- with only one video
- on a brand new YouTube channel
- 100% organically (0 external promotion)
Making this channel the first in history to achieve this.
Here's how.
First of all, this video was NOT "made by AI" but "made with AI".
I didn't find a magic prompt and *poof*, the whole video was made.
I used AI as a tool, just like you'd use a shovel to dig.
Many people romanticize AI but it's far from being as easy as they think.
I'm very experienced in YouTube strategy & content creation, and this is by far what weighted the most in the outcome.
AI just helped me go faster.
1/ The topic
I'm well-versed in finance, trading and economics. I know the subject well & can easily spot AI hallucinations (wrong or made-up facts).
I also consume that kind of content, so I have a good idea of what this market is looking for story/information-wise.
2/ Ideation
I used ChatGPT 4 to assist me in the ideation process, and it is a game changer for me.
I'm very good at spotting strong ideas, but finding them from scratch is another story.
Not anymore with ChatGPT.
I can now ask precisely what kind of ideas I'm looking for with a specific frame and generate a list in seconds.
All I need to do is select and execute the best ones, that's literally it.
So I've played with it, generated about 50 ideas and ranked them all on a spreadsheet.
3/ The hook
I've made about 20 prompts for the hook, mixed them and added my own touch.
Here's the final version with editing:
4/ Script
By far the most time-consuming task.
It took me about 2 full days to do it, but it's because I made a lot of mistakes in the process.
The first iteration is always the steepest and I've learned a lot from this one.
What I did was first build the skeleton of the story then build it paragraph by paragraph once the direction was clear.
It took a lot of wrestling with ChatGPT to find the right style, but once I had it, it went faster.
About 70% of the script is prompt engineering & 30% manual since ChatGPT is really good at following a certain style.
It saves a lot of time, but it still has a "GPT taint" to it, which quickly creates a recognizable & redundant pattern.
I tried to break it as much as I could.
5/ Voiceover
The voice is my friend's voice but we used @elevenlabsio "voice cloner" feature for this project.
It took quite some time to build a voice in the tone & style we wanted, and tbh the result is impressive (considering how early this feature is).
Unfortunately, the voice tone and intonation aren't super customizable yet and still have an "AI" vibe sometimes.
There's nothing we can do about it, but I'm confident these tools will only get better.
Surprisingly, the majority of people didn't even notice it was an AI.
6/ Editing
For the edit, I outsourced the job.
This was by far what took the most time because I went through different editors and ditched a lot of content.
Once I found the right fit, I still had to edit about 20% on top of it to be the way I wanted.
I've spent about $2k in total in editing (proofs of concept are priceless) but if we only take into consideration the final product, it was about $450.
6/ The final product
Even though the final product wasn't perfect (and I knew it), my goal was to build a proof of concept.
I wanted to confirm the demand from my market research.
I always build my channels with the market, improvement will be made iteration after iteration.
7/ The packaging
Regarding the title, just like the hook, I prompt-engineered about 20 times and added my own touch.
For the thumbnail, I used Midjourney to generate an avatar of "Mrs Watanabe", and my personal design skills to create the thumbnail.
The whole journey was documented on the Investor's Kitchen (the discord community I run) from day 0 before the video was uploaded.
Skin in the game.
If you're tired of reading generic breakdowns of "why this MrBeast video went viral" and want to understand YouTube at its core, The Investor's Kitchen will save you many scars along the way.
YouTubers, you're probably not reading data in your analytics correctly.
Here are 7 biases you're blind to:
I have been contacted by people working at YouTube to provide my feedback on the analytics, but before I do detailed feedback (soon), let's start with a quick overview of how reading your analytics can fool you.
Throughout most of history, humans have evolved to make quick decisions with limited information.
While beneficial in survival scenarios, it becomes problematic in our modern context, especially when interpreting data where a more careful & analytical approach is required.
If you still struggle to dissociate legit people from charlatans, here's a mental model that will not only help you answer this question in less than 30s, but will also save you a lot of money in your life.
(Thread)
"In theory, there's no difference between theory and practice.
In practice, there is."
Yogi Berra
In this thread, we will discuss the concept of "skin in the game" popularized by Nassim Taleb.
Simply put, skin in the game is when someone faces the consequences (good or bad) of his actions.
If you bet $1000 from your pocket, you're skin in the game.
Why YouTube gurus are dangerous for content creators.
(thread)
Our brains are not wired to fathom the complexity of this world because it is mostly random.
To make sense of such a chaotic world, we have mental shortcuts and biases.
This helps us make decisions faster without constantly thinking about what to do next.
While this is generally beneficial (for survival), it's destructive when looking for signal.
I've selected a non-exhaustive list of biases YT gurus are blind to, resulting in misleading content creators and making them make mistakes that can become lethal for their channels.