How to Grow Your YouTube Channel
Two numbers decide almost everything, and neither of them is the one on your channel page.
Two numbers decide whether YouTube shows your video to people who have never heard of you: the click-through rate on the thumbnail, and how long viewers stay once they click.
YouTube offers a video to a small audience, measures both, and either widens the offer or stops. Subscriber count is not an input to that decision.
So growing a channel means making videos that get clicked and then watched, repeatedly, on a subject narrow enough that the people who liked one will like the next.
Everything else — posting schedules, keyword stuffing, asking for likes — operates at the margins of those two numbers.
The part that is counterintuitive is that a narrow subject grows faster than a broad one, and it is worth understanding why rather than taking it on faith.
YouTube's recommendation works by association. It learns which videos are watched by the same people, and it uses that to decide what to put next to what. A channel where every video appeals to the same audience builds a dense cluster of associations: watching one of your videos becomes strong evidence for recommending another.
A channel covering five unrelated subjects builds five weak clusters instead. The viewer who came for one subject is not a signal for the others, so each video starts from close to nothing.
The practical form of this is uncomfortable for most new creators, because it means your tenth video on the same narrow topic will outperform your first video on a new one, and the tenth feels repetitive to make.
It also explains the common pattern of a channel that gets one video with a million views and never repeats it. The video found an audience the rest of the channel does not serve, so the audience arrived, watched one thing, and had no reason to stay. A viral video on a channel with no cluster behind it converts almost nobody.
The thumbnail and title are the product
Nothing gets watched that does not get clicked, and the click happens before anybody has seen a frame of your video.
YouTube Studio reports impressions click-through rate per video. The useful comparison is against your own channel's average rather than any published benchmark, because the figure varies enormously by subject and by where the impression appeared.
A thumbnail works when it poses a question the title answers, or the other way round. A thumbnail that duplicates the title wastes half the space.
And test on a phone at the size it will actually appear. Most thumbnails are designed on a large screen and viewed at the size of a postage stamp, which is why so many of them are unreadable where it counts.
Watch time, and the first thirty seconds
The audience retention graph in Studio shows exactly where people leave, and almost all of the drop is in the opening.
The fix is structural rather than cosmetic: start at the thing the title promised. Introductions, channel branding and asking for subscriptions before any content has been delivered are the most reliable way to lose a third of the audience in twenty seconds.
For a longer video, the retention graph also shows the mid-video dips, and those are usually a section that could be cut entirely.
Watch time in absolute minutes matters as well as percentage, which is why a well-retained ten-minute video outperforms an equally well-retained three-minute one for the same subject.
What subscribers are actually for
A subscriber is not a guaranteed viewer. YouTube shows new videos to a fraction of a channel's subscribers and expands based on how they respond.
What the subscriber count does is give YouTube a reliable first audience to test each new video on. A channel with a thousand engaged subscribers gets a faster, cleaner signal than one with ten thousand who do not watch.
That makes the ratio more useful than the total. Views per subscriber on recent uploads tells you whether the audience is real.
It also makes the count matter to viewers rather than to the system: a channel with a visible following is one a stranger is more willing to give ten minutes to, and that is a conversion effect rather than an algorithmic one.
Shorts, and what they do and do not feed
Shorts are a separate recommendation surface with their own feed, and growth on one does not automatically transfer to the other.
A Shorts audience subscribing does not reliably watch long videos, so a channel that grows on Shorts can end up with a large subscriber count and weak long-form performance — which looks like the algorithm turning against you and is not.
Shorts are useful for being discovered and poor for being understood. The sensible use is as an entry point that points at a specific long video, not as a parallel channel.
Shorts watch time does not count toward the monetization threshold in the same way, which is the other reason to treat the two as different products.
What wastes the first year
Buying into a posting schedule you cannot sustain. Consistency helps because it produces more attempts, not because YouTube rewards punctuality.
Optimizing tags. They have been close to irrelevant for years; the title, the thumbnail and the spoken content carry the meaning.
Chasing trends outside your subject, which produces exactly the orphaned viral video described above.
And rebranding the channel repeatedly. Every change resets the association cluster YouTube has built, and the cluster is the asset.
Questions people ask about growing a channel
How many videos before a channel takes off?
There is no number. What matters is whether consecutive videos serve the same audience, because that is what builds the associations YouTube recommends from.
Does posting more often help?
Only through volume of attempts. YouTube does not reward a schedule, and a rushed video that retains badly costs more than a skipped week.
Do tags still matter on YouTube?
Barely. Title, thumbnail, spoken content and the first lines of the description carry nearly all the meaning the system extracts.
Should I make Shorts to grow?
They are a separate surface with their own audience. They are good for discovery and poor at converting to long-form watch time, so treat them as an entry point rather than the plan.
Why did my viral video not bring subscribers?
Because the audience it found is not served by the rest of the channel. A hit on a channel with no cluster behind it has nothing to convert viewers into.
Is click-through rate or watch time more important?
Neither alone. A video has to be clicked before retention can be measured, and high retention on a video nobody clicks never gets tested.
Keep reading
- subscribers on a channel — A visible following is what makes a stranger give an unknown channel ten minutes, and subscribers on a channel addresses that half of the problem.
- where views come from on a new channel — The distribution surfaces are fewer than people assume, and where views come from on a new channel names each.
- why Shorts behave differently — The two surfaces are separate products, and why Shorts behave differently covers the one that confuses most channels.