what five million views are actually worth.
That’s the right question. Views are the easy part to measure. Turning them into revenue is the hard part, and most people stop experimenting there.
I made a big career decision to address this, and I’ll get into it next week.