How much text belongs on a thumbnail

Start from the size it is actually shown at

You design at 1280 or 1920 pixels wide. Your viewer sees roughly 360 pixels wide on a phone home feed, about 210 in a desktop search result, and about 168 in the suggested column beside a video they are already watching.

Take that smallest number seriously, because the suggested column is where a very large share of impressions happen. At 168 pixels wide, your 1280-pixel canvas has been reduced to about 13 % of its design size. A headline set at 100 pixels tall in your editor arrives as 13 pixels tall on screen.

The threshold that matters

Below roughly 10 to 12 pixels of cap height, text on a screen stops being read and starts being recognised as a shape. Readers do not decode it; they see a grey bar and move on. That is the line you are designing against.

Working backwards: to clear 12 pixels at the 168-wide size, your type needs to be about 90 pixels tall on a 1280-wide canvas — roughly one eighth of the height of the whole image. That is far bigger than most people's instinct, which is why so many thumbnails fail.

Which gives you the word count

Type that big can fit about three or four short words across the frame in one or two lines. That is the entire budget. Not a guideline — a consequence of the arithmetic above.

So the discipline is not "write less". It is "decide which three words carry the whole idea". A video titled I tried living on two euros a day for a month becomes 2 € A DAY on the thumbnail, and the title carries the rest. The thumbnail and the title are one message split across two surfaces, not two copies of the same message.

The title does the work the thumbnail cannot

This is the part people miss when they cram a sentence onto the image. The title is right there, in full, in readable body text, immediately below or beside the thumbnail. It has no size limit and it is always shown.

Duplicating the title inside the thumbnail wastes the most valuable surface you have. Use the thumbnail for the hook, the number, the emotional payload — and let the title supply the specifics.

Write the two together, not one after the other

The thumbnail and the title are shown at the same moment, in the same block, and the viewer takes them in as a single unit. So the useful question is not what should the thumbnail say, but what should each half say so that the two together say everything.

A workable division of labour: the thumbnail carries the part that is concrete and instant — a number, an object, a face, a contrast. The title carries the part that needs words to be precise — the qualifier, the place, the condition, the promise. 2 € A DAY on the image, I tried living on two euros a day in Madrid for a month in the title. Neither half repeats the other, and neither half is complete on its own.

The check is quick. Read the pair the way a viewer would, then ask what the thumbnail added that the title had not already given you. If the answer is nothing, those words are costing you your largest piece of visual real estate for no return. Note too that titles are often cut short in narrow layouts, so the front of the title is the part you can rely on being read — another reason for the two halves to carry different information.

When you genuinely need more words

Occasionally the concept needs a contrast: before / after, cheap / expensive, 2020 / 2026. Two or three words on each side of a divided frame can work, because each cluster is small and each has its own space and its own visual anchor.

What does not work is a full sentence flowing across the frame. If you find yourself needing a comma, you have already lost.

A process for cutting a phrase down to three words

Start by writing the sentence you actually want to say, at whatever length it comes out. Trying to be brief from the first keystroke tends to produce a phrase that is vague rather than one that is short, and vague is worse than long.

Then delete in this order. Articles, prepositions and auxiliary verbs go first — at a glance they carry no meaning. The subject goes next, if the subject is you, because the viewer can already see that it is you. Adjectives go after that, unless the adjective is the point. What tends to survive is a number, a proper noun, and one strong verb or noun.

Read what is left. If it is still more than four short words, ask which single word is the hook, and whether anything else in the line could be shown instead of written — a currency symbol, an arrow, an object held up in the photograph. At this size images are cheaper than words, and every word you convert into an image buys height for the ones that remain.

Finally, set the survivors at full size on the canvas and look at them small. A phrase that reads well in your head sometimes turns out to need a comma or a second line before it makes sense, and that is the signal to go back and cut again rather than to shrink the type.

Test it instead of guessing

The reliable check takes five seconds: shrink your own design until it is about the width of your thumb on your phone, then look at it without leaning in. If you cannot read the words at a glance, no viewer scrolling past will either.

That is exactly what the live preview panel in this editor does — it renders your design continuously at the real sizes YouTube uses, and the readability score flags the specific problem when your type is too small or your words are too many.