Script Length, Runtime & Word-Count Math

Start with the arithmetic, because most length decisions get made by feel and feel is how people end up with a video that doesn't fit anything. Budget fifteen hundred to sixteen hundred words for every ten minutes of finished video. Call it seven hundred and fifty words per five minutes. Your voice speed shifts it a little either way, but that's the ratio you plan around.

Long-form is the revenue engine of the whole channel. In practice that's fifteen hundred to thirty-five hundred words, landing you between ten and twenty-five minutes. Shorts are a different product entirely: thirty to fifty-nine seconds, nine by sixteen. Long-form is sixteen by nine, three minutes as an absolute floor, eight or more if you want it to be worth the effort.

Now the trap almost everybody falls into once. When you ask ChatGPT for a script, ask for a word count. Never a character count. Two thousand characters sounds enormous and is about three hundred words, which is ninety seconds of video. If you want fifteen minutes, you're asking for twenty-two hundred to twenty-eight hundred words, and more than that is fine.

What isn't fine is forcing a model to hit a number it doesn't have material for. Tell it to make the script longer and it won't go and find more story. It pads. It restates the point it made two paragraphs ago, adds throat-clearing in front of everything, stretches sentences that were already the right length. Padding is precisely where retention falls off a cliff. If the script came back short, the fix is more research, not more instructions.

So how long should the video be? Partly the platform decides. YouTube pushes anything over thirty minutes onto TV screens, because television placement carries bigger ads and YouTube would rather be selling those. Very long videos blow up on TV. Roughly ten-minute videos do better on mobile. Around the twenty-minute mark performs best right now, because YouTube is competing with Netflix for session length. One channel running twenty to twenty-seven minute videos made thirty-one thousand dollars in its first thirty days.

But every niche has its own sweet spot and you find it by testing. Twenty to thirty minutes worked on one history channel. A forty-five minute video on the same channel didn't get pushed anywhere near as hard. The shortcut is your competitors. Set your word count against the length of the videos already winning in your niche instead of picking a number you like the look of.

Documentary-style work is its own case. Twelve to fifteen genuinely good minutes, and if the topic is bigger than that, split it across several videos rather than stretching one. Partly that's attention. Partly it's safety, because the longer a video runs the more surface area there is for something inside it to trip a policy problem, and multi-hour videos are much harder to keep strike-free.

On a brand new channel, go shorter than your instinct says. Five or six minutes, not twenty or thirty. You're trying to get to roughly fifteen videos fast, because that's the point where the data means something and the niche has either proved itself or hasn't. Length is a decision you earn after that.

When you do extend, extend in steps. Test slightly longer than what's already working rather than jumping straight to an hour. One clean way to add runtime: take two short scripts covering the same story and merge them into a single thirty or forty minute video. Your retention percentage will drop. That's fine. Percentage isn't what pays you, average view duration is, and as long as that clears the threshold the revenue holds up.

The reverse works too if you're solo and posting once or twice a day. Cut a twenty-minute piece into two ten-minute videos and write a fresh intro onto part two.

Count the words before you write. Not after.