YT YouTube2Text

YouTube2Text → Transcripts → The Complete Guide to the YouTube Algorithm

The Complete Guide to the YouTube Algorithm — Transcript

by vidIQ · 4,309 words · 662 segments · language en · Watch on YouTube

Open in transcriber Summarize with AI Download SRT/VTT

Full transcript

0:00Why did your video get 300 views when a
0:03channel just like yours did a similar
0:04video and got 30,000 views? Why is it
0:07that a channel with 50,000 subscribers
0:08can outperform a channel with 5 million
0:10subscribers? Why would a new video do
0:12nothing and then 3 days later suddenly
0:14explode in views? Well, the answer to
0:16all three of those questions is the
0:17same. The algorithm isn't doing what you
0:20think it's doing. And that gap between
0:21what you think is happening and what's
0:23actually happening is costing you views,
0:25subscribers, and growth every single
0:27day. So, today we're going to close that
0:28gap and explain the YouTube algorithm at
0:30three different levels: beginner,
0:32intermediate, and expert. Each one is
0:34going to go deeper than the last, and by
0:36the end of this video, you'll have a
0:37better understanding of this system than
0:39most creators on all of YouTube. So,
0:41let's start with the basics. And this
0:42time we're not going to water it down
0:43like we have in the past to save time.
0:45The YouTube algorithm is not one thing.
0:46It's a collection of systems working
0:48together, each one with one job: find
0:51the right video for the right person at
0:53the right moment. That's it. And to do
0:55that, it asks two questions about every
0:57single video on the platform. Question
0:58one, [music] will this person click?
1:00Question two, if they click, will they
1:03actually watch? Those two questions map
1:05to two numbers in your YouTube Studio
1:07dashboard right now. Click-through rate,
1:09which is CTR, and average view duration,
1:12which is AVD. If you're a small channel
1:14trying to track as few numbers as
1:15possible, make it these two numbers. But
1:17here's what most beginners get wrong.
1:19They think the algorithm promotes their
1:20video to an audience. It doesn't.
1:23YouTube's own engineers have said this
1:24directly. The algorithm doesn't push a
1:27videos to people. It finds videos for
1:30people. But even then, before the
1:31YouTube algorithm can find the right
1:32person for your video, it needs to
1:34understand what your video is actually
1:36about. Todd Beaupre, YouTube's head of
1:38discovery, confirmed this publicly.
1:39Knowing a video is about home workouts
1:41or Minecraft isn't enough. There's only
1:43so much the algorithm can get from the
1:45title and description alone. It needs
1:47richer signals, consistency of language,
1:49clarity of framing, and real audience
1:52behavior to back it up. So, it reads
1:53your title, your description, your
1:55chapters, it transcribes your audio, it
1:57watches who clicked, who stayed, who
1:59left. All of that becomes one signal
2:01that answers one question: who is this
2:03video for? And this is why some of the
2:05fastest growing channels in YouTube are
2:06built on radical simplicity. Let's take
2:08a look at the Organic Chemistry Tutor.
2:10The subject matter is genuinely hard:
2:12calculus, physics, organic chemistry.
2:14But every title is completely plain. How
2:16to find the domain of a function. How to
2:18solve quadratic equations. No metaphors,
2:21no mystery, just a direct description of
2:22exactly what the video teaches. The
2:24algorithm never has to guess who it's
2:25for. The result is over 10.5 million
2:28subscribers in one of the most
2:29technically demanding niches on the
2:31platform. But math is scary, so let's
2:32also take a look at Daily Dose of
2:33Internet. The channel itself is a
2:35complete signal. A quick dose of
2:37interesting internet content every day.
2:39The title, content, channel name, it's
2:42all one consistent message. YouTube
2:44never had to guess. That clarity
2:46compounded across hundreds of videos
2:47drives over a million views on every
2:49single upload. Contrast that with
2:51unclear signals. A creator titles a
2:53video, "This changed everything for me."
2:55Life hack, shocked face in the
2:56thumbnail, 3 minutes of backstory before
2:58we get to the actual content. The
2:59algorithm genuinely can't tell if this
3:01is a vlog, a finance video, a lifestyle
3:03channel. And a confused algorithm
3:05doesn't like to take risks. This is why
3:07niche clarity matters so much early on.
3:09Every off-topic video essentially resets
3:11the picture YouTube is building of your
3:13audience. But every consistent video
3:15sharpens it. And we don't have exact
3:17numbers, but presumably 10 consistent
3:19pieces of content in a row starts to
3:21tell a story. Make it 50 and YouTube has
3:23a reliable model of your channel. Who
3:26watches? What do they care about? How do
3:28they behave? Now, once people click, you
3:30have to deliver. And this is where watch
3:32time and retention come in, but a lot of
3:33beginners confuse the two. Watch time is
3:35the total amount of minutes people spend
3:37watching. If 1,000 people each watch 5
3:40minutes of a video, that's 5,000 minutes
3:42of watch time. YouTube cares about watch
3:44time on the platform because more watch
3:45time means more ad revenue. Retention
3:47though is different. It's the percentage
3:49of your video that the average viewer
3:50actually watches. If in a 10-minute
3:53video, people drop off after 4 minutes,
3:54that's 40% retention. YouTube shows this
3:57on a curve, and that curve tells a
3:58story. A steep drop in the first 30
4:00seconds means your opening didn't
4:01deliver on what the thumbnail promised.
4:03A gradual slope is normal. But a sudden
4:06cliff halfway through the video means
4:08something specific happened. A topic
4:10shift, a tangent, a pacing drop. YouTube
4:12notices every single one of these. And
4:14YouTube doesn't look at retention in
4:15isolation. It compares your video to
4:17other videos of a similar topic and
4:19length. A 6-minute video 60% retention
4:22can outperform a 12-minute video with
4:2350%, even though the longer one has more
4:26raw watch time. By the way, speaking of
4:28length in terms of content, one more
4:29thing to note here. YouTube recently
4:31announced that TV watch time had
4:33overtaken mobile for the first time
4:35ever. People are watching YouTube like
4:36they do Netflix, leaned back, long watch
4:39sessions. That's why the average video
4:41length from the top 100 YouTube creators
4:43has nearly doubled in recent years. If
4:44your audience watches on TV and your
4:46videos are 5 minutes long but have the
4:47potential to be longer, you might be
4:50leaving real watch time on the table.
4:51So, the beginner framework is this.
4:53Clarity first. Be crystal clear about
4:55what your video is, who it's for, and
4:57what you're promising. Then your
4:58thumbnail gets the click. Your opening
5:00keeps them watching. They watch as much
5:02as possible, earning you watch time, and
5:04the algorithm is taught who your videos
5:07are for. So, by the next time you
5:08upload, a much clearer picture is
5:10painted for your channel. If you get all
5:12that right, you are no longer fighting
5:13with the algorithm. It is now a tool
5:15that you're using to your advantage.
5:16Moving on now to a different type of
5:18content creator. Someone in the range of
5:20maybe 5,000 to 100,000 subscribers. If
5:23you're in this category, you understand
5:24CTR, you know what retention is, and
5:25you're doing the things you're supposed
5:27to be doing. And even still, it is very
5:29possible you feel stuck. If that sounds
5:31like you or a far-off future you don't
5:33want, then that's what this section is
5:34for. And let's start with something that
5:36most creators don't think about. Where
5:38are your views actually coming from?
5:40Open YouTube Studio right now. Click on
5:42analytics, then traffic sources. You're
5:44going to see three main categories.
5:46Browse, that's the home page. Suggested,
5:49which is the sidebar next to a video you
5:51watch. And search. Most creators look at
5:53the screen and think, "Oh, this is
5:55interesting data." It's not. Look beyond
5:57the face value. Each source tells you
6:00something completely different about how
6:01YouTube sees your channel, and each one
6:03rewards completely different content
6:05behavior. Browse traffic comes from your
6:07video showing up on somebody's YouTube
6:09home page before they've searched for
6:11anything. They've just loaded the
6:12website. That's YouTube spending its
6:13most valuable real estate on your video.
6:16But, there's two common misconceptions
6:18that are worth clearing up here. The
6:19first is that browse is only for the big
6:21channels, and it isn't. You've probably
6:23noticed small channels appearing on your
6:25home page from time to time, and that
6:27right there is YouTube doing its early
6:29discovery tests for new content. What
6:31they're doing is seeding a new video to
6:33a small group of people whose watch
6:34history suggests that they would
6:36probably enjoy that kind of video. They
6:38do this for channels of all sizes, and
6:40that kind of makes sense, because
6:41otherwise, everyone would literally have
6:43zero subscribers, and that would be it.
6:44Speaking of subscribers, the second
6:46misconception is that browse is driven
6:47by subscribers. It is not. YouTube has
6:50shifted to a viewer history model. If
6:52someone has watched your last two videos
6:53and didn't hit subscribe, YouTube is
6:55going to put your next video on their
6:57home feed at some point anyway, because
6:59they watched your last two videos.
7:01Meanwhile, a channel that has a million
7:03inactive subscribers isn't going to get
7:05that much browse traffic. Just having
7:07subscribers does not earn you browse
7:09traffic. Recent, consistent viewership
7:11does. Now, suggested traffic comes in
7:14play when your video is recommended
7:16alongside somebody else's video. For
7:18example, if you're watching this on a
7:19browser and you look off to the side, I
7:21think the side, there's going to be a
7:22bunch of videos over here. These are all
7:24suggested videos. I really hope they're
7:26on this side of the screen. Otherwise,
7:27I'll look very silly right now. If
7:28you're watching on mobile, you can
7:29scroll down and there'll be suggested
7:30videos at the bottom. I think this is
7:31the bottom. Most creators think getting
7:33suggested traffic is purely about a
7:35topic match. Make a video about the same
7:37subject as a larger channel, and YouTube
7:39routes their traffic your way. That's
7:41partially true, but the updated reality
7:43is a bit more precise than that. Again,
7:45you can see this in play for yourself
7:46right now. Essentially, YouTube has
7:48shifted away from matching topics to
7:50matching sessions. The question it's
7:51asking now isn't just is this video
7:53about the same thing? Instead, it's
7:55asking if somebody watches video A,
7:57would they also like video B? Again, the
8:00difference here is subtle, but
8:01important. The clustering is also got
8:03more specific. It's no longer about
8:05categories like gaming or cooking.
8:07YouTube now groups by micro niches, a
8:08specific combination of topic, tone, and
8:11audience. A relaxing architecture
8:13channel might get suggested next to a
8:14productivity channel, not because the
8:16topics overlap, but because the audience
8:18is the same person. Which is awesome
8:20news because if your content fits that
8:21vibe, then you could end up appearing
8:23next to channels 10 times the size of
8:24yours. That is how small channels break
8:26through without having a single viral
8:28moment. Next up, there's search traffic,
8:30and this is pretty obvious. This is
8:31people searching for the thing that you
8:33talk about on YouTube. Search rewards
8:35clarity and consistency of language. You
8:37using the same words that your audience
8:39would use to describe a problem that
8:41your video solves. Search traffic can be
8:43slower, but incredibly durable. A
8:45well-optimized video can drive views for
8:47years. And now we get to talk about the
8:49metric that sits above all three of
8:51these, the one YouTube cares about more
8:53than all of them combined, and that is
8:55satisfaction. YouTube runs millions of
8:57viewer surveys every single day,
8:59randomly asking people what they thought
9:01of videos they recently watched. Those
9:02responses feed directly into the
9:04recommendation system. YouTube is not
9:06just measuring whether people watched,
9:07it's measuring whether or not they're
9:08glad they watched. I want you to search
9:10for a video called We Sent Garlic Bread
9:12to the Edge of Space, Then Ate It by Tom
9:15Scott. The title says exactly what it
9:16is. Someone put garlic bread in a
9:18weather balloon, sent it 35 km into the
9:20stratosphere, brought it back down, then
9:23ate it. This is one of the most viewed
9:25videos on the entire channel with tens
9:26of millions of views. Let's imagine for
9:28a second that you only know the title of
9:30this video. What do you think the
9:32average YouTuber today would do with a
9:34video titled We Sent Garlic Bread to the
9:36Edge of Space and then ate it. Most
9:37creators making video like that would
9:38probably do a whole bunch of stuff. They
9:40would film the entire planning process.
9:42They would vlog the journey to the
9:44launch site. They would drag out the
9:45suspense and turn a 6-minute concept
9:48into a 20-minute experience. Sure, this
9:51will add to the run time of the video
9:52and potentially get you more watch time.
9:54Maybe people even make it to the end of
9:55that video. But with all of that
9:57padding, going from click to biting down
9:59on garlic bread might make people feel
10:01like they were a bit shortchanged. And
10:03when the time comes, they may leave a
10:05pretty mediocre review on that
10:06satisfaction survey. Meanwhile, here's
10:08what Tom actually did. The video starts
10:10with a piece of garlic bread, and within
10:11the first 5 seconds, it's already on its
10:13way to space. The promise made in the
10:15title is kept right at the front of the
10:16video. No unnecessary build-up, no
10:18filler, just the concept executed
10:21cleanly from start to finish. He
10:22sacrificed real watch time, but gained
10:24something worth far more. Millions of
10:26people clicked, watched the whole thing,
10:29and saw something genuinely surprising,
10:31and left feeling satisfied. This idea
10:33hopefully changes the way you think
10:35about creating content on YouTube. And I
10:37hope the takeaway here is, "Oh, I guess
10:39he's saying make shorter videos."
10:40Because it's not. A video with strong
10:42watch time, but low satisfaction can get
10:45suppressed over time. The point being is
10:47that every minute you add to your video
10:49should be worth the viewer's time. The
10:51ultimate goal is raising satisfaction.
10:53If somebody leaves your how-to video at
10:56the 3-minute mark because they got
10:57excited and automatically just wanted to
10:59try the thing they just learned from
11:00your video, that isn't necessarily a bad
11:02thing. The algorithm is getting smarter
11:03about context. So, something like this
11:05no longer reads as a failure. This is
11:07what creators call good abandonment.
11:09Now, there's one more thing I want to
11:10talk about before we get on to the
11:11expert level of the algorithm, and that
11:13is format saturation. If you've been
11:15using a consistent format with the same
11:17thumbnail style, same title structure,
11:19and same content angle, and recently it
11:20stopped working for you, there's a
11:22reason for that. And the clearest
11:23example out there in the real world
11:25right now is the 100 days in Minecraft
11:27format. We believe a creator called Luke
11:29TheNotable popularized this concept.
11:31Within months, the packaging became
11:33completely saturated. On the left side
11:35of the thumbnail, day one, weak
11:36character. On the right side, day 100,
11:38glowing character. The title, I survived
11:41100 days in Minecraft. And it worked
11:43brilliantly. Thousands of small and
11:45intermediate gaming channels copied it
11:46exactly, and for a brief window, they
11:48grew. Now, imagine you're the YouTube
11:49algorithm, and you have a Minecraft
11:51fan's homepage to build out for them.
11:53Minecraft fans also like other stuff, so
11:55it's not just going to be Minecraft
11:56videos on that feed. It's likely that
11:57the YouTube algorithm is only going to
11:59surface one 100 days video on that home
12:01feed at that time. And we'll just
12:03imagine that it has two to choose from.
12:05A 100 days video from a channel with
12:074,000 subscribers, and a 100 days video
12:10from ForgeLabs, who has 6 million
12:12subscribers. If you're the algorithm,
12:13you don't have to flip a coin here. You
12:15You're going to default to the channel
12:16with the larger data set. YouTube has
12:18years of proof that ForgeLabs keeps
12:21people watching and keeps them
12:22satisfied. The thing is, YouTube doesn't
12:24have just two candidates in that
12:26category to choose from. Go ahead and
12:28search for 100 days in Minecraft right
12:29now. Those thousands of small channels
12:32are nowhere to be found. Buried the
12:34moment the giants decided to start
12:36playing the same game. This
12:37templatization is at a breaking point,
12:39and the lesson isn't that complicated.
12:41If you look identical to a channel that
12:43is 10 times the size of you, the
12:45algorithm will just show the big channel
12:47instead. But there is good news, because
12:49the big channels are locked into their
12:50formats because their audience expects
12:52consistency. But you're not locked into
12:54anything. So, find the angles that the
12:56big channels aren't covering. Adopt a
12:58thumbnail style that stands out against
13:00the clones, and framing that triggers
13:02genuine curiosity instead of format
13:05fatigue. YouTube will reward that
13:07because a refreshing video is exactly
13:09what the satisfaction algorithm is
13:10looking for. So, here's one thing I want
13:12you to do this week. Open Studio, go to
13:14analytics, hit traffic sources, and take
13:16a screenshot of that breakdown. Then,
13:18ask yourself one honest question. Does
13:20what I'm creating actually match where
13:23my viewers are coming from? If browse is
13:25your biggest source, but you haven't
13:26posted consistently in weeks, then you
13:28are burning the most valuable real
13:30estate YouTube could give you. If search
13:31is dominating, but you're not able to
13:33turn those viewers into subscribers,
13:34then what you're doing is building
13:35someone else's library, not your own
13:36channel. But, here's something we still
13:38haven't explained. Why is it that
13:39sometimes a video will do nothing, and
13:41then 3 days later, suddenly explode? And
13:43why does it sometimes feel like tiny
13:45channels are outperforming ones that are
13:4710 times their size, and somehow yours
13:49isn't? The answer to these questions
13:51isn't the word luck. It's engineering,
13:54and that's what the expert section is
13:55for. So far, we've covered CTR,
13:57retention, satisfaction, traffic
13:59sources, and all of that is the
14:01interface layer, what you can see in
14:03YouTube Studio. This section is about
14:04what's running underneath all of it. A
14:07while back, a team of engineers at
14:08Google published a research paper
14:09explaining exactly how YouTube's
14:11recommendation system works at the
14:13machine learning level. It's publicly
14:15available. Most creators have never read
14:17it. We're going to walk through the key
14:19parts of it right now. The system runs
14:20as a two-stage pipeline. Every time
14:22someone opens YouTube, it has less than
14:24a second to decide what to show them,
14:26simultaneously for over 2 billion
14:28people, mind you. So, the first thing it
14:30has to do is make the problem smaller.
14:32Stage one is called candidate
14:34generation. YouTube has over 14 billion
14:36videos. It can't score all of them for
14:38every user in real time. That would be
14:40computationally impossible. So, instead,
14:42it uses a logic called co-visitation.
14:44People who watch video A tend to also
14:46watch video B. That pattern gets used to
14:49rapidly narrow the entire catalog down
14:51to a few hundred candidate videos per
14:53user. Your video doesn't get ranked
14:54until it makes the list. And if you post
14:56inconsistently across a vast array of
14:58unrelated topics, then the system can't
15:00find a reliable group of people to test
15:02you against. Now, for this behavior, you
15:05don't get penalized, but you do become
15:07what these engineers would call a null
15:09candidate. You are invisible to the
15:11pools where your audience actually
15:13lives. Ryan Trahan is a good real-world
15:15example of this. Today, he has over 23
15:17million subscribers, but for a stretch
15:19of his career, his channel was
15:20completely stalling. Not because his
15:22videos were bad, but because the
15:24algorithm was simply confused.
15:26>> it's very evident on my channel that I
15:28have pivoted so many different
15:29directions. I made a video copying Emma
15:31Chamberlain's Instagram photos, and then
15:33I kind of kept easing more into that
15:35like mainstream social media culture
15:37type of content, and then eventually I
15:40got into the commentary world being
15:41inspired by people like Cody Ko.
15:43>> Running vlogs for athletes, Emma
15:44Chamberlain style content, Cody Ko
15:46commentary, three completely different
15:48audiences on one channel. The system
15:50couldn't build a reliable fingerprint
15:52for him. He wasn't being penalized, he
15:54was just invisible. That was until he
15:57committed entirely to one format,
16:00narrative-driven budget and survival
16:02challenges. One audience, one consistent
16:05signal, every video pointing in the same
16:07direction. His channel broke out of its
16:09plateau and never looked back. The same
16:11algorithm, it didn't change, the data he
16:13was feeding it did. The system takes
16:15those few hundred candidates and scores
16:17each one. Watch time, retention, CTR,
16:19probability, and satisfaction signals
16:22all get weighted together into a single
16:24predicted score. But the question it's
16:26really asking is this, if we show this
16:28video to a specific person right now,
16:30will they come back to YouTube tomorrow?
16:32YouTube calls this expected value over
16:35time. They would rather someone close
16:36the app early feeling satisfied and
16:39return the next day, than stay for two
16:41hours, feel drained, and never open it
16:44again. A flat retention curve caused by
16:46padding, where people stay but the
16:47satisfaction survey comes back weak,
16:49will eventually get deprioritized in
16:51favor of content that actually delivers.
16:54The system isn't fooled by watch time
16:55alone. Now, here's the mechanic behind
16:58the question we opened this entire video
17:00with. Why does a video do nothing for
17:02three days and then suddenly explode?
17:04Well, the system has a tension that's
17:06built into it, and YouTube's engineering
17:08team has a name for this, the
17:09explore-exploit problem. Exploit means
17:12the system recommends a video it's
17:14already confident about, proven
17:15performers with strong historical
17:17signals, safe bets. But if YouTube only
17:19ever exploited, new videos would never
17:21surface and the platform would stagnate.
17:23So, it explores, giving a small slice of
17:26traffic to unproven videos as a test.
17:29Every video you upload gets an
17:30expiration window. The system shows it
17:32to a small initial audience and watches
17:34what happens. If the video over performs
17:36relative to what the system predicted,
17:37allocation expands. More people see it.
17:40They respond well, it expands again and
17:42again. That expansion takes time,
17:43sometimes 48 hours, sometimes a week.
17:45But here's what makes this even more
17:47important to understand. The system's
17:49prediction for your video isn't based on
17:51some universal standard. It's based on
17:53your own channel's history. Your last
17:55[music] 10 videos set the baseline
17:57expectation your next video has to beat
17:59to trigger expansion. You can see how
18:01this is playing out right now, once
18:02again, in the Minecraft niche.
18:04CaptainSparklez has 11.4 million
18:06subscribers, a genuine YouTube OG on the
18:08platform since 2010. Forge Labs has
18:11around 6 million, but Forge Labs 100
18:13Days videos routinely pull in millions
18:16of views while CaptainSparklez is
18:18posting to over 11 million subscribers
18:20and getting a fraction of that per
18:21video. The subscriber count doesn't
18:23explain the gap. The baseline does. Most
18:26of CaptainSparklez's subscribers
18:27>> [music]
18:27>> came from a different era of YouTube.
18:29When the algorithm tests his new content
18:31on that subscriber base, the vast
18:33majority of them scroll past it. If they
18:35scroll past enough, it's not going to
18:37regularly submit his content to them at
18:39all. The click-through rate drops
18:41against his own baseline expectation and
18:43the system takes that as a signal to
18:45pull back distribution. A smaller
18:46channel with a highly engaged audience
18:48doesn't have this problem. So, the
18:50baseline is just lower for them. The
18:52over performance signal is easier to
18:54trigger and when it triggers, expansion
18:56happens faster. This is why comparing
18:57your view count to channels that are
18:59larger than you isn't going to tell you
19:01anything of value. They're not your
19:03competition. Your competition is your
19:05own past performance. So, here's one
19:08thing you can do this week that other
19:09creators are not doing. Open Studio, go
19:12[music] to content tab, switch to
19:13advanced mode and pull your last 20
19:16videos. I want you to build a simple
19:17spreadsheet with four columns: title,
19:19CTR, average view duration, and views in
19:22the first 7 days. That average is your
19:24baseline, the exact bar the algorithm
19:26has set for you based on your history.
19:28Now, find your outliers, which would be
19:30the two or three videos that have over
19:32performed that baseline in all four
19:34categories simultaneously. The reason we
19:36can't look at just two of these
19:37categories is simple. Say one of your
19:40videos has like a lot more views than
19:41usual, but a really low click-through
19:42rate. That means that it probably got a
19:44push from somewhere else, maybe a Reddit
19:46post or an email newsletter. But the
19:48videos that over performed across all
19:50four categories at the same time, that's
19:52your signal. Was it the topic, the title
19:54structure, the thumbnail? Was it
19:55something you did differently in the
19:56first 30 seconds? The algorithm has
19:59already told you with this data that it
20:01is willing to bet heavily on that
20:03combination. Your job from here is to
20:05find out what those things were that
20:07worked, and then find a way to repeat
20:09them deliberately. Don't chase the
20:11highest view count video on your
20:13channel. You need to be looking at the
20:14videos that performed the best relative
20:17to your own baseline. That is where the
20:19algorithm has the most confidence in
20:20you, and that is where you should be
20:22going next. But here's the thing,
20:24understanding the algorithm doesn't
20:26actually fix the problem that most
20:27creators are facing today. Because the
20:30frustrating reality of YouTube is that
20:32sometimes videos that are genuinely
20:33good, well-researched, well-optimized
20:35pieces of content still seem to get no
20:39views relatively speaking. And the
20:41reason that could be happening has
20:42nothing to do with what we covered
20:43today. If that sounds familiar [music]
20:45to you, then I encourage you to watch
20:47this video next. You're not going to
20:49want to miss it.

About this transcript

This page contains the full transcript of The Complete Guide to the YouTube Algorithm by vidIQ, generated from the public captions YouTube serves with the video. The transcript has 4,309 words across 662 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.

What you can do with it

Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to choose a caption track available on the video.

Free YouTube transcript tool

YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, caption language selection, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.