MrBeast's Video Production Playbook

HOW TO SUCCEED IN MRBEAST PRODUCTION (or How to Succeed on YouTube)

Came across MrBeast’s recommendations for his employees. The #1 YouTuber in case you didn’t know. There’s a lot about general workflows and creativity, but I was interested in his approach to making videos. I myself like to shoot stuff sometimes.

Read the whole thing and jotted down the most interesting parts. I’ll attach the original in the comments.

CTR, AVD, AVP The three metrics you should care about are:

• Click-through rate (CTR)
• Average view duration (AVD)
• Average percentage viewed (AVP) To make your videos successful, you need to get your AVD and AVP as high as possible.

Title/Thumbnail

• Must match expectations set by the title/thumbnail
• In general, the more extreme, the better

First Minute

• Maximum engagement
• The most important minute of every video
• There should be good lighting at the start

Minutes 1 to 3

• This is where you should go from talking to doing (as a rule)
• Show crazy dynamics

At Minute 3 Do a re-engagement hook

Minute 3 - Minute 6 This is the next most important part of the video — the goal is to make people fall in love with the story.

At Minute 6 Another re-engagement hook

Second Half Content

• Usually not the best content
• Never signal that the video is ending
• Place for long explanatory segments

A strong payoff at the end of the video boosts retention. The ending should always be abrupt.

Do something no other creator can do.

If something goes wrong during filming, always try to turn it into content.

Consume daily whatever will help you create better content. You can’t be inspired by what you don’t know.

BRAND DEALS (ads) - THAT’S CONTENT!

The End of Programming as We Know It

It just hit me and I couldn’t stop thinking about it.

Remember those threads about how we’ve forgotten how to make good software? The ones where people marvel at programs written 20-30 years ago and complain that development isn’t what it used to be, engineers are getting weaker, everything’s being simplified, and the levels of abstraction just keep growing.

I think this trend will only intensify. In 50 years, people will stop programming altogether. Well, not in the way we’re used to. It’ll be more of a toy or hobby for fans of retro aesthetics. It’ll be some mix of visual editors with flowcharts and a dash of text prompts to describe the overall task and possible nuances. Writing code by hand? What nonsense.

Humans will just control the result, tweak the automation, make sure everything connects properly.

And they’ll completely stop thinking about where bytes go, which framework is being used, which language, syntax, and so on. That was always just extra information needed only to be able to write what you envisioned. Objectively just applied knowledge with no other value.

And knowledge of algorithms will be devalued too. There just won’t be any use for it. The robots do the heavy lifting.

We’ve already invented so much stuff that even a dumb machine can create apps at a junior-to-mid level. What can we expect from a smart one?

All the necessary knowledge is already there. The technology to use any knowledge will come too. As well as the ability to create new knowledge.

And that’s a good thing 💪

Lopaka Gallery Launched and a UX Surprise

We recently added a gallery to Lopaka where all public projects are displayed.

There are a ton of awesome designs there, check it out: https://lopaka.app/gallery

It’s super useful to observe exactly how users are using the editor. Really nice to see that people are actually using our product to the fullest.

One interesting observation: all user graphics are stored in projects, and each project consists of screens. A project can have a name (shown on the main page) and each screen can also have a name (shown in the screen list).

And here’s the thing: nobody gives a damn about screen names! It’s just there, and nobody (95%) ever edits it.

First Customer, Giant PR, and a Beautiful Day

This morning I opened GitHub and merged a giant pull request.

I’ve been working on it for the last two months. Had a great time, got a ton of enjoyment from engineering all the cloud features for Lopaka.

Today, for the first time in my life, someone bought a subscription to my software! Unreal.

Huge thanks to everyone who supported and gave feedback.

Special thanks to Denis aka deadlink for his huge contribution to the project.

Thanks to my beloved wife, who helped research the market.

Thanks to Pasha Zhovner for the reposts and support.

What a beautiful day!

I Cloned My App in 30 Minutes Using AI

Recently stumbled upon an internet simulator: you type in any (even non-existent) URL or search query and get something generated by a neural net: https://websim.ai/

You can check out a ton of stuff on the homepage. Works really well.

The coolest part is that you can generate any app there (with enough patience). For example, in half an hour and about ten follow-up prompts, I made a copy of Lopaka: https://websim.ai/c/URr4umb4PKmu44Jr2 It’s not perfect, but that’s just THIRTY MINUTES OF WORK 🤯

Video review: https://www.youtube.com/watch?v=a4nXGnumD1U

By the way, I’m not stressing about AI replacing us. It’s fine, we’ll find something more interesting to do. Online schools, though — they’re gonna have a rough time.

Or that everything around us will be flooded with generated content. Also don’t care. My mom (65) happily watches generated YouTube videos and listens to AI retro covers of Yesenin, not even suspecting how it’s all made. She gets a lot of joy out of it, forwards them to her friends. And that’s how we’ll all be.