YouTube Banner Size in 2026: The 500-Pixel Inset That Decides What Anyone Actually Sees

Ask what size a YouTube banner should be and you will get 2560 × 1440 in about four words, from roughly forty sites, all of which stop there. It is a true answer and close to a useless one, because the file you upload is not the image anybody sees. The banner is the only asset on YouTube that gets cropped four different ways depending on the screen it lands on, and the crop is severe: a phone throws away more than eighty per cent of what you uploaded.

That is why so many channels have a banner that looked correct in the upload dialog and arrives with the channel name sliced through the middle on mobile. It is also why the safe-zone advice is such a mess — search for it and you will find two different numbers presented as rival facts, 1235 × 338 in some places and 1546 × 423 in others, with nobody explaining why both exist.

This is that question worked out properly: what YouTube actually publishes, why the two safe areas are the same safe area, the four crops in pixels, the text size the geometry demands, and — the part almost nobody writes about — what a channel page is for now that it is a conversion surface rather than a shop window.

The numbers YouTube actually publishes

Start with the source rather than the template packs. YouTube's channel branding documentation specifies the banner in terms of a minimum, not an ideal, and it names the safe area explicitly.

Spec	Value	What it means in practice
Minimum upload size	2048 × 1152 px	The floor. Anything smaller is rejected or upscaled badly on a television.
Aspect ratio	16:9	Same shape as a video frame — and misleading, as the next section explains.
Minimum safe area	1235 × 338 px	The centred region guaranteed to survive on every device, at that minimum upload size.
Maximum file size	6 MB	Generous for a still image; only reached by uncompressed PNG exports.
Accepted formats	JPG, PNG, BMP, non-animated GIF	JPG at high quality is the sane default. Animated GIFs are not accepted.
Profile picture	800 × 800 px, square	Rendered as a circle, usually at under 100 px across. More on this below.

The number missing from that table is 2560 × 1440, which is what every guide, template pack and design tool preset uses. It is not a contradiction. 2560 is exactly 1.25 × 2048, so the recommended size is the stated minimum scaled up by a quarter — same shape, same proportions, more resolution to hand a 4K television. Uploading at 2560 × 1440 is right. Believing it is a different specification from the one YouTube documents is what causes the safe-zone confusion.

The two safe areas are one safe area

Here is the arithmetic that resolves the argument. Take the documented safe area of 1235 × 338 inside the documented minimum of 2048 × 1152, and express it as proportions of the frame:

• Width: 1235 ÷ 2048 = 60.3% of the image width
• Height: 338 ÷ 1152 = 29.3% of the image height

Now apply those same proportions to a 2560 × 1440 upload: 60.3% of 2560 is 1544, and 29.3% of 1440 is 422. Which is, to within a rounding error, the 1546 × 423 that the other half of the internet quotes.

Upload size	Safe area	Share of width	Share of height	Safe-area ratio
2048 × 1152 (documented minimum)	1235 × 338	60.3%	29.3%	3.65:1
2560 × 1440 (what everyone uploads)	1546 × 423	60.4%	29.4%	3.65:1

One rectangle, two scales. If you are designing at 2560 × 1440 — and you should be — the number you need is 1546 × 423, and the older figure is simply the same box measured on a smaller canvas. Any guide presenting them as a disagreement has copied both from somewhere else without dividing one by the other.

The four crops, and the 500-pixel rule

YouTube does not scale your banner down for smaller screens. It crops it — takes a centred slice and discards the rest — which is why the damage is invisible until it happens. Four surfaces, four slices out of the same file:

Surface	Region shown	Discarded each side	Discarded top and bottom	Share of your pixels
Television	2560 × 1440	0 px	0 px	100%
Desktop	2560 × 423	0 px	508 px	29.4%
Tablet	1855 × 423	352 px	508 px	21.3%
Mobile	1546 × 423	507 px	508 px	17.7%

Read the last two columns of the mobile row together, because they contain the most useful fact in this article and it is not stated anywhere else. The phone crop discards 507 pixels from each side and 508 pixels from the top and bottom. Those are the same number.

Which means the safe area is not a mysterious proportion to be looked up. It is your 2560 × 1440 canvas inset by 500 pixels on all four sides. Draw a 500-pixel margin round the whole frame in your editor, keep every piece of text and every logo inside what remains, and you have satisfied the safe zone exactly, without a template, on any tool that can show you a guide.

The other figure worth holding onto is that final column. A phone shows 17.7% of the image you uploaded — under a fifth. Most of the effort that goes into a banner is spent on pixels that only a television will ever display.

You are not designing a 16:9 image

The upload is 16:9 and everything about that ratio is a trap. The region that is always visible has a ratio of 1546:423, which is 3.65:1 — closer to a cinema letterbox than to a video frame, and closer still to a website header. That changes the design problem in three concrete ways.

Vertical composition is unavailable. Anything stacked — a logo above a tagline above a schedule — needs vertical room the strip does not have. On a phone the visible band is 423 pixels tall out of 1440. Two lines of readable text is the ceiling, and one line plus a mark is the comfortable case.

The corners are decoration. Every corner of your banner sits outside the safe area on all three non-TV surfaces. Signature graphics, social icons and secondary logos placed there are, for the great majority of viewers, not there at all.

Photographic backgrounds fight the crop. A photograph composed for 16:9 has its subject somewhere in the frame; the mobile crop takes a horizontal band through the middle and usually finds a torso, a wall or a sky. If you want a photographic background, treat the visible strip as the composition and everything above and below it as bleed — the same discipline that governs a website hero image.

The practical version: build the banner at 1546 × 423, get it working there, then place that finished strip in the centre of a 2560 × 1440 canvas and extend the background outwards to fill it. Design inside-out and the crop can never surprise you. If you need to move between the two canvases quickly, the image resizer has the 2560 × 1440 preset.

How big the text has to be

Banner text obeys the same physics as thumbnail text: it is glanceable reading, not reading, and it has to survive being rendered at a fraction of its design size. The arithmetic is straightforward once you know what the phone does with the strip.

A phone shows the 1546-pixel-wide crop across the width of the screen — call it 390 CSS pixels on a typical handset. That is a scale factor of 390 ÷ 1546, or about 0.25. Your banner is displayed at roughly a quarter of the size you are designing it at, and the visible band ends up about 107 CSS pixels tall — on a screen around 65 mm wide, a strip about 18 mm deep in the viewer's hand.

Cap height in your 2560-wide file	Share of the 423 px band	Rendered on a phone	Physical height
40 px	9.5%	10 px	~1.7 mm
60 px	14%	15 px	~2.6 mm
90 px	21%	23 px	~3.9 mm
120 px	28%	30 px	~5.1 mm
160 px	38%	40 px	~6.9 mm

Nielsen Norman Group's work on glanceable reading — words encountered in a microsession rather than read deliberately — tested letter heights of 3 mm and 4 mm, and found the larger condition measurably easier. Map that onto the table and the threshold lands between the 90-pixel and 120-pixel rows. Text under about 90 pixels of cap height in a 2560-wide banner is below the size at which a phone viewer can take it in at a glance.

At 90 to 120 pixels of cap height, with sensible margins inside a 1546-pixel-wide strip, you have room for roughly twelve to sixteen characters on a line in a heavy sans, where a capital's advance width is close to its cap height. That is two or three words, occasionally four. The same conclusion the thumbnail font arithmetic reaches, for the same reason: the rectangle is small, and every word you refuse to cut takes size away from the words you kept.

Which quietly disqualifies most of what creators put in banners. An upload schedule, a tagline, three social handles and a business email cannot coexist at a readable size in a strip 18 mm deep. They can coexist at an unreadable one, which is what most channels have done.

The asset that actually gets seen is the other one

Here is the reframe that should change how you allocate an afternoon of design work.

Your banner appears in exactly one place: the top of your channel page, seen by people who have already decided to visit it. Your profile picture appears next to every video in search results and suggested feeds, in the subscriptions feed, on every comment you leave, on every community post, next to your name in live chat, and on the channel page itself. One is a destination asset. The other travels with everything you publish, to people who have not chosen to look at you at all.

For nearly every channel, the profile picture is seen orders of magnitude more often than the banner — and it is usually the asset that received five minutes and a cropped selfie.

The specification is 800 × 800 square, rendered as a circle at well under 100 pixels across in most surfaces. That is the design constraint, and it is brutal: a circle roughly the size of a lower-case letter on this page. What survives it is a single high-contrast shape — one face cropped tight to the head, or one letterform, or one silhouette. What dies is a full-body photo, a wordmark with more than about three characters, a busy background, or a thin outline.

What a channel page is actually for

The old mental model of a channel page is a shop window: somewhere to look impressive. The useful model is narrower. A channel page is where a viewer who has just watched one video decides whether you are worth a second one, and where a viewer who arrived from a comment or a mention decides whether you are worth any at all.

YouTube counts that traffic under Channel pages in the Reach tab, and there is no published benchmark for what share of views should come from it — impressions and traffic-source splits never leave your own Studio, which is why every "average" figure you have seen for this was invented. Check your own Reach tab, and check the subscription source report alongside it, since the channel page shows up there too. The subscriber mechanics post covers how to read that report properly.

What matters for design is the sequence a first-time visitor moves through, which is the same on every channel: banner, name and handle, description line, then the first shelf of videos. Four surfaces, in that order, and the whole decision is usually made before any scrolling happens. The banner's job in that sequence is not decoration and not branding in the abstract — it is to answer what is this channel about in the two seconds before the eye moves down to the videos.

Which gives you a test for banner content. Three patterns pass it:

Pattern	What it contains	Works when
Positioning line	Three or four words naming the subject and the angle	The channel name is not self-explanatory
Proof	One credential, result or recognisable artefact	The niche is crowded and trust is the barrier
Identity	A face or mark at scale, with no text at all	The channel name and avatar already carry the subject

And a list of what fails, all for the same reason — it consumes the strip without answering the question:

• Upload schedules. "New videos every Tuesday" is a promise that ages into a liability the first week you miss, and it does not help anyone decide to watch now.
• Social handles and email addresses. Illegible at 25% scale, and there are dedicated link fields directly beneath the banner that render as tappable links.
• Subscriber milestones. Dated the moment the number moves, and invisible at the size they are usually set.
• Collages of thumbnails. The videos are already on the page, at a size where they can actually be read.
• Anything in the corners. Covered above: for most viewers, it does not exist.

The rest of the header, in order of what it does

Beneath the banner sit four elements that do more work than the image above them, and that most creators set once and never revisit.

Your handle is now the channel's address

The @-handle has replaced the old custom URL as the canonical way to link to a channel, and it now appears wherever your channel does — mentions in comments, community posts, live chat, Shorts. It is one of the few things on the page that is genuinely permanent in practice: YouTube caps changes at twice in a fourteen-day window, and once an old handle is released, someone else can take it. Treat it like a domain name rather than a nickname, and make it match the channel name closely enough that someone hearing one can guess the other.

The description is read one line at a time

The channel description is truncated on the page — a first line, then a link to see the rest, which almost nobody clicks. So the first sentence is the whole of the description for practical purposes. Write it as a standalone claim about what the channel does and for whom, not as the opening of a paragraph that resolves later.

Links render as a row, so order them

The links you add in customisation appear directly under the header, with the first ones visible and the remainder collapsed. That means the ordering is a real decision: whatever you most want a new visitor to do goes first, and the rest can queue behind it.

The tabs split your work whether you like it or not

Channel pages separate long-form videos, Shorts and live content into their own tabs. If your Shorts and your long-form serve different audiences, that separation is doing you a favour — a viewer who arrived from a Short can find the long-form work without wading through fifty vertical clips, and the reverse also holds.

Below the fold: trailer, featured video, and twelve shelves

The Home tab is the part of the channel page you can genuinely programme, and YouTube's layout documentation sets out what is available: a channel trailer shown to people who are not subscribed, a separate featured video for returning subscribers, a personalised For You shelf, and up to twelve custom sections you order yourself.

Three things about that arrangement are worth knowing before you spend a weekend on a trailer.

First, the trailer and the featured video are different slots for different audiences, and they can hold different videos. The trailer is an introduction for strangers; the featured video is what a subscriber who came back sees. Filling both with the same upload wastes one of them.

Second, the trailer is shown once. Once a viewer has watched it, they will not be shown it again — so it is a one-shot introduction rather than a permanent billboard, and it should be short enough that a stranger will finish it.

Third, and least appreciated: the default section order is not a recommendation. A new page defaults to shelves like Shorts, uploads, playlists and public subscriptions in a fixed order, and for many channels that puts the weakest evidence at the top. The shelf order is the one part of the channel page that behaves like editorial judgement — your strongest playlist, or the series that best explains the channel, belongs above your most recent upload if the most recent upload is not your best argument.

The television, where the whole banner finally appears

There is one surface that shows your banner in full, and it is the fastest-growing one. Nielsen's monthly distributor measure has had YouTube in first place among all US television distributors for a run of consecutive months, at 12.5% of television watch time in January 2026, up from 10.8% a year earlier.

On a television the channel page is browsed with a remote, at a distance, and the banner occupies the whole frame rather than a strip. This is the one context where the pixels outside the safe area do work: they set the tone of the page before anything else loads. It is also the context where a banner that is a phone-sized logo floating in a field of flat colour looks like an unfinished web page.

The resolution is the inside-out method from earlier, taken seriously. Design the 1546 × 423 strip for phones, then treat the surrounding region as a considered background for the living room rather than empty margin — a gradient, a texture, a continuation of the photograph. Two designs, one file, no conflict between them, because the crop boundary is a hard line you already know the position of.

A twenty-minute channel page audit

Everything above collapses into a checklist you can run on your own channel this afternoon.

1. Open your channel page on a phone. Not the desktop preview — an actual phone. Note what is cut.
2. Measure your banner's largest text. If its cap height is under 90 pixels in a 2560-wide file, it is decorative rather than readable.
3. Delete anything in the banner that is not positioning, proof or identity. Schedules, emails and social handles come out.
4. Draw a 500-pixel margin on all four sides of the canvas and confirm every important element sits inside it.
5. Export your profile picture at 96 × 96 and look at it beside five competitors. One idea, or redesign it.
6. Read your channel description's first sentence alone. Does it say what the channel is and who it is for?
7. Reorder your links so the one that matters is first.
8. Check whether your trailer and your featured video are the same video. If so, change one.
9. Reorder the Home shelves so your best argument sits above your most recent upload.
10. Open the Reach tab, find Channel pages in the traffic sources, and note the number so you have a before to compare against.

Most of that is half an hour of work on assets that have been sitting untouched since the channel was created, which is the usual case: the banner gets made once, in the first week, when the creator has the least idea what the channel is about.

None of it will move views on its own. The channel page is a conversion surface, not a distribution one — it improves what happens to the traffic you already have rather than creating more of it. The rectangle that decides whether the traffic exists at all is still the thumbnail, and the arithmetic there is harsher: a phone renders it at about the size of a postage stamp, as the thumbnail size guide works out. If you are producing those one at a time in an editor, Thumblore exists to remove that bottleneck, because the creators who test three thumbnails per video are almost never the ones hand-building each variant.

But the channel page is the cheapest fixed asset you own. It is designed once, it is seen by every person who takes an interest in you after a video ends, and unlike almost everything else on the platform, no algorithm decides whether it gets shown. Getting it right costs an afternoon and then keeps working, which is a better return than most of the things a creator is told to do weekly.