Notes on YouTube

I just quit YouTube. And as a parting gift of sorts, I’ve compiled all of my notes into a blog post that is hopefully cohesive.

These notes cover when I ran my channel Andrewkfb between January 2023 and March 2024 for context.

This post is split into two parts. Big Notes is where you’ll find information covered in the video (it’s basically a transcript of it). Extra Notes is where you’ll find everything that didn’t fit in the video.

Big Notes

Start Sooner rather than Later

• Anytime you watch a YouTube tips video, you’re going to hear about the debate of quantity versus quality of the videos. Here’s my take on it.
• Start with quantity. Make as many videos as you can as fast as you can, but only for a short while.
• Why? Because by doing that, you’re going to figure out what quality is.
• When you first start, you have no context, no feedback from the audience. You really just have your own gut intuition to go off of, and while that’s helpful, it’s not enough to make it sustainable.
• The sooner you start making videos, the sooner you’re going to figure out what kind of quality your viewers are expecting. Then you can slow down the upload schedule and take your time to make a banger video for your audience.
• Speaking of the audience…

Start with the Audience, Not the Content

• If you want to start a YouTube channel, you probably have an idea of what you want to make. That’s fantastic!
• But before you create the video, I ask that you think about the type of person who will watch your videos. Try to be specific.
• In my case, I thought that “people who like movies” was specific enough, but as it turns out, I was wrong.
• I attracted people from specific franchises who were only interested in those particular names, like Star Wars, One Piece, or Godzilla.
• You might be thinking, “Shouldn’t I be trying to attract a wide audience?” That’s what I thought, but I found that trying to appeal to a wider audience meant that when specific people came to my channel, I didn’t quite scratch the itch for them.
• If I could start over again, I would have refocused the channel on something even more specific like Godzilla, movie subscription services, or Hollywood commentary videos or whatever. The channel would have had a solid focus to build a smaller but more loyal audience, which I can later expand into a broader one as the channel grows.
• But alas, we live and learn.
• If that sounds vague, well, how about this?

Compose a Video Recipe

• This goes hand-in-hand with nailing down your specific audience.
• Think about when you watch a TV show or listen to a podcast. They generally evoke the same emotions, cover similar topics, and do so in a similar style every time you watch an episode. You want your videos to be the same way.
• When a viewer comes across one of your videos and likes it, they’re probably going to want to see more like it. But if they start looking through your catalog and it’s not really that similar, then they’re probably not going to subscribe.
• A textbook example is the man himself, MrBeast. When I say “MrBeast video,” you’re probably going to think “big money prizes,” “philanthropy,” “obnoxious editing.” While I do not suggest copying MrBeast for many reasons, you want a similar effect. You want people to look at your channel and be able to identify a few key attributes that make your channel memorable and make them want to keep watching your videos.
• You do that by forming a video recipe. Start with things like similar topics, similar video length, similar style, and similar music. The ingredients you carry over from video to video are up to you, but the important thing is consistency.
• Although regarding consistency, that comes with one caveat.

Experiment, but not too much

• Because you might be wondering well, how do I form my video recipe? Well that’s where you get to make videos and experiment!
• Allow yourself some freedom to try new things and, if necessary, pivot.
• Maybe you wanna create an “AI Tools for Noobz” channel with 5 minute long videos each covering a specific tool. But then you realize that most of these AI tools are just ChatGPT skins so what’s the point? Well, maybe make longer videos commentating on the development of several AI models? Or maybe you realize that AI is an incredibly competitive topic whereas ranking your favorite lollipops is not.
• This is the kind of discovery you can only make once you’ve started making videos and only you can decide what to do with it.
• Again, you have to strike a balance, I doubt the AI crowd is gonna care about your lollipops, ya know what I mean?
• Though there is one area I advise you to experiment liberally and that is…

Experiment liberally with Community Posts

• I think Community posts are gonna be huge for YouTube going forward.
• They’re pushing them on the homepage and giving creators more tools to make them. They’re even experimenting with community spaces, which can basically turn your channel into a subreddit. Personally that idea gives me a headache when it comes to moderation and trolling, but I digress.
• Point is that there’s lots of ways to use community posts to get reach for your channel. Polls with images are the best way that I’ve found, some of which got more engagement than many of my videos lol.
• So as you build your channel and look at how to grow it, use Community posts as much as you can.

Modern YouTube SEO comes from Transcripts and Subtitles

• Speaking of reach, SEO! Everyone’s favorite topic. I hate it, but it’s still important.
• Let’s get some things out of the way. YouTube looks at the text in your titles and descriptions for keywords that will help them match your video with potential viewers. They used to use tags, but they’re getting phased out as the algorithm gets smarter.
• But the most underrated way to give the algorithm a better idea of who your video appeals to is with the transcripts and subtitles. I’m not talking about YouTube’s automated stuff. I mean you should get a tool that can transcribe your video for you, and upload the transcription with your video.
• There are lots of tools for that purpose. I use one called Descript, not sponsored just highly recommend it. It takes care of my transcriptions and subtitles for me. It also has some AI features like background removal, audio cleanup, and a few other things.
• And now that I brought up that, I should probably talk about AI.

Use AI Wisely and Ethically

• As of 2024, Youtube wants you to label your content when it has AI in it, although I think there’s more too it than that, best to double check their statements on it.
• Whatever your feelings about AI, it’s here.
• And there’s a lot of pro AI and anti AI opinions out there. I just wanna give my own thoughts as of… August 2024.
• Like I said earlier, I used Descript’s AI features to help me work faster and smarter. What I didn’t do is use AI to do my work for me.
• There’s tons of videos about how to use ChatGPT to write a script, Eleven Labs to record a narration, and InVideo to automatically edit a video for you with minimal effort. I think “YouTube Automation” was the buzzword for that. All fun in theory, except they usually turn into garbage videos.
• I think AI is useful a tool, and it can make a good video better, but I have yet to see it make a bad video good. Ya know what I mean?
• Anyways, rant over, that’s how I feel. So back to YouTube and my final note.

Think of it like a Podcast

• Don’t START a podcast, just think of your channel like one.
• I can’t be the only person who noticed that Podcasts do well on YouTube and I think I figured out why.
  • They have a consistent upload schedule and selection of topics.
  • The format of each episode is usually similar between all of them
  • The best moments can be broken down and turned into Shorts if you want.
  • If you have a small update or question, you can just do a community post without needing to make a full video on it.
  • They tend to be longer and do well in terms of watch time and retention (if you do it right).
  • They don’t require advanced editing, it just needs to be edited enough to be engaging.
• I think that when I come back to YouTube, this is how I will do things and I wished I thought of it sooner.
• I think it gives you a basic video recipe that you know already works but that you can mold to your liking and give your own creative spin to.
• Again, it doesn’t need to BE a Podcast, just…think of it like one.

Extra Notes

Watch & Search a LOT of YouTube

• Doing so will help you understand the market/niche you’re interested in making content in. You’ll can gain insights into what is popular, what works, and what doesn’t, helping you to make your videos stand out.
• You can spot gaps where viewers want more videos or spot ways you can do it differently or better.
  • I’ve seen some channels where they video topics aren’t that original, but they tackle the Titles and Thumbnails in a way that’s more engaging to that particular realm of videos.
• It’s also helpful for inspiration. Mixing elements from various realms can spark new ideas and creative approaches.
  • Instead of doing a regular TikTok compilation, how about you add some more editing like gaming clip compilations tend to do?

Content Recipe Notes

• Based on my analytics, viewers tend to gravitate towards specific video lengths (livestreams/replays are different, I’ll talk about them later):
  • 20–40 seconds (for Shorts)
    • Ideal for quick engagement and reaching new audiences on the Shorts Feed or Search.
  • Around 3 minutes
    • Concise videos that get straight to the point are great for viewers with shorter attention spans on the Homepage.
  • Around 5 minutes
    • YouTube’s “classic” length imo that’s a great default length for creators just starting out.
  • Around 10 minutes
    • Perfect for videos with more details in a way that keeps interest from viewers.
  • Around 20 minutes
    • Longer videos are great for comprehensive content, like in-depth tutorials, reviews, or discussions.
• I loved experimenting with different content styles on my channel, but recently realized this was a bad idea.
  • Viewers were having a hard time understanding what my channel was about.
  • If viewers liked a specific type of video that I no longer made, they wouldn’t stick around.
• Random aside, most of the subscribers from regular videos came from the video page, meaning they didn’t explore my entire channel.
• In theory, having a recipe could help with the video production process. I spent chunks of time trying to find inspiration for a new type of video to make, and ended up not making one at all sometimes. If I had a recipe for my channel, I feel like I would’ve had a better jumping off point and easier production process.

Don’t Look at the Analytics

• Avoid checking your analytics every day. It can get time-consuming and mentally draining very quickly.
• Once a week is optimal because it lets you gather enough data to see meaningful trends without becoming obsessed with daily fluctuations.
• It’s easy to get caught in the loop of watching the views trickle in when you’re first starting out. But it can lead to an unhealthy focus on the numbers rather than the quality of the videos. This loop can lead to more anxiety than happiness over time.
• While we all want our channels to blow up, you’ll know when it happens through notifications and comments. You don’t need to constantly check your analytics to know when something is performing well.
• I remember uploading a short that “blew up” to 12k views in 24 hours, and the high from that really motivated me. However, when subsequent videos didn’t perform as well, I felt the lows much more.
• By checking your analytics at set intervals (once a week imo), you can remove emotion from the equation and view the numbers more clearly.

Some Views Are More Valuable Than Others

• We’re always chasing views with the idea that they have value, but that’s not always the case, especially if you’re trying to make money on YouTube.
• CPMs aren’t a good metric for valuing views since they can vary so wildly due to factors you can’t always predict or control (topic of the video, time of year, etc.).
• Even if you do YouTube purely out of joy, this still applies. Engaged viewers can lead to a more fulfilling content creation experience.
• I find that the longer the video the more “sticky” the subscriber. Shorts subscribers
• The best way to turn views into money is to use the attention views get you into something more straightforward that DOES make you money. This can include affiliates, merchandise, or selling your own book.

A “well designed” thumbnail is not the same as an enticing thumbnail

• Thumbnails aren’t about being well designed or smart fully composed images. They are about standing out from the sea of other thumbnails in a way that makes them enticing.
• In this sense, the thumbnail and title should work together to sell a viewer on what the video is about and what value they’ll be getting from it
• YouTube has made two updates/evolutions that help small creators in this regard:
  • Creators can now test different thumbnails (up to 3) to see which one resonates more.
  • Small channels generally have greater visibility on the homepage and related sections than they did a 2 years ago. I suspect YouTube has wanted to replicate TikTok’s success by giving smaller creators a better chance of popping off. But the increased impressions gives inexperienced creators a better chance of honing their craft and finding an audience.

Block Negative Words

• There’s a feature that lets you add recurring negative terms and slurs to a list, and any comment that includes those words is automatically hidden.
• It blocks some of the basic stuff by default, by I’d add names of politicians or more obscure slurs.
• Sometimes you just need to block mundane stuff because you’re tired of dealing with it (for me it was Scott Pilgrim, if you know then you know).

Shorts Have an Entirely Different Meta

• The algorithm for Shorts operates VERY differently from regular videos, prioritizing videos that capture attention within the first 3–5 seconds.
  • “Successful Shorts need a strong hook and quick, engaging edits to retain viewer’s attention” seems obvious, but in practice I find that it not only works but can be easy to do with some practice.
• Thumbnails have virtually no impact on Shorts since they are mostly seen in the Shorts Feed. The main exception to this is if the Short shows up in search.
  • There’s still the hack of adding a custom thumbnail image into the Short via editing then trimming it out in YouTube Studio, but honestly I don’t think it’s worth the effort.
  • I haven’t tested this extensively, but each time I tried this hack for a custom Shorts thumbnail it didn’t significantly impact the video’s performance.
• The same goes for titles, though be thoughtful about them as they still serve some SEO purposes, so crafting thoughtful titles is important for discoverability.
  • On a similar note, I’ve found using hashtags in the titles or descriptions of Shorts to be hit or miss. They would never show up as a significant view driver, probably because no one uses the hashtag feature.
  • Also, adding “#fyp” to your short makes you look incredibly stupid since the “For You Page” is on TikTok not YouTube, so dismiss any gurus who say you should add that hashtag or anything similar like “#viral.”
• Shorter Shorts (in my experience this is 10–25 seconds) generally perform better as they are more likely to be watched in their entirety.
  • My guess is that, statistically, this shows up as a higher percentage viewed in YouTube’s eyes, so the algorithm will see this as a positive signal and push it more.
  • I imagine the “ideal Short length” is gonna vary between niches/topics, though I have no data to prove that.
• Shorts have a feature where you can link another video (right above the title if you didn’t know. Your results may vary, but here’s what I’ve observed:
  • Sending people from Shorts to another Short works 30% of the time. Less than I was hoping for.
  • Sending people from Shorts to a long video works less than 5% of the time. This is appalling, basically useless unless the Short has a million views.
• The shorts algorithm (at least in the year 2023) is modified in 2-month cycles, based on my data.
  • And what changed in each cycle was completely unpredictable.
  • For example, one cycle had it that my videos would not be put into the Shorts feed until exactly 6 hours after they were posted. I had to time my releases accordingly.
  • I think radical swings like that have died out, at least from what I’ve observed on my channel, but I put this out as a warning in case anyone else experiences something like this.
• Even if you do everything right, it’s challenging to make a significant impression on an audience with only Shorts.
  • They are very good for making a lot of content and experimenting with recipes and niches.
  • But I recommend they be used as part of a larger strategy that includes regular videos, community posts, possibly podcasts or livestreams.

Live Streams Are Hard, But They Might Be Worth It

• Livestreams seem to get a visibility boost from the algorithm.
  • Even with people outside your subscriber base, similar to Community Posts but at a much smaller scale.
• Vertical livestreams seem to be heavily favored in the Shorts feed at the moment, but I see that dying down.
• Always test a stream before going live; something WILL go wrong.
  • Technical issues are almost inevitable. A test run helps you catch and fix problems in advance.
  • After testing, make sure to delete any test streams to keep your channel tidy and professional.
• Livestreams are particularly effective at garnering watch time.
• The “starter level” of the YouTube Partner Program seems to favor channels that primarily livestream.
  • The first level of it unlocks Super Chats, Memberships and Shopping. These are features helpful to all creators but live-streaming-focused ones would benefit the most.
  • I theorize that it was set up like this on purpose so that whenever people leave Twitch they’ll recognize YouTube as the better platform for them.

Community Posts Are a Hidden Gem

• To the surprise of no one, Memes do really well.
• There’s a lot of reposting going on, which is dumb.
  • This is not only lazy/derivative, but it misses the potential for meaningful interactions with your audience and potential viewers.
• They get a TON of reach from people who’ve never seen your videos before, so try to compose them accordingly.
  • I’ve seen some creators use them for quick updates (myself included) and Posts do seem to be effective at that.
  • But I also see a lot of channels getting subscribers simply for making funny posts or sharing news in a similar tone/manner as Twitter/Threads.
• And unlike Shorts, you can meaningfully link to something and it will get more impressions.
  • Oddly enough, standard web links in Community posts often perform better in driving traffic compared to embedded YouTube video links.

Notes on YouTube Podcasts

• Like I said earlier, YouTube is becoming a great place for podcasting because the podcasting formula suits the algorithm very well.
  • They usually have a straightforward recipe that allows for deep conversations and great stories that can earn great watch time from viewers/listeners.
• Episodes can be recorded via livestream.
  • I’ve seen some creators double dip into this trend by recording their podcast episodes during livestreams, which adds those benefits to podcasting.
  • It seems to work really well for engaging the audience and for time efficiency (since you can just use the livestream VOD and edit it if need be).
• The podcast playlist lets you send it to YouTube Music.
  • YouTube Music is basically Spotify at this point, for better and for worse.
  • By using the podcast playlist feature, you can make your episodes available on YouTube Music.
• Be selective with your podcast playlists
  • Some channels set EVERY playlist as a podcast. I think this dilutes the impact and confuses viewers.
  • Think about it: if you go to the Podcasts section of a channel and see 50 playlists there, are you gonna listen to any of them?
  • When it comes to how a potential subscriber sees your channel I think less is more, so only using the feature when it makes sense

Hope all this was helpful to someone out there. If it was, let me know. Take care and I’ll see you next time!