Chapter 1 · Lesson 2

How YouTube Actually Works

The algorithm in plain English: CTR, watch time, retention, session time, and the recommendation loop that decides who sees your video.
The YouTube Recommendation Loop Every video runs this loop — successful ones complete it many times. Viewer opens YouTube Impression thumbnail + title shown Click? CTR signal Watch + Retention how long they stay Algorithm Score CTR × retention × session More impressions to similar viewers (the recommendation loop) Session time = what they watched next on YouTube
The YouTube recommendation loop: CTR earns the click, retention holds the viewer, and a strong algorithm score triggers more impressions.

How the algorithm "thinks"

You don't need to understand machine learning to make sense of YouTube. The algorithm has one job: show each viewer the next video they're most likely to watch and enjoy. That's it. Every metric you'll hear about — CTR, retention, session time — is just a signal YouTube uses to predict whether your video will satisfy a particular viewer.

Forget the idea of a secret formula or "hacks." Think of YouTube as a matchmaker between videos and viewers. Your job as a creator is to make videos that match real viewer intent — and the metrics simply tell you how well you're doing it.

What is an impression?

An impression is one time your thumbnail and title appear in front of someone on YouTube — on the homepage, in suggested videos, in search results, or in notifications. Impressions are how viewers discover you exist. No impressions = no clicks = no views.

Crucially, impressions are decided by the algorithm based on who YouTube thinks the right audience is. Early on, you'll get a small "test pool" of impressions. What you do with them decides everything.

CTR — Click-Through Rate

CTR = clicks ÷ impressions. If your video shows up to 1,000 people and 60 of them click, your CTR is 6%. CTR is mostly a verdict on two things: your thumbnail and your title. The video itself doesn't matter yet — they haven't seen it.

What's a "good" CTR?

For most videos, anything between 4% and 10% is healthy. Below 3% usually means a packaging problem (thumbnail or title). Above 12% can sometimes mean clickbait that hurts retention later. CTR is best read in context — compare a video to your channel average, not to a stranger's.

Tip
Don't obsess over absolute CTR. Compare your CTR on similar topics over time. If it drops, your packaging slipped. If it rises, you're learning.

Watch time and audience retention

Watch time is the total minutes viewers spent on your video. Retention is the percentage of the video the average viewer watched. They sound similar but say different things.

  • A 10-minute video with 70% average retention = 7 minutes of watch time per viewer.
  • A 30-minute video with 30% retention = 9 minutes of watch time per viewer.

The second video produced more watch time even though its retention was worse. YouTube notices both numbers. Retention tells YouTube whether the video delivered. Watch time tells YouTube how valuable the experience was overall.

The "average view duration" you'll see in Studio

YouTube Studio shows you Average View Duration (in seconds/minutes) and Average Percentage Viewed (as a %). The percentage is the cleaner signal early in a channel's life. Aim for 40%+ on long-form videos and 60%+ on Shorts as a working baseline.

Session time — the underrated metric

Session time is how long a viewer stays on YouTube after watching your video. If they leave the platform right after, your session contribution is low. If they go on to watch three more videos (yours or anyone else's), you've made YouTube a happier place — and the algorithm rewards videos that do this.

This is why "end-of-video CTAs that lead to more YouTube content" (end screens, next video suggestions, playlists) outperform "subscribe to my newsletter" CTAs from an algorithmic standpoint.

The recommendation loop

Here's what actually happens when you publish a video:

  1. YouTube shows your video to a small test audience (a few hundred impressions).
  2. If CTR + retention are healthy, YouTube expands the test audience.
  3. If signals stay strong, YouTube starts recommending it more broadly — homepage, suggested videos, browse traffic.
  4. This compounds. The bigger the audience, the bigger the next push, and so on.
  5. Eventually the topic exhausts or competing content gets prioritized, and the curve slows down.

This is why some videos "explode" weeks or months after publishing. The loop didn't catch fire immediately — it caught fire after enough viewers signaled "this is worth recommending."

Heads up
The first 24–48 hours after publishing matter, but they aren't everything. A "dead" video can come back to life if it's evergreen and well-packaged. Don't delete videos that flopped on day one.

Why some channels grow fast

Fast-growing channels almost always do three things at once:

  • Tight packaging (high CTR): thumbnails and titles that match a clear, curious viewer intent.
  • Sharp scripting and editing (high retention): hooks in the first 5 seconds, no padding, constant payoff.
  • Topic clusters (high session time): each video naturally leads to the next, keeping viewers on the channel.

It's rarely "one viral video." It's a system that produces healthy signals video after video, allowing the algorithm to trust you.

Key takeaway
YouTube rewards videos that keep viewers watching and bring them back. CTR gets the click, retention holds them, session time grows the channel, and the loop does the rest.

Diagnosing a video at a glance

When a video underperforms, you can usually pinpoint the issue using these three metrics:

  • Low impressions? Topic/SEO mismatch — YouTube doesn't know who to show it to.
  • High impressions, low CTR? Thumbnail/title problem — the packaging isn't earning clicks.
  • Decent CTR, low retention? Script or pacing problem — viewers click but don't stay.

We'll dive deeper into this diagnostic system in Module 11.

Common misconceptions

Common pitfalls
  • "More views are always better." 10,000 views with 15% retention is worse than 3,000 views with 60%. The algorithm sees both.
  • "Subscribers will save my channel." Subscribers help, but the algorithm primarily ranks individual videos on their own signals. A video that fails will fail even if you have 100k subs.
  • "Tags fix discoverability." Tags are a minor signal. Your title, thumbnail, and the first 30 seconds of your script drive impressions and retention.
  • "Posting time is the secret." Posting time is a minor factor. A great video posted at 3 AM still wins; a mediocre video posted at the "perfect" time still flops.
Lesson Summary
YouTube's algorithm is simpler than it looks: show each viewer the next video they'll enjoy. Every metric is just a signal toward that goal. CTR earns the click, retention holds the viewer, watch time and session time tell YouTube your video is worth recommending — and the loop amplifies what works. Master the signals, and you stop chasing the algorithm and start working with it.