How I got 2 million views on Instagram for my app
the hook and demo format, step by step, and why it stopped working.
Four months ago I launched a speaking confidence app called Magnetic. It was doing about 10 downloads a day.
Then one video format took off on Instagram. 32million views in a month, 200 downloads on my best day, and about $4,000 in revenue so far. All of it organic.
Here is exactly how I made the videos, and the mistake I made after.
The format
It's called hook and demo.
The first 3 seconds are me looking surprised, hand over my mouth, with a strong hook in the first 3 seconds. Then it cuts to a short screen recording of the app with a voiceover.
This format still works on Instagram but is less effective on tiktok UNLESS you have a very novel app.
1. Write 21 hooks
21 is 3 videos a day for a week. Its important to post as many videos as you can to quickly find a winning hook.
Before writing anything, I spend 20 to 30 minutes watching what is already going viral in my niche. I then copy or rewrite them for app.
My two best hooks:
- "25 years of being shy, and now I find this??"
- "25 years of fearing public speaking, and now I find this??"
The hook has to point at the demo or signal something is coming next (known as a curiosity gap). A relatable line on its own can get views, but the cut to the app feels random, and your skip rate will rocket (which is very bad).
I used Claude for a first batch and then rewrote every single one until it sounded like something I'd say out loud. AI hooks are pretty bad out of the box, and its usally better to just copy what is currently working.
Side note: you can copy hooks from unrelated niche and surpirsingly they often work, it just takes a little bit of curiosity.
2. Film 21 reactions in one sitting
Record 3 seconds of you covering your mouth, looking shcoked or best of all (epescially for women) is crying. You can easily do this in 10 minutes.
3. Record one demo
One screen recording, reused in all 21 videos. Keep it under 10 seconds. Record the screen first, then do the voiceover after with a proper mic. Experiment with what type of demos work best for your app, some options are to
- Screen record your app.
- Use over the shouldershot of your app on a real phone (my demo).
- Snapchat reaction, then flip camera and show you using the app (most natural).
4. Stitch and post
In the beginning I used Claude and CapCut to stitch each reaction onto the demo, then added music and captions in the Instagram editor. The full batch of 21 takes me about an hour.
Now I have my own app where I upload the clips and it does everything, editing, captions, etc and I can download it directly from my phone. All I do is add music before posting. Claude can make you something similar or I can open source if people want.
Then post your video 3 times per day
5. Get people to comment
This is a new meta on instagram but not many people are talking about it.
Every caption says, "Magnetic” and I'll send you the link. An automation DMs them the App Store link straight away. I use LinkDM, and ManyChat does the same job. (LinkDm is way cheaper, not sponsored).
Comments push the video further, and the person gets a direct link to your app.
Before this, I had a video hit 100k views that got about 20 downloads. Views with no path to the app are worth very little. Also sometimes you can go viral with a video and have very poor conversion (an article for another time).
What happened
Three videos went viral at the same time. Downloads went from 10 a day to between 80 and 120, with 200 on the best day. I made 2k in my best month and the app was brand new! I had the worst onboarding so conversion rate was very bad, but that was an app problem not cotnent.
Then it died
After about two weeks, the format started dropping. The more I posted it, the worse it did. I'm back at around 10 downloads a day.
The format dying wasn't the mistake. They all die eventually.
My mistake was that I only started looking for the next one after the first had stopped working. And my way of looking was scrolling Instagram for hours and hoping.
What I needed was a way to see which videos in my niche were doing far better than that account's normal. Small accounts with massive outliers, not big accounts with big numbers. That is surprisingly hard to find by hand (and takes so much time).
What I'm doing now
So that's what I'm building next. It's called SearchTok.
You describe what you're looking for in plain language, and it finds the real TikTok and Instagram videos that are beating their account's normal. It also plugs into Claude and ChatGPT.
I'm building it in public on YouTube, and the waitlist is open at searchtok.app. I will post my first video next week, the series is called ‘Building app until I hit $4,669 MRR’, follow so you don miss it.
If you're running hook and demo for your own app, steal all of the above. Just start looking for your second format before the first one dies :))
