The ruler: how many words fit into one minute
The honest answer is a range, not a single number. Comfortable narration lands between 130 and 160 words per minute. Under 130 the video drags and viewers start scrubbing forward. Over 160 the words still arrive but the meaning stops landing. For explanatory content, 140 words per minute is the safe center, and it is the number worth building your entire planning around.
Format moves the ruler before you write a line. Documentary and storytelling sit at 120 to 135 words per minute, because the emotion lives in the pauses and rushing kills it. News, list videos and fast explainers push up to 165, because the viewer arrived expecting density and slow delivery reads as padding. Choose your number by format, not by habit.
Language moves it too, and almost nobody accounts for that. The same idea usually takes 10 to 20 percent more words in Portuguese or Spanish than in English. A 1,400 word English script becomes roughly 1,550 to 1,680 words in Portuguese for the same runtime. When you localize a channel, translate the target duration, never the word count.
That detail is usually filed as a translation chore, and it is actually where the cheapest growth on a channel hides. If a format already works, a second language is not a second production, it is the same production re-timed and re-voiced. FalconVid is built around exactly that: you duplicate a project into another language and pay only the difference instead of producing a new video from scratch, with narration available in 63 languages, and the script is written to the target duration in the new language rather than to the old word count. The 10 to 20 percent swell stops being a manual re-timing job and becomes a setting.
- Comfortable narration range: 130 to 160 words per minute
- Explanatory content: 140 wpm is the safe planning number
- Documentary and storytelling: 120 to 135 wpm, pauses included
- News, lists and fast explainers: up to 165 wpm
- Portuguese and Spanish need 10 to 20 percent more words for the same runtime
- Duplicate a project into another language paying only the difference, narration in 63 languages
The table: words by duration at 140 words per minute
This is the working table. It assumes continuous narration at 140 words per minute, which is the pace of a normal explanatory video that is not racing and not crawling. Stop guessing before you write. If you arrived from a text you already have instead of a blank page, why a prompt is not a script lays out the order that works: text, then structure, then scene.
Read it forward to plan and backward to diagnose. Forward: you want 10 minutes, so you need about 1,400 words. Backward: your last video came out at six minutes and change, so count the file. A 900 word script is 6 minutes and 26 seconds at 140 wpm. It was never a ten minute script, it was a six minute script written hopefully.
Two adjustments follow immediately. If your format sits at 120 wpm, multiply the numbers by 1.17. If it sits at 165 wpm, multiply by 0.85.
- 5 minutes: about 700 words
- 8 minutes: about 1,120 words
- 10 minutes: about 1,400 words
- 15 minutes: about 2,100 words
- 20 minutes: about 2,800 words
- 30 minutes: about 4,200 words
- Slow format at 120 wpm: multiply by 1.17. Fast format at 165 wpm: multiply by 0.85
What steals screen time and is not a word
The table counts speech. A finished video contains a lot of runtime that no word occupies: breathing pauses between sentences, deliberate silence after a strong line, a music sting between blocks, the intro, the outro, and b-roll that plays with no narration over it. None of it appears in your word processor, and all of it appears in the timeline.
In practice this adds 5 to 15 percent on top of the raw calculation. A 1,400 word script that measures exactly 10 minutes of pure speech renders as a 10:30 to 11:30 video. Do the arithmetic in the direction you need: to land safely under 10 minutes, write around 1,270 words. To land safely over, write 1,400 and let the overhead push you.
The critical part is that this error only runs one way. Overhead makes videos longer, never shorter, so a script written to the exact number always overshoots. Decide once whether your target is a ceiling or a floor, then write to that side of it.
Finding your own overhead percentage is the tedious part, because the only honest way to measure it is to render, time the file and rewrite. Done by hand that loop costs an evening every single time, which is why most people do it twice and then stop. It gets much cheaper when rendering is not the expensive step: in FalconVid the AI researcher, scriptwriter, narrator, editor and sound designer work on the same video in parallel instead of in a queue, so a long video is finished in up to 30 minutes, and several run side by side, 2 at a time on Starter and 5 on Pro. Calibrating your real overhead becomes a same day loop instead of a project.
- Breathing pauses and sentence gaps: invisible in the script, real in the timeline
- Deliberate silence after a strong line: seconds each, and they add up
- Music stings between blocks and intro plus outro: usually 20 to 40 seconds total
- B-roll playing with no narration over it: pure runtime, zero words
- Budget 5 to 15 percent on top of the raw word count, and always in the longer direction
- Long video finished in up to 30 minutes, so measuring your real overhead takes a day, not a month
How to spread those words inside the script
Knowing the total is half the job. The other half is distribution, because a video with the right word count and the wrong shape still loses the audience. Give the hook 5 to 8 percent of the total. On a 1,400 word script that is 70 to 112 words, roughly 30 to 48 seconds, and it is the single highest leverage paragraph in the file.
The body works in blocks of 150 to 250 words. One block equals one idea equals one image change. At 200 words per block, a 1,400 word script is about seven blocks, which means at minimum seven visual changes. Blocks over 300 words are where retention curves bend down.
Close with 5 percent, so about 70 words or 30 seconds, covering the payoff and the call to action. Anything longer and the retention graph falls off a cliff before the end card.
- Hook: 5 to 8 percent of the total, 70 to 112 words on a 10 minute script
- Context: enough to frame the promise, never a biography of the topic
- Body blocks: 150 to 250 words each, one idea and one image change per block
- A 1,400 word script equals about 7 blocks and at least 7 visual changes
- Close and CTA: about 5 percent, 70 words, 30 seconds, then stop
When the target duration becomes a channel setting
Everything above is a manual discipline, which means it decays after the tenth video. The alternative is to stop treating length as something you estimate per script and start treating it as a property of the channel. In FalconVid the target duration is a channel setting: you say the videos are 10 minutes, and the system writes to that measure instead of writing blindly and hoping.
The rest follows from the same setting. The script is cut into blocks, media is matched to each block so the picture changes when the idea changes, and narration is generated in any of 63 languages, which is where the 10 to 20 percent word swell between languages stops being your problem. You approve the content calendar, and the videos are produced and published on their own to YouTube, Instagram, TikTok, Rumble and Facebook.
The numbers depend on the quality mode far more than on the plan, so they are worth stating exactly. A 12 minute video costs 1,008 credits in economy mode, 8,676 in balanced and 26,760 in premium. Starter is $47 a month with 15,000 credits: 14 videos if you stay in economy for the whole month, or a realistic working plan of around 10 to 12 a month with most in economy and the one that carries the channel pushed up to balanced. Pro is $97 with 30,000 credits, 29 economy videos and 5 channels, and every creation feature is on every plan, so no capability is locked behind a higher tier: what moves with the plan is volume, channels, concurrent generations, the Senior Analyst and support.
Volume comes from the same place as the consistency. Videos are generated side by side rather than one at a time, 2 concurrent on Starter, 5 on Pro, 10 on Business, 25 on Agency and 50 on Scale, and there is a 7 day guarantee. For a faceless channel publishing on a fixed rhythm, the point is not that a machine writes faster than you. It is that the length stops drifting from video to video while several videos are being made at once.
- Target duration is set once at channel level, not estimated per script
- The system writes to the measure and matches media to each block
- Narration in 63 languages, so the cross language word swell is handled
- You approve the calendar and the videos publish themselves to 5 networks
- 12 minute video: 1,008 credits in economy, 8,676 balanced, 26,760 premium
- Starter $47 for 15,000 credits: about 10 to 12 videos a month mixing modes, 14 in pure economy
- Generations side by side: 2 on Starter, 5 on Pro, 50 on Scale, 7 day guarantee
Why 8 minutes became everyone's target
The 8 minute mark is the most repeated number in YouTube planning and the most often misquoted. Historically, 8 minutes was the threshold where a video could carry more than one ad break. The rule today is stated in terms of the video being long enough to fit them, but the habit stuck, and crossing 8 minutes is what opens mid-roll ad insertion for the upload.
In words, 8 minutes at 140 wpm is 1,120 words. That is exactly the case where you should not write to the exact number, because landing at 7 minutes 52 seconds defeats the entire purpose. Write 1,200 to 1,250 words and let the overhead carry you comfortably past the line, then verify on the rendered file rather than trusting the estimate.
What you should not do is inflate. A video padded from 6 real minutes to 8 hollow ones retains worse, and average view duration is the metric that decides whether the next video gets shown at all. The clean solution is upstream: pick topics with enough substance for 8 to 12 minutes, so the length is a consequence of the material and not a stretch.
Both halves of that get easier when duration is a setting and the topic is not a guess. Telling a FalconVid channel that its videos are 10 minutes puts every upload past the 8 minute line by design instead of by luck, and the Spy reads the channels already monetizing in your niche and extracts the pattern, so you choose subjects that genuinely hold 8 to 12 minutes rather than staring at a blank page. Padding is almost always a symptom of a thin topic, and the topic is decided before a single word gets written.
- 8 minutes at 140 wpm equals 1,120 words of pure narration
- Crossing 8 minutes is what opens mid-roll ad insertion on the upload
- Write 1,200 to 1,250 words so you clear the line with margin
- Verify on the rendered file, never on the estimate alone
- Never pad: a hollow 8 minutes costs more in retention than the ad break returns
- A duration set at channel level clears the 8 minute line by design, and the Spy picks topics that hold it
AI narration changes the math, in both directions
A synthetic voice does not get tired, does not speed up when it is bored and does not slow down in the final third of a long session. That constancy is worth real money in planning: a human reader can miss the estimate by 15 percent across a 20 minute script, while a synthetic voice at a fixed setting lands within a few percent of the calculation almost every time. The ruler becomes trustworthy.
The trap is speed. Nudging playback above 1.1x is tempting because it shaves a minute off without touching the text, and it reliably costs retention on the same script. The words arrive intact and the comprehension does not. If the video is too long, cut words, never milliseconds.
One last calibration. Voices do not all run at the same rate at the same setting, and the difference across languages is larger than most people expect. Generate one full video per language, measure the actual words per minute on the rendered file, and lock that number into your template.
Two things make that calibration stick instead of decaying by video twenty. FalconVid narrates with ultra realistic voices across 63 languages at one fixed pace for the whole channel, so the words per minute you measured on video one still hold on video one hundred. And when a finished file does land long, the Studio is the honest tool for it: you watch the first cut, shorten the intro, drop a block or swap a piece of media, and it re-renders. That is cutting words, which works, instead of cutting milliseconds, which does not.
- Synthetic narration keeps a constant pace, so estimates land within a few percent
- Human readers can drift 15 percent across a 20 minute script
- Playback above 1.1x costs retention even with identical text
- Too long? Cut words, not milliseconds
- Measure the real wpm of your voice per language once, then write to that number
- One fixed pace per channel, and the Studio to drop a block instead of speeding playback