How to write a YouTube script that fits the time

What a video script is for

A script for a video is not a film script. It is a short list of sentences in a fixed order, so that you know what you are going to say before you press record, and so that you can cut the video to a length without deciding what to leave out while talking.

The order that works is the same one the free template on this site uses, and the same one our own measurement supports. The hook is the first sentence: why keep watching. The setup is one or two sentences of context or a promise. The main points are the value, one idea per line. The call to action is what to do next: follow, comment, watch part two.

Writing the hook first is the part people get wrong. If the hook cannot be written on its own, the video does not have one yet.

• Hook: the first sentence, why keep watching. One sentence, no preamble.
• Setup: one or two sentences of context, or a promise about what the video answers.
• Main points: the value, one idea per line, in the order you want them heard.
• Call to action: what to do next, in one sentence.

How to write one in five steps

Here is the order we worked in, and the numbers each step produced for our own 76 word script.

1. Decide the length first, in seconds. The template offers 15, 30, 45 and 60 second targets and longer ones, and each comes with a word count: 38, 77, 116 and 155 words at the measured pace of 155 words per minute.
2. Write the hook on its own. Ours was 11 words and the tool timed it at 4.3 seconds.
3. Write the setup and the main points. Ours were 14 and 26 words, timed at 5.4 and 10.1 seconds.
4. Write the call to action last. Ours was 25 words, timed at 9.7 seconds.
5. Paste all four parts into the free template and read the total. Ours came out at 0:29 for 76 words, which is the number to compare against your target length.

Source: our own run on October 25, 2026 in headless Chromium against the live template, with our four parts typed into the four fields and the one minute target selected. The word counts are ours; the per part timings are the tool's, at its 155 words per minute.

The word count is a guide, not a measurement

We recorded our own script with a local text-to-speech model and compared the length of the recording with the tool's estimate. This is the number to take seriously and the number to be careful with.

Source: our own measurement on October 25, 2026. Method: the script was written by us, the timings came from the live template in headless Chromium, and the recording was made with a local text-to-speech model at its default speed and measured with ffprobe 6.1.1. The recording and the log are shipped with this page.

Read it the right way round. The word count model did not fail: 76 words divided by 155 words per minute is 29.42 seconds, and the tool reported 0:29. The model simply does not know how fast your particular voice reads. Our synthetic voice read at 213 words per minute. For a person reading normally, expect the estimate to be close; for a fast or clipped delivery, expect the real recording to be shorter than the estimate.

The numbers to write against

The template measures the pace of 21 videos and publishes the words that fit a given length at three speeds: the slowest quarter of them, the median and the fastest quarter. Here is the table it prints.

Source: the live video script template page, which publishes these figures from its own measurement of 21 videos (read October 25, 2026). The same page reports that the first sentence of 10 short videos took a median of 3.0 seconds and 9 words, and that 4 of those 10 opened with a question; it also reports 161 words per minute for short explainers against 125 for briefings, so the pace depends on what kind of video you are making.

Use the slowest column to be safe on a video with a hard length, and the middle column for anything else. If your script is over the number, cut words rather than speed up: the model gives the same advice, that it tells you how many words to cut when you are over your target.

What we did not test

We measured one script and one recording. Here is what that does not cover.

We did not test a person reading. Our recording was synthetic speech from a local text-to-speech model, and it read faster than the model in the template assumes. The comparison tells you how far the estimate can be off, not what your own delivery will be.

We did not test a filmed video. We measured the audio length of the script, not whether a video with that script holds attention, and nobody watched it.

We did not test the 155 words per minute figure itself. It is the site's published measurement of 21 videos, we did not repeat it.

We did not test shorts, live streams or videos with a sponsor read in the middle, and we did not test other writing structures than the four parts above.

What to do next

The situation you are in decides the next step.

• Your script is over the time limit: cut words from the main points first, and keep the hook as it is.
• Your script is under the time limit and the video feels thin: add one more main point rather than padding the hook.
• You do not know how long your delivery takes: record the hook on your phone and time it, then use that as your own baseline.
• The video is for a sponsor and the read has to fit: write the read as its own part and time it separately.
• You want the file to be timed for you: the free template times each part as you type, and keeps the draft in your browser only.