A few months ago, I was in SF talking to a bunch of AI creators who focus primarily on Instagram.

Some of them had audiences roughly the same size as mine, yet they were earning more than 10x as much as creators I know on X with similar audiences.

Since I built my content business on X, that blew my mind.

It probably sounds like a stupid outside motivator, but it was an effective one lol.

And the more I thought about it, the more it made sense.

So I decided to go all in on short-form for 60 days.

I even took a short hiatus from this newsletter because I wanted to sprint fully on learning it and come back with real results to show you.

So here are my results after 60 days:

My goal is to hit 70,000 followers this year, so I’m about one-sixth of the way there.

After 53 videos, here are the biggest things I learned.

A strong topic carries almost everything else.

A great hook, trend, or edit rarely saves an idea that people do not care about.

Before I make a video, I score the idea across five things:

I use Claude or Codex to scrape a larger batch of ideas and help me rank them across those criteria.

I still make the final choice because current models have a weak feel for taste and what people in a niche are starting to care about.

Coming from X as a text-first creator, I always thought of the hook as just the first line of the script.

But in a video, people take in THREE things at the same time:

The most surprising lesson for me was that the spoken hook is usually the least important of the three.

The order I think about it now is on-screen text, then first visual, and then the spoken line.

The viewer often understands what kind of video they landed on before I finish my first sentence.

People did not opt in to see your video, so you have to give them a reason to invest their time immediately.

A strong opening usually does at least one of these:

I also frontload the main value proposition or the most shocking fact into the first two spoken lines.

In other words, just state “the main thing” in the tightest, most dramatic, scroll-stopping way immediately.

Early on, I made the mistake of adding too much context before I reached the main interesting part.

The point can be incredible, but if 99% of viewers have already swiped away, they will never see it.

Good audio makes a bigger difference than I expected, especially if your room has a lot of reverb.

I use Adobe Podcast (not sponsored btw) to clean up the recording, and it is free and stupidly effective.

For the camera, a newer iPhone is already enough.

You can also use a small handheld camera like an Osmo Pocket, which is what I use.

Views are the easiest number to obsess over, but the metric that matters is conversion.

The main conversion I track is how many viewers become followers and eventually customers.

I connected Codex to Instagram through browser use (Codex browser use is so OP now btw) and had it analyze my latest batches of 10 videos.

To set it up, install Browser from the Codex Plugins Directory, then log into your IG account on your browser and tell Codex to navigate there.

Claude Code can use the Claude in Chrome extension for a similar workflow.

I mainly look at:

One video with around 400,000 views generated far more followers than another video with 1.2 million views.

That is the kind of format I want to scale because it creates reach and turns that reach into a relevant audience.

This was a hard pill to swallow coming from text-based X, where the words can carry the entire post.

You can have the best idea in the world, but if you cannot also find great visual footage, the chance of the video going anywhere is extremely slim.

Now I check the available demos, images, and clips before I commit to the script.

If the visual material is not there, I usually skip the idea.

I publish 10 videos, analyze the batch, form a few hypotheses, and use the next 10 to test them.

Even a video doing twice the usual views is worth studying because it can reveal something useful about the topic, opening visual, or structure.

I treat the process like an iterative video game.

You need at least 50 videos before making a big judgment about whether short-form works for you.

Most people stop while they are still collecting the first useful data.

My first drafts are almost always 20% to 30% too long.

If I already explained something, implied it in the previous line, or added it only to smooth a transition, I cut it.

I use AI to help identify and critique those lines, but I make the final call based on whether each sentence moves the story forward.

Every line has to earn its place.

Many people will say you should “just yap” on camera, but that only works when you are extremely good at leaving out everything that does not help the point.

After 60 days, I have found AI is most useful once I already have a direction and real material to give it.

AI is good at:

AI is still bad at:

The media-sourcing workflow alone has cut a ridiculous amount of time from editing.

Once the script is finished, I ask Codex or Claude Code in the same chat:

That is pretty much the entire prompt.

When the editor opens the folder, all the raw material is already there in the same order as the script.

A picture says a thousand words, and a video is a thousand pictures.

When people see your face and hear you talk often, trust compounds much faster.

I have met Instagram creators with around 10,000 followers who make more from their businesses than creators with 200,000 followers on X in the same niche.

I seriously underestimated how large that difference could be.

The way I see it:

You can build a complete sales funnel directly inside Instagram Stories, which is much more powerful than anything I’ve been able to do on X.

IMO it’s an excellent, more intimate and visual complement to a newsletter.

Feed videos bring people in, stories turn that attention into customers.

After 2 months in Vienna, we are back in Berlin.

The main feeling I have is that I want to call one place home again.

I want to grow roots, invest in long-term friendships and community, and build a place that actually feels like home.

We already know Vienna will not be that place for us.

We will spend two months in Berlin, try one final city, and then decide where to settle.

See you next week,

Ole

P.S. I’m deciding what to teach in my next live workshop, and I’d love your vote: