I built the PERFECT YouTube Intro using MrBeast's Data
A MrBeast Intro Case Study (Python/NLP/Excel/Tableau)
Introduction
"The thing that people undervalue the most are the first 10 seconds of a video".
During this full interview with Colin and Samir Jimmy (aka MrBeast) goes on to explain on that his intros are short, concise and give the audience everything they need. They have a laugh over a video that MrBeast himself watched from Colin and Samir’s channel, and got bored 5 seconds in.
They also talk about the first 40 seconds being the time where he expands from the intro trying to insert as many concepts, and challenges as possible. Later on the video developing on them to create plot twists.
And he's not alone.
Our attention span has been declining, and we are constantly being bombarded with new information; meaning that we are always ready to diverge attention. So if you own a YouTube channel, what can you do to increase the chances of people staying and watching the whole video?
My goal is to determine the key factors that make MrBeast's intros so effective, and tell you how to apply them to your own channel.
To do so, I planned on collecting the transcribed text from every MrBeast’s YouTube intro, and run NLP models on them to extract as much information from that as possible.
Table of Contents
- Collecting and Processing the Data
- Analysing the Data (Click here for the graphs and the explanations)
…….▹ MrBeast through the Years
…….▹ Use of words to promote TENSION
…….▹ Words to enhance RISK
…….▹ Grabbing your Attention through CONTEXT
…….▹ The BEST way to start an Intro
…….▹ So the ideal intro is…
…….▹ Length of an Intro
- Final Thoughts (Click here if you just want solutions for your channel)
Collecting and Processing the Data
Using Python in Jupyter Lab, scrapping via the YouTube API and a transcript API; a code was built to extract the first 15 seconds of transcription from every MrBeast video ever posted. This turned out to be inefficient because the transcriptions contained several problems, such as:
- Brackets with closed captions descriptions (ex: [Narrator], [Soft piano music]).
- Typos or complete mistakes.
- Reactions from people in the video, that don't directly make part of the words of the introduction (but make a huge difference when it comes to the impact through editing).
- Intros that were longer than 15 seconds, thus making the intro incomplete.
Therefore, I decided to go over 100 of the most recent videos and review my initial dataset manually. Taking advantage of that decision to really cut down the text to things that truly define the intro. And also taking note of how long each intro lasted in seconds, in order to completely explain the initial plot.
To finish of, I ran an NLP model using the NLTK and Spacy libraries to try to determine word order patterns. All my code for this project and a complementary project about YouTube driving factors for growth are on this GitHub repository.
Enough context, let's go the juice:
Analysing the Data
MrBeast through the Years
Graph 1 shows the crazy evolution of his channel during these past few years. Before this time he was already averaging hundreds of thousands of views, but for a few years he had almost no views on his channel. He told himself he would become successful on YouTube at any cost.
Maybe you can use this as encouragement. 🤪
Obviously the greatest driver to his success was the content of the videos, and the way he presented in general. He mastered the thumbnail and title game, but he admits that the intro is what people undervalue the most. So let's check what we can learn from him:
Use of words to promote TENSION
He developed his style throughout the years to incorporate more tension. By tension I mean that he is always trying to increase the stakes. Graph 2 sheds light on his use of conjunctions in sentences. Conjunctions are some of the most used words to "connect" basic sentences, and better explain their context.
As you can see below, he uses the word "and" at a much higher frequency than other conjunctions, and the average use of "and" keeps growing through the years.
On a normal scenario, we often use words like "then" or "also" to keep adding information to a story, otherwise it just sounds too intense. But he specifically calculates to use "and" so that it keeps increasing the tension and the stakes, even in situations where grammatically wouldn't make sense. Here's an example of an intro were he used "and" 4 times:
"I got 100 of my subscribers and gave them each $10,000. And I also rented one of the largest malls in the world, and locked them inside. And for the next 10 hours, I am gonna hunt them down."
As you can see, this wouldn't necessarily get an A+ as in intro on an essay at your english literature class. Yet, it performs very well on bringing more tension to the videos. From the intros studied, 61% use the word "and". So don't forget to use it on yours!
Words to enhance RISK
On the other hand, words like "if" and "but" help to insert the "risk" into the story that he is presenting to the audience. Graph 3 brings interesting insights on that matter:
As we can se above, 18% of intros use Time as a risk to move the plot forward. Coincidentally 18% of intros use Money to do so. Risk is key on opposing to the goal of our "hero" (here talking about the hero's journey).
My view to simplify it: the hero wants something and faces a risk to get it. They have to decide if the risk is worth, because to get what they want, they pay a price and learn from it.
So in every intro, try to keep these concepts in mind:
- Explain the context to create empathy and interest
- Have an amazing goal for your hero to reach
- Insert a risk that makes people question if it's worth it
- Give the reward to the hero, and show how they have changed
And now you might be saying out loud: "Listen here… I'm not making videos about amazing journey's and spending millions of dollars on people to do weird fun stuff. I own a channel about fashion. You crazy man"
You can still apply these same ideas:
- Explain the context of where you could use a piece of clothing
- Invent the challenge of managing to use that piece of clothing somewhere that no one would think of
- Show how risky it is to try to combine these concepts, and how people react to it
- Get people's positive reactions to show confident you looked like on that situation (reward)
Once again, this is just an example. But the idea is: put yourself (or something) in a journey, and make of yourself (or that something): the hero.
Grabbing your Attention through CONTEXT
The first point of this list we just peeked at was: context. There are several ways to give context with images, sounds and editing. But the most effective way to insert context through words is by using:
"THIS IS…"
Let's take a look at Graph 4 to see how often he uses this sequence of words, and what is their impact:
As you can see, 58% of videos had the word "this" at least once in the intro, while 12% had the word "these". They both work just like "a huge pointing finger" would if you were trying to show and explain something to a friend on a street. When we use "this", we instantly grab the attention of the viewer to understand what we are talking about and what else is there to know about it. Being the reason why it's so useful in an introduction, to make sure the viewer remains interested. But should we use it in every intro? And how?
The BEST way to start an Intro
The Spacy library has the interesting feature of creating dependency trees through tokenisation of words. A dependency tree gives a classification for every word and how it relates to others in that sentence. Image 2 is an example of one made for an entire intro in this data set.
From our dataset we can pull the patterns of words used to start the intro, as seen on Graph 5:
We can notice that 34% of the sentences begin with the pronoun "I", and 16% with "we". Remember I mentioned the use of "This is" to give context? A great strategy is to use “This is” as a intro starter when the goal of the hero needs more context before making sense.
We can now take a look at how well this intros perform when it comes to average Views, Likes and Comments down here on Graph 6:
Once again we see the intros beginning with the pronoun "I" performing better than the remaining options. "This is" comes up right next to it, followed by the pronoun "we". All of these act as perfect "context givers".
An interesting situation comes up as an outlier that averaged many likes and comments on the "We+word+verb".
This "word" in between the pronoun and verb is very often a word that enhances the power of the verb.
An example of this is: "I just bought this house and…"
So the ideal intro is…
I + verb… / I + intensifying word + verb…
or
This is…
On Image 3, we can see a Dependency tree for the first sentence of what I believe to be his best video intro:
Length of an Intro
Hopefully by now you understand what Jimmy meant when saying that people undervalue the introduction. Explaining how important the first 10 seconds are. However, could we say that there is a perfect intro length to truly explain a plot and get better views?
Graph 7 shows the relationship between intro length and the average main stats: Views, Likes and Comments.
Looking at this we would see that 8 second long videos yield some of the best possible results. And with the exception of an outlier at 16 seconds, we could say that the ideal range is between 6 and 9 seconds.
A strong argument for short intros is that "if you can't explain simply, than the video is not simple enough to be understood".
Final Thoughts
Suggestions from the data pulled in this study:
- Giver your hero some well defined: goal, risk and reward. And use context to make it very clear to the audience what they mean.
- Introduce all of the above as concise and direct as possible, ideally from 6 to 9 seconds, but preferably in less than 12 seconds.
- Use risk words like "if" and "but" and give the audience a reason to empathise with the risk, wondering if it is worth to take it.
- Use the "This is" sequence of words to give context as soon as possible, just like a pointing finger would.
- Start your intro with the words "I+verb" or "This is" to give context.
- Increase the tension on your intro as soon as possible. Use the word “and” to do so, and avoid words like “then” or “also”.
My personal suggestions, taken from years of experience of improving several stats to clients in the Youtube platform:
- Follow the hero's journey structure for the video but also to guide your Intro.
- Use text, sound and animations to enhance your words. This is the most perfect intro he has ever made in my opinion.
- Give the audience the satisfaction of the reward to the hero with amazing visuals and emotions.
- When editing, add people's reactions to the introduction, to increase tension even more. Here's a video example of that.
- In case you thought that making such short intros won't work for you, try using this intro of his as reference.
- If you want to master the use of data to increase your channels growth, check my complementary study on that.
- If you truly want to master the craft of well made intros and videos, study the following channels: Vox, Johnny Harris, Mark Rober, Veritasium and Cleo Abram.