YouTube Tag Generator

YouTube tags are the least visible field on the upload page and the one people spend the most anxious effort on. They are metadata: hints about what the video is, sitting alongside the title, description, and - far more importantly - what viewers actually do when the video is recommended to them.

This generator builds a tag set mechanically. Enter your keywords, one per line or comma-separated, and it returns every keyword on its own plus every possible pairing of two keywords joined into a phrase. Alongside the list it tracks the total character count against YouTube’s 500-character allowance across all tags combined.

The mechanical part is the value and the risk. Combination generation covers phrasings you would not have typed, and it also produces combinations that are not phrases anybody would search. Reading the output critically is most of the work.

The output is every single keyword plus every unordered pair, which means the count grows quadratically. Three keywords give six tags. Five give fifteen. Eight give thirty-six, and ten give fifty-five. The character budget is 500 across everything, so the practical ceiling arrives well before you run out of ideas - somewhere around six to eight keywords for typical phrase lengths.

This is why the character counter matters more than the tag counter. Fifty-five tags averaging fourteen characters each is 770 characters before separators, comfortably over the limit, and the tool will tell you how far over rather than silently truncating. When you are over, removing one keyword removes itself plus every pair it appears in, which is a much larger reduction than removing one tag.

The corollary is that keyword choice should be ruthless before generation, not after. Four well-chosen keywords producing ten tight tags is a better set than eight loose keywords producing thirty-six that need pruning down to fifteen. Start narrow.

Every pair is generated in one order - first keyword, then second, as they appear in your list - so the phrases you get depend on the order you typed. "kettlebell workout" and "workout kettlebell" are different tags, and you will only get the one your input order produces. Reordering your keyword list and regenerating is a legitimate way to get the phrasing you actually wanted.

More than that, many pairs are simply not things people say. From "beginner", "sourdough", and "24 hour", the pair "beginner 24 hour" is a phrase nobody searches, while "beginner sourdough" and "24 hour sourdough" both are. Copying the whole set and deleting the nonsense phrases takes a minute and is the difference between metadata and noise.

It is worth being calm about the stakes. Tags are a minor ranking and discovery signal on modern YouTube - the title, thumbnail, first lines of the description, and viewer retention do the heavy lifting. Well-chosen tags help with misspellings and niche search phrasings; a perfect tag set does not rescue a video nobody clicks.

Your input is split on newlines and commas, each fragment is trimmed, empties are dropped, and the survivors are deduplicated - so typing the same keyword on two lines yields one keyword. Order is preserved from your input.

Generation then runs in two stages. Every keyword is added to the tag set as-is, then the list is walked in nested fashion adding each keyword joined to each later keyword with a single space. Because the inner loop starts after the outer index, pairs are produced once each in input order rather than twice in both orders, and no keyword is paired with itself.

The final list is joined with a comma and a space to produce the copyable string, and that joined string’s length is what is measured against the 500-character allowance - so the separators count toward your budget, as they do on YouTube. Everything runs in the browser as you type; there is no model, no keyword research lookup, and no check on search volume, competition, or whether anyone has ever searched a generated phrase.