wono

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.

It will be divided into two parts:

1/ Biases when analyzing successful creators/videos

2/ Biases when analyzing analytics

(In the link in my bio, I have a whole section called "Vaccine against YouTube gurus", if you want to read more about it)

----------------------
1. Biases YouTube gurus are blind to when analyzing successful creators/videos
----------------------

🔸 Survivorship bias:

The most classic one.

Focusing on successful creators (MrBeast, Airrack, Ryan Trahan...), while ignoring those who failed and concluding that their success was due to specific factors.

Stuff like:

"Mrbeast genius strategy"
"Mrbeast retention is why he's successful"

You get the idea.

99.9% of these are pure noise, they highlight one person who succeeded at something while ignoring thousands who failed.

Here's a great (short) video on the subject:

🔸 Hindsight bias:

Believing that the success of a creator was predictable after it happened when in reality, it was impossible at the time.

YouTube fake gurus are great at predicting the past but horrible at predicting the future.

You've seen many:

"I've studied every MrBeast video/interviews.."

But have you seen anyone predict the next MrBeast with the same level of precision and be right?

That kind of content is basically explaining the lottery numbers after their release.

Bullshit.

🔸 Attribution bias:

Overemphasizing internal factors, such as:

- thumbnail
- retention

 And underemphasizing external factors, such as:

- "remarkability"
- supply/demand dynamics (attention markets)

Many overanalyze to sound smart/farm credibility.

"MrBeast uses a surprised face in the thumbnail"
"Ryan Trahan color theory genius"

A video performs well because it's unique and fresh (remarkable).

Everything else is mostly nothing but overfitting concepts to flex.

🔸 Selection bias:

Analyzing only successful videos and drawing conclusions that don't generalize to other videos.

You need to analyze both sides of the coin to get the big picture (studying successes AND failures).

Most of these fake gurus out here never built a youtube channel from scratch, making them even more prone to this bias.

----------------------
2. Biases YouTube gurus are blind to when analyzing analytics.
----------------------

🔸 Confirmation bias:

Looking for data that confirms preconceived notions about what makes a video successful rather than considering alternative explanations.

A classic one. when they try to make sense out of CTR/AVD when in reality it's mostly random.

🔸 Availability bias:

Relying too heavily on the most easily accessible data without considering the broader context of the video's performance.

That's why they try to fit everything under CTR/AVD/Retention, when most of the time it's due to market conditions (supply/demand).

For instance, these two videos have roughly the same number of views, but they compete in two different markets (the first one is a fun gameplay, the second one a deep analysis).

Obviously, the entertainment market is larger than the "serious essay" market.

🔸 Narrative bias:

Tendency to interpret information as being part of a larger story or pattern, regardless of whether the facts actually support the full narrative.

"CTR/AVD are bad"
"AVD is low"

This is complete nonsense for understanding a video's performance.

More here:

🔸 Attribution bias (again but for analytics this time):

Overemphasizing internal factors, such as:

- Retention graph
- CTR/AVD

And underemphasizing external factors, such as:

- "Remarkability"
- Supply/demand dynamics (attention market)

🔸 Sampling bias:

Drawing conclusions based on a non-random sample of data and overlooking other factors, such as only analyzing videos on channels that already have a large number of subs or videos of roughly the same duration.

For example, a lower CTR can only be due to the length of the video &amp; not to the thumbnail's performance because the shorter a video, the more likely a viewer will have the time to watch it right away (= 📈CTR).

Here's an example with videos from 2 different channels of mine:

🔸 Simpson's paradox (by far the most overlooked):

A statistical phenomenon where a trend or pattern observed in a group of data is reversed or disappears when the data is divided into subgroups.

This one is a bit complex but super important.

Most creators are fooled by Simpson's paradox in metrics representing an average (especially CTR, AVD &amp; the retention graph).

I'll make a separate thread with more details to explain this soon.

Fooled by randomness:

The tendency of people to mistakenly attribute significance or meaning to random events or to assume that there is a pattern or trend where none actually exists.

This is the surprise du chef, I see it everywhere.

For example:

"I noticed an increase in CTR after I changed the thumbnail"

When in reality it's only due to CTR fluctuations that would've happened anyway.

That's it for this one!

I could keep going like this all day because there are so many biases to talk about, but the point of this thread was to give you some tools to protect yourself from so-called experts who are completely clueless.

They are misleading many creators (small &amp; big), and the only way to prevent this is to know these biases.

Be careful of who you take advice from because remember that in the end, this is your career that is at stake.

My private (but free) newsletter will launch in a few weeks, link &amp; conditions to be whitelisted below:

👉

You know I'm a somewhat of a psychology master 😏

If you're a creator on YouTube not getting the views you want, it's not because:

- Your retention is bad
- Your thumbnail is bad

But because you do not understand "attention markets".

It took me 12 years to figure it out, but I'll explain it to you in 5 minutes 👇

1. Attention Basics

Attention is a limited resource. I like to define it like this:

Attention = interest + time

Take you as an example, you clicked "show more" on this thread, meaning I already have your interest, but now I'm also getting your time.

As you are reading this, you can't simultaneously give your attention anywhere else (interest + time), it is impossible.

If that were to happen, I'd first have to lose your interest, which would result in losing your time as well.

In other words, I'd lose your attention.

All good? Great.

Attention is a currency.

If you spend $10 at your local restaurant, that's $10 all nearby restaurants didn't win.

Replace $10 with 10 minutes and you have the attention market.

We call that:

2. A zero-sum market

Think of it like a pie, if one slice is bigger, the other slices are smaller.

When one video gets one view (interest), all other videos lose that view.

And when the same video gets watch time (interest + time), all other videos lose that watch time.

Now I want you to understand this:

A zero-sum market is not hard to beat but EXTREMELY HARD to beat consistently.

Why?

Because any gains you make (views) come at the expense of someone else, leading to a highly competitive environment where people fight to survive.

Here are a few examples of zero-sum markets/games:

- Commodities market
- Sports betting
- Gambling
- Trading
- Poker

If you have "played" in one of the above for long enough, you know what I'm talking about when I say:

👉 "it's extremely hard to beat consistently".

Have you already won a few sports bets/gamblings/poker games?
Sure, probably.

Were you in profit after 100 bets/gambles/poker games?
Extremely unlikely.

Why? Because they are zero-sum games and also something we call:

3. The power law

In simple terms, a minority will have a lot, while the majority will have very little.

This is observable in many competitive fields/markets, such as wealth distribution for example.

This is even more exacerbated in a zero-sum market where people compete for limited resources (money, attention, power...).

On YouTube, a minority of videos get A LOT of attention while a majority gets very little.

This is why I called it "the attention market".

When you understand:

1. Attention is a limited resource
2. It's a zero-sum market
3. The power law

You will also understand that to beat a zero-sum market where many people already have massive leverage, you need something very few people can access (such as your personality).

The narrative:

Getting more people to click &amp; making them watch longer = more views is a fairy tale.

Why?

Because that's not how you beat a zero-sum market.

Can most people get better at making thumbnails and titles?
👉 Yes

Can most people get better at storytelling?
👉 Yes

If everyone has access to a resource in a zero-sum market, it can't be the answer to beat it.

Plain &amp; simple.

It must be something the market doesn't have access to or is hard to achieve for most.

If better packaging can only lead to success, then why do most startup products fail?

Aren't they made by marketing experts?

If better storytelling can only lead to success, then why do most movies fail?

Aren't they made by storytelling experts?

Some are built for years and end up being complete failures.

Do you think a nice thumbnail/title &amp; good storytelling on which you worked a few weeks at best, will help you get more views when it couldn't help professional marketers &amp; storytellers after years of polishing?

It won't.

They are definitely great tools for building an audience, that's undeniable.

But get more views? Fairy tale.

To beat a zero-sum market, the solution has to be counter-intuitive AND hard to pull consistently.

And that solution is:

4. Be remarkable, not average

Now we have something:

- Counter-Intuitive
👉 You must not think like 99%

- Hard to pull
👉 Ideas can't be rushed and are hard to execute

- Constantly changing
👉 It can only be remarkable once, after that it becomes average

"But wono, @MrBeast said this".

Yes he did, and to me he's wrong.

He has a great intuition for remarkability and is right about many things.

But on this specifically, he is wrong (mostly because of the survivorship bias).

@MrBeast did something remarkable to beat the market at first to get attention (counting to 100 000).

This was hard and counter-intuitive, 99% of the market didn't think of it.

He then did it again by donating $10k to a homeless person, again risky &amp; counter-intuitive.

After finding a gap in the market he then built leverage to beat it on a consistent level.

Same for @Dream:

Before his channel, the Minecraft market was mostly PVP soup/time-lapses.

He came up with something remarkable and snowballed from there.

From my long experience as a creator on youtube (12 years, 1500+ videos) in different attention markets, I have observed that:

Youtube cares way more about "attention clusters" (remarkable)
than pure metrics (CTR/AVD/Retention graph).

I'm also lucky to have experience with other zero-sum markets and various other projects involving market dynamics &amp; economic concepts.

This allowed me to experiment with them and see how they played out on YouTube.

The narrative:

"get more clicks and make people watch more" = more views will only work:

- If the idea is remarkable enough to get more views

- Or if there is already an established audience.

Outside of that, won't work.

Thumbnail/Title 👉 tools to make something remarkable, more remarkable.

Storytelling/retention 👉 tool to build an audience

None of these tools will make something not remarkable, remarkable.

Just like a tool can fix a car, but can't be the car.

Remember that YouTube wants you to believe this narrative because it's in their own interest.

People watching longer will be more beneficial at scale for them than it will be to you.

YouTube would be way less sexy if they promoted it as a zero-sum market very hard to beat.

TL;DR:

1. Attention is a limited resource
2. It's a zero-sum market
3. The power law applies
4. To beat the attention market you must stand out by being remarkable, not average
5. Remarkable today, average tomorrow

That's it for this thread!

We just scratched the surface of attention markets with the basics, but hopefully, this gives you a better idea of it as I mention that notion often! 🫡

Yes, but it is important to mention that the idea is easily 90% of the work. If not 99% for some cases.

Chasing virality on YouTube will eventually kill your channel.

I made this mistake and learned it the hard way.

I've been a full-time creator for 8 years (and counting) thanks to audience building.

I share in this thread how I did it so you can avoid depression.

👇

Forget views, forget subs, forget engagement.

Building an audience is all about catching attention and managing it.

It's about building a system that allows you to manage the attention of your viewers so you can leverage it to reach your goals.

Views, subs etc.. are proof of attention, not attention itself.

They are tools to help you measure the attention your content/channel earned, kept, or lost.

But first, let's start with the basics:

1. Attention Quality

This is the most important thing you must remember from this thread:

Not all viewers are equal.

Most creators tend to think that:

1 view = 1 view

This is the WORST belief you can have as a content creator.

Here's why:

When you build an audience, you have 4 kinds of viewers:

• Blue: they don't know your content exists
• Yellow: they've seen your content at least once
• Orange: watch your content regularly
• Red: die-hard fans, watch everything

(Or from cold to hot)

The attention (in itself) is the opposite:

It starts 🔴 (hot) but ends up 🔵 (cold) over time, as weeks pass unless your strategy to keep it red is strong.

The quality of attention varies from one viewer to another.

Red viewers watch more content and watch longer.

In the long run, 10 🔴 viewers will produce more watch time than 100 🔵 viewers.

Pareto's Law:

👉 80% of the results come from 20% of the viewers.

The goal when building an audience is to convert as many 🔵 → 🔴 viewers as possible.

BUT (and this is where most creators fail), it must be done so that 🔴 viewers are not lost in the process.

Optimizing too much for views will affect the audience (🔴 viewers).

2. Attention Management

When a blue viewer comes across your content and ends up hitting the subscribe button, he sends you a message:

👉 I enjoyed this video and would like to see more like this one.

At this point, he's yellow, not orange yet, with "red attention".

To keep the attention red, you need a strategy to manage it.

The nightmare for a creator is to have a majority of yellow viewers, who turn into ghost subscribers over time (yellow viewers with blue attention).

Like here:

3. Format

A format is a recipe (or a formula).

It is a structure of content with its codes, crafted so the viewer can know what to expect in the future.

This is by far the best way to sustain attention in the long run.

For example, MrBeast's format would be:

- Entertainment involving extreme challenges and a lot of money
- 10-15mn long on average

Dream:

- Finish Minecraft as fast as possible while hunters try to kill him
- 40-50mn on average

Casey Neistat:

- Vlogs with storytelling
- 5-10mn on average

elwono (one of my channels):

- Fun Dota 2 strategy with storytelling
- 20-30mn on average

The majority of successful channels have a specific format or farmed attention with one format.

It is a proven method that works at scale.

A format will grow your base audience organically while maintaining a viewer orange/red for as long as possible (years, not weeks).

When you build an audience:

👉 You are not looking to go viral at all costs, but to keep the ratio blue &gt; red the highest possible.

Viral videos bring many blue viewers, and if your library can't convert them, they will end up being ghosts forever like here:

Sure, it feels great for the ego to go viral and must be a part of your strategy, but not the priority.

You can trust me on this:

I have hundreds of viral videos, and what allowed me to survive 12 years on YouTube was not these videos.

You judge the health of an audience by its ability to convert blue viewers into orange/red.

I started "elwono" in mid-2019 and I applied most of my current knowledge about audience building.

It was built to maintain as many orange/red viewers as possible.

Results 4 years later:

- 66 videos in total
- Average views: 532k
- Only 7/66 videos below 200k views
- None of the last 34 videos had less than 200k views (except the part.2 of a 568k video at 199k)
- 33/66 videos at 500k+
- 21/66 at 600k+
- 8/66 at 900k+

The goal of a format is to consolidate and grow the bottom of your views, not the top.

That's where you should focus 90% of your strategies because it means you're growing your base of red viewers.

4. Be a long-term player

Youtube gurus are obsessed with views because they didn't experience the long-term game, failing to tell you that:

Attention is highly volatile &amp; PERISHABLE.

You must NEVER forget that. Let me say it again: NEVER!!!

Farming red viewers take time.

Yes, going viral is cool, but if you have no strategy to keep them in the long run, the attention around your content will fade and produce ghost subs.

Give yourself at least a year to establish a solid foundation of red viewers.

Convert blue viewers, let them get used to your content, and be patient.

Farm them one red viewer at a time don't be in a rush especially if you're a small creator.

A library of the same format with a high reach &amp; conversion rate will give you results you can't even fathom.

To sum it up:

1. Viewers are not equal, 🔴 viewers are the real deal

2. Build a strategy to farm 🔴 viewers

3. Craft a format to increase the conversion rate from 🔵 to 🔴 viewers

4. Play the long-term game with a compounding audience

These are just the basics, but hopefully they give you an idea of where to go now.

I will post more content like this in the future so subscribe to @wono_strategy if you want to see more.

Here is another thread you should be checking:

Yes, viral videos are great when your strategy to convert blue &gt; red viewers is on point. 

They are a great boost.

But in the opposite case, they are nothing but a mirage.

Understand that you will never achieve a 100% remarkable rate; it is impossible, and no one has ever done so.

The best is to focus on the audience first and foremost and see "remarkability" as a bonus.

You should definitely aim for it, but not at all costs.

Be authentic and genuine, not trying to be someone else.

Fresh new angle (disruptive, unique, bold claim etc)

If you're a content creator on YouTube, you probably heard this:

Higher retention = More views

Well you've been lied to, this is 100% wrong ❌

Here's why and what will truly get you views:

(thread)

I've done many things in my life.

Besides being a content creator, I also have a good experience in market trading (a lot of football trading &amp; a bit of crypto).

This is why my vision of YouTube is "market-oriented" and why I came up with the concept of "Attention Markets".

1. "Niches" vs. "Attention Markets"

I don't use "niche" but "attention market" because the former is static and lacks the supply/demand dynamic.

• "Basketball" is a niche

• "People who understand basketball" is an attention market

It's a subtle yet critical difference.

If you make basketball content with "niche" in mind, you will make content for basketball fans.

But if you keep "attention market" in mind, you'll never forget that there's a whole world of people who may not watch basketball, but they still get it.

2. Zero-Sum Market

This is something so fucking essential to understand, and YouTube gurus are clueless about it.

That's why they are parroting each other non-stop, spreading pieces of advice that are both wrong and dangerous.

Such as:

- Higher retention = more views/impressions

- If people watch until the end = more views

All of the nonsense that people say when they don't understand how a zero-sum market works.

Here is how it works:

When someone opens up YouTube, they're here to spend some attention (watch time).

Let's say they've got 60 minutes to spend (the demand).

The algorithm shows them a selection of videos (the supply) that they're most likely to click on.

In this example:

8 videos are selected by the algorithm.

They compete against each other for a portion of the 60 minutes of watch time available.

When a video gets to compete, we call that "an impression."

Every single one of these 8 videos receives one impression.

To keep things simple without getting into complex dynamics, let's assume each video is 1h long.

In this configuration, we have:

- 1h of attention available (demand)

- 8h (8 x 1h) of content available (supply)

It doesn't matter how the viewer spends his attention, there will be losers.

Either:

- One video wins the whole 1h watch time &amp; 7 loses

- 8 videos get 7.5mn each (or any other combination)

- The viewer decides to leave youtube &amp; spend his attention elsewhere, nobody wins

That's a zero-sum market.

Attention is a limited resource because it is tied to time, and humans' time on earth is limited.

Now some of you will say:

"But wono, he still can watch them all later, everyone can win!"

Yes, but nope.

Here's why:

Once spent, these 60 minutes are gone forever.

The 5 minutes you spend reading this thread come at the expense of everything else in terms of attention.

That's 5 minutes you'll never get back, and once you're done, you'll be 5 minutes older.

Your time is limited.

Yes, you can spend 5mn more elsewhere, but that's 5 MORE minutes of your time.

If you spend $5 somewhere, sure you can spend another $5 somewhere else, but that totals $10, which is $5 MORE of your wealth.

On YT the attention demand is limited, it is not infinite.

That means if one video competes for 1h and loses, it will never be able to win them back.

That hour is gone and the next hour is another competition, they are two different battles.

Back to our example:

the "attention pie" at this very moment (where each video gets 1 impression), is limited to 1h.

If someone gets a bigger slice (or the whole pie), it's taking away from everyone else.

It's a zero-sum market - if someone wins, someone else gotta lose.

Now, let's see why "retention = views" is wrong, or rather, why it can't be true.

3. Attention Market Size

Depending on what content you produce and how you produce it, every video will compete in its own attention market.

If you trap a video inside a super tiny market, no matter the retention, you'll never outreach it. (like my basketball example above)

For example, if you make a guide on how to play a game only 100 person play, good luck going viral.

There are many ways to trap your content in a small attention market that are far more subtle than this obvious example.

99% of the time, this is the reason behind low views.

4. Markets are incredibly efficient

Anyone who's spent a good amount of time in a zero-sum market knows this, especially swing traders.

Markets are extremely efficient and incredibly hard to beat consistently.

The power law applies in these markets, I already covered the power law here if you haven't read it check it out:

90% of all uploaded YouTube videos will never reach 1,000 views.

Only 0.77% of videos get 100k+ views.

And I can tell you I know something about it, I uploaded 1500+ videos on youtube for a total of 260M views (173k/video average).

I crossed the data left &amp; right and retention alone had nothing to do with "more views."

In a zero-sum market where the power law applies, the competition is so high that by pure common sense, the answer to getting more views can't be retention.

Because:

Anyone can learn story-telling, retention techniques, etc.

There are infinite free and premium resources out there about that, anyone can master it.

This is why many filmmakers (who, by the way, are master storytellers) spend 4 years working on a film you'll never hear of.

Because the "power law" applies, a minority of videos will get 88.4% of the total views.

To beat such a market and join this minority, you must have an edge (competitive advantage) that only you (ideally) can access.

If everyone has access to something, it becomes average, &amp; average is boring.

The opposite of average is remarkable, as Seth Godin puts it:

"something worth making a remark about".

If your video were a car, "remarkable" would be the engine &amp; "retention" would be the comfort.

If your video has something remarkable (doesn't have to be the whole video), it will get views if it's clear enough in the title &amp; thumbnail.

Here are a few examples from a channel I ran in 2017–2019 in a highly competitive market (pro gaming gameplays):

I made the thumbnail/title about an impressive decision-making moment by a pro player in Dota 2 "World Cup" final.

Many didn't see it live because it happened too fast.

18mn total video, ~40 sec remarkable moment

Result: 508k views (with 103k subs at that time)

Same here:

World cup match, again I focused the title/thumbnail on a specific impressive play.

11mn gameplay, ~30 sec remarkable moment

Result: 300k views (100k subs)

Same here:

19mn video ~12s remarkable moment

362k views (103k subs)

Here (another channel):

One of the worst videos I've ever made in terms of quality (rushed editing, shit retention).

It was a glitch hard to execute (needed 5 coordinated people to do it), so people wanted to find out about it.

21mn → 3 remarkable mn

1.7M views → 160k subs

I could continue like this all day long, as I have over 1500 videos and like 100 different angles.

But my goal is to give you a big picture so let's continue with the last pieces of the puzzle:

Noise vs Signal 101 on YouTube.

"CTR &amp; AVD are not perfect, but ignoring them is stupid."

"The truth is somewhere in between"

Right?

🤦

(I swear the fact that I have to make this thread is extremely concerning)

Mega thread 👇

Okay, bear with me as I try to calm down while I write this, I apologize in advance if I talk like an arrogant piece of shit.

Incompetence and ignorance becoming mainstream drive me crazy.

*Deep breath*
😑😤😮‍💨
😑😤😮‍💨

Hope it will hold.

Sorry if I fail.😂

By the end of this thread, you will clearly understand why these assertions are nothing but incompetence (or ignorance) in disguise:

• CTR &amp; AVD are not perfect but ignoring them is stupid

• WhY iS yOuTuBe PuShInG tHeM iF tHeY aReN'T vIeWs iNdIcAtOrS? (sorry couldn't help)

Let's start with the VERY BASICS:

👉 What cannot be measured, cannot be improved.

You improve by:

1. Removing noise
2. Identifying signal
3. Apply what the signal tells you
4. Improve your understanding of it through trial &amp; error

👉 Accuracy VS Precision

You expect a metric to be both: accurate AND precise.

Why? Because a metric is used to MEASURE something QUANTITATIVELY.

If it's not precise AND accurate, it's not a metric.

CTR &amp; AVD are METRICS.

Do you know what this means?

They MEASURE something quantitatively in a precise AND accurate way.

But what exactly? We'll see in a moment, bear with me.

Last element 👉 Context.

A metric only makes sense in context, otherwise it's just useless information.

For instance, if I say:

28°

It doesn't make any sense without context.

Let's recap:

👉 You can only improve what you can measure

👉 You measure something with an accurate (and precise if possible) metric

👉 You apply it in a specific context

This leads to the following:

→ I want {x} goal,
→ in {y} context,
→ using {z} metric.

A bit fuzzy?

Here are 2 easy examples:

1️⃣
→ I don't want to be cold (goal);
→ Should I wear a coat today (context)?
→ What's the temperature outside (metric)?

2️⃣
→ I don't want to get fined (goal)
→ This road has a speed limit (context)
→ I should watch my speed (metric).

All good?

It's just common sense stuff so far.

When we use a metric, we expect it to be accurate and precise so that we can rely on it in a specific context.

Now let's look at two metrics loved by youtube gurus:

CTR &amp; AVD

Are they metrics? Absolutely 👍

Are they working fine? Absolutely 👍

BUT

what are they measuring EXACTLY?

(Remember, metrics measure QUANTITATIVELY something)

They measure (precisely and accurately) this:

CTR: Measures how often viewers watched a video after seeing an impression.

AVD: Measures the total watch-time of a video divided by the number of video views (including replays) a video gets.

If they also are an indicator of views (in their current state) then why is it not working?

I'm sorry let me say it again:

*ahem*

WHY IS IT NOT WORKING???????????????????????????????????

(proofs at the end of this thread)

Do you want to know how many clicks you had per impression?

→ Look at CTR

Do you want to know how much watch time on average a video accumulated?

→ Look at AVD

That's what they measure quantitatively right now. (In their current state, creator side)

Higher CTR &amp;/or AVD = more views is a FAIRY TALE. (again, in their current state)

Do you know why my dear clueless youtube gurus?

BECAUSE IT DOESN'T WORK IN REALITY!!!!!!!!!!!

I'm sorry was I arrogant again? Excuse my french 🤦

Higher CTR = more clicks per impression? Yes ✅
Higher AVD = more watch time per view? Yes ✅

Higher AVD = More views? No ❌
Higher CTR = More views? No ❌

(in their current state)

At this point, this is where youtube gurus say:

"But are you stupid? That's not what we said, we said they were important metrics to watch. They are only important in context."

And this is where their level of incompetence goes exponentially up.

Not understanding what a metric is?

That's fine, it's a noob mistake.

You were fooled by the youtube analytics UI highlighting CTR/AVD (a mistake imo), you have no experience (or didn't learn from it), fair enough no big deal.

Everyone makes mistakes, admit it and move on.

However, saying:

"They are IMPORTANT metrics to watch to get views"

AND

"they are only important in context"

in the same sentence, that's a VERY dangerous kind of incompetence.

As dangerous for creators as a 2 y/o holding a loaded gun.

Why? Because youtube DOES NOT provide the context. 😂

There is a word for people who claim to have the ability to read beyond the observable reality: "oracle".

(There is a real life analogy a few tweets below so you can understand what I mean)

Now the rhetorical loaded gun:

"The truth lies in between" is a strong (often wrong) popular belief milked by demagogues.

Many people use that as a virtue-signaling posture to show they are wise and speak in a measured way (when in fact they are simply clueless).

It is a powerful rhetorical tool (used a lot by politics), because unless you know how to counter it, it is an argumentative trap.

No matter where the truth really is, the person using it can't lose.

It's a great way to keep the high ground without taking risks.

This is very easy to counter though, especially when we are talking about non-abstract things (concrete and measurable stuff).

Take for instance 1+1 = 2

The truth is not in the middle, the result isn't 1,5.

"Ofc youtube provides context, you have other metrics!"

I'm not talking about that context. (Because if we talk about that context, views are 100x better for signal)

I'm talking about context BEHIND these metrics (CTR/AVD).

Specifically:

1️⃣ How much do they weigh in the balance for the algorithm and in which conditions EXACTLY

2️⃣ We have the GLOBAL average but what is behind this average exactly? (there's no information right now)

Without at least one of these, we can't rely on them.

Hard to follow?

Check this very easy-to-understand analogy, illustrating why context behind average is crucial, with a metric everyone understands 👇

→ Temperature. 🌡️

So explain to me:

HOW can anyone understand an average without knowing what's behind it?

It's impossible at scale.

It's like trying to understand a puzzle of 1000 pieces with only 2.

Averages are good if you look for a general tendency or have a rough idea of something.

But when you look for precise information, like:

→ "Why this video performed badly?"

it is terrible because it hides what you need to know (signal) behind an average (noise).

In that case, you want to know exactly what factor caused that.

You want to be able to look at CTR and tell:

👉 the problem is my thumbnail or title

 or

👉 the problem is something else

You look for signal, not a rough idea.

Here is another example:

Say we want to compare 2 channels:

First one:

• 10 videos at 1M views

Second one:

• 9 videos at 1k views
• 1 video at 9 991 000 views

Both have 1M average views.

Are they the same?

If you only look at the global average, yes.

If you look behind it, no.

Now if you tell me that:

CTR is supposed to measure how well your thumbnail and title catch attention.

AVD is supposed to measure whether viewers enjoyed/found your video entertaining/interesting or not.

Right now they both fail at it, too much context is missing.

Now it's time to prove the conclusion of this graph wrong since they've been copy-pasting it left and right (from the thread below).

Here, @LowKeyJude says:

• "that demonstrate a correlation between AVD*CTR &amp; views"

• "based on the data, there is a correlation between AVD*CTR"

• "Completely dismissing these metrics is misguided and harmful for other creators who read your tweets"

Full version below:

Now let's do some maths (if you're allergic don't worry, I'll make it easy to understand)

I want you to focus on this number from the graph: R² = 0.096 (pic below)

Now, first, read the rules in the pics below.

The model predicts the outcome 9.6% of the time (R² = 0.096)

Do you know what this means?

That it does NOT predict the outcome 90.4% of the time.

Ouch.

The conclusion of this graph is, in fact, the exact opposite:

90.4% of the time, there is no correlation.

Correlation!!!!

We are not talking about CAUSATION yet!

I could end this thread right here and it would be game over.

But it wouldn't be fun, right @LowKeyJude?

You brought your friends over to the party, so why stop here? 🤗

Now let's spice things up:

Let's admit the R² was not 0.096 but 0.99999.

Well, guess what, it still wouldn't prove the causation.

Why? Because of 4 things:

1. Spurious correlations

The more variables, the more we can find correlations that are not causations.

Here is an example:

(more examples here: )

CTR &amp; AVD (same for the retention graph) are just 2 metrics among {x} others.

In other words, 2 pieces of a puzzle youtube provide to creators.

Since we do not know all the variables youtube look at, we can't prove AVD*CTR is not a spurious correlation.

Even at a 99.9% correlation.

Just like my potato = fish example, it can correlate without being causation (as long as the fishing area is full of fish).

What is even sadder is:

The correlation between:

• The number of people who drowned by falling into a pool

and

• Films Nicolas Cage appeared in

is 7 times more relevant than AVD*CTR 😂 (correlation speaking, if we occult causation)

2. Simpson's Paradox

The only information we have about the videos in this data set is:

"2400 videos (gaming niche)"

Here is how you can manipulate data with Simpson's Paradox:

And in this example, key elements are missing:

• how many channels (and their name)
• how old they are
• how many subs
• how many videos (per channel)
etc..

If the videos were just selected based on being "gaming" they hold a lot of noise, hiding signals even more.

For instance:

• 100k views with 10 000 000 subs = noise (audience)
• 100k views with 1000 subs = signal (remarkable)

Same number of views, different conclusion.

3. Logarithmic view manipulation

I won't go too much into detail here because it would be too technical (not the goal).

It's very visual tho, look at this graph:

• Linear (normal) scale on the left
• Log scale on the right

More info here:

This graph is on a log scale.

This means that on a normal scale, the dots would be way more spread (up and down).

Look at the difference of views between these dots, they are not far away in terms of distance, but the views are 10x higher each time:

⚠️ I'm not saying the author did that on purpose to deceive the reader. Again, I have nothing against the author.

The log scale can be useful for many things, especially to read exponentials, but it gives a false sense of high accuracy.

4. Manipulation using averages

Averages are great for many things but can also lead to completely absurd conclusions.

Here is a dead-simple example of an absurd conclusion using average:

That's the problem with averages, they hide variations.

It's fine if you want to study the average, but terrible when you want to study outliers (high performing videos).

In the example above, it only took 1 outlier to completely fuck up the relevancy of the average.

In this graph, the average is between ~8k &amp; 80k views.

The problem is, the average carries too much noise (for the reasons I already mentioned).

Again, we want to study outliers, not the average videos on youtube.

And now to nail the coffin here is a graph my associate made and shared with me.

Translation below for non-math people.

This prediction uses their AVD*CTR model to get a 500k views video.

Red is the area where it predicts more than 500k views, white less than 500k and the red dotted line 500k.

Taking into account the fact that nearly all videos over 5k views on YouTube have a CTR below 20%.

According to the AVD*CTR model, you need an AVD higher than 35 minutes (2100 secs) to get a video above 500k views.

Conclusion:

👉 Videos under 35 mn with over 500k views on youtube are extremely rare

Alternative conclusion: AVD*CTR is bullshit.

Believe which one you want.

But it's not over lol.

The author of this thread mentions a paper from Google, and says "Google outlines this", in reference to AVD*CTR = expected watch time per impression.

So I downloaded it.

And that's not what the paper says at all.

Here is the full context:

• In red is the quote used in the thread (screen above)

• In yellow, the context

The paper here says youtube uses two neural networks:

 • "candidate generation"
 • "ranking"

Or in simple words:

• This is a list of 100 videos this user is likely to like ("candidate generation")

• Ranked from 1 to 100 by "likely-ness" ("ranking")

(100 is arbitrary)

For the user. (pic below)

In other words, the paper talks about "ranking per user", because obviously, each user is unique (personal watching patterns and center of interests.)

Saying "Google outlines this" (CTR*AVD = expected watch time per impression) is wrong.

There's a difference between:

Expected watch time per impression PER USER (what the paper says)

and

Expected watch time per impression that is GLOBAL, blending all users together

It's explicitely stated in this FAQ about CTR: