The “40% Rule” for Video Scripts: Why Page-Timed Drafts Run Long on Camera

Creators love runtime formulas because they feel concrete. Pick a target length, multiply by an average speaking rate, and your script looks finished.

For script timing for video, that shortcut is useful right up until you perform the thing.

A draft can look perfectly timed on the page and still run long on camera. You pause to breathe. You repeat a phrase because the first version felt flat. You add a bridge like “so,” “actually,” or “the thing is.” You slow down on a product name, a contrast, or the line that carries your verdict. None of that feels dramatic while recording. All of it takes time.

A quick clarification before we go further: the “40% rule” here is a mnemonic, not a measured average. It is not a documented benchmark that every script expands by 40%, and it is not a claim that you should always cut exactly 40%. It is a memorable way to think structurally when your takes repeatedly overrun. The point is to stop treating timing problems as tiny word-level issues when they are often beat-level issues.

Common industry guides do give useful starting ranges. Swarmify’s guide to video scripts suggests roughly 125 to 150 words per minute for finished video, Mindstamp’s corporate and training script guide uses about 150 spoken words per minute as a planning estimate, and Vidyard’s script timer lets you estimate duration by slow, average, or fast delivery speed rather than pretending there is one fixed rate. That is helpful planning guidance. It is still only a baseline.

Independent speech research points in the same direction. In a classic review of speech tempo, linguist N. J. Enfield summarizes findings showing that reading aloud tends to run faster than spontaneous speech, with read speech often around 150–190 words per minute and spontaneous speech often closer to 110–180 words per minute, depending on task and speaker (Enfield, “The Anatomy of Meaning: Speech, Gesture, and Composite Utterances” summary of tempo findings). The exact numbers vary by study and situation, but the practical takeaway is stable: performed, explanatory, semi-spontaneous delivery often expands beyond page-based estimates.

That is why a script is not truly timed when it reads well. It is timed when it speaks well.

If you want a quick baseline before you do the real out-loud pass, the Script Word Counter is useful for first-pass estimation.

Why page estimates fail once a script becomes performance

Silent reading compresses effort. Your eyes can absorb a dense sentence almost instantly because comprehension is not the same as speaking.

Delivery is slower because spoken language has to carry extra jobs in real time:

• breathing
• emphasis
• listener processing time
• transitions between ideas
• natural signposting
• slight resets when a line feels stiff aloud

That gap is not just a creator anecdote. It lines up with both scripting advice and speech research.

Swarmify’s guide recommends writing for the ear, not the eye, with short sentences and read-aloud rewrites. Captions’ short-form scripting guide says to read the script aloud with a timer before filming because expected and actual runtime often differ, especially early on. Those are practical creator rules. The research angle helps explain why: read or tightly prepared text and spontaneous explanatory speech do not behave identically.

That matters even more in creator-led formats like tutorials, reviews, explainers, and commentary. These are rarely delivered like neutral narration. They are usually extemporaneous performances built on prepared text. You are staying on message, but you are also sounding human. That means extra pauses, restarts, softer qualifiers, and emphasis beats are likely to appear even if you never “go off script.”

There is also a second problem: not every sentence costs the same amount of spoken time.

On paper, two paragraphs can look equally manageable. On camera, one may be cheap and one may be expensive. Product comparisons slow you down. Caveats invite extra wording. Technical claims often need a simpler second phrasing. The word count may stay close, but the spoken load grows.

So when creators say, “The script was the right length,” what they often mean is, “The script matched an average WPM estimate.” That is not the same as saying the performed delivery fit the slot.

What the “40% rule” actually means

Used well, the 40% rule is a big-edit mindset.

It does not mean every draft needs a literal 40% cut. It means that when you repeatedly run long, the fix is often more structural than cosmetic. You may need to cut whole arguments, examples, or setup beats rather than shaving adjectives and hoping for the best.

Most creators respond to overruns in one of four ways:

1. speak faster
2. trim a few filler words
3. plan to “fix it in the edit”
4. keep everything and rush the ending

Sometimes that works for a tiny overage. But if the real issue is that your 3-minute slot contains 4 minutes of spoken thinking, word-level cleanup will not save it.

The better question is:

Which ideas deserve spoken time, and which ideas only deserve page space?

That shift is the heart of the rule.

Spoken time is a budget. Every claim, example, caveat, transition, joke, and summary line is competing for it. Once you start thinking that way, you stop protecting every sentence equally.

And usually the video gets better. The strongest points have breathing room. The transitions stop sounding panicked. The conclusion gets to land cleanly instead of being squeezed into the final seconds.

Worked example: Mara’s “3-minute” script that became 4:12

Let’s make this concrete.

Mara is a fictional tech reviewer with a fixed 3-minute slot in a weekly roundup. She drafts 390 words, using a common rule of thumb: roughly 130 words per minute puts her near 3 minutes.

That looks reasonable on paper.

On camera, her take lands at 4:12.

Nothing unusual happened. She did normal creator things:

• paused before her verdict
• repeated a contrast for emphasis
• softened a harsh opinion
• added small bridges like “actually” and “so”
• restated one technical point more plainly
• slowed down during the buyer recommendation

In other words, she performed the script like a person.

Here is the rough original shape:

Hook: This budget microphone looks like a bargain, but I think most people are comparing it to the wrong alternative.

Point 1: The build quality is better than expected, especially the stand and front controls.

Point 2: The sound is clearly better than laptop audio, but the top end gets brittle in untreated rooms.

Point 3: The software is simple, though setup has two annoyances first-time users should know.

Point 4: The real question is not whether it is good, but whether it is better than waiting for the next tier up.

Conclusion: If you need a clear upgrade from a headset mic, it makes sense. If you want something you will still enjoy a year from now, I would wait.

On paper, this seems concise. On camera, each beat expands. “The build quality is better than expected” becomes an example plus a reaction. “The top end gets brittle” becomes an explanation plus a disclaimer so the judgment does not feel too blunt.

That is the real issue: page density hides performance cost.

Mara’s paper edit table

This is where the article usually needs to get more practical, so here is the actual edit logic in table form.

That table shows the key principle: she does not save time by deleting random words. She saves time by deciding which beats get screen time at all.

Here is a more developed before-and-after comparison.

BEFORE
This budget microphone looks like a bargain, but I think most people are comparing it to the wrong alternative. The build quality is honestly better than I expected, especially the stand and the front controls, and that gives a really strong first impression. The sound is definitely better than laptop audio, and I don't want to undersell that, but in a more reflective room the top end can start to sound a little brittle, which I think some people are going to notice more over time. The software is pretty simple, although there are two setup annoyances first-time users should know about. And really the question isn't whether it's good, because I think for the price it mostly is, it's whether it's better than waiting and moving up one tier.

AFTER
This mic feels more premium than it sounds.
The build beats expectations.
It easily improves on laptop audio, but reflective rooms make the top end harsh.
So the real decision isn't “is it good?” It's “buy now or wait one tier up?”
Buy it if you need the upgrade now.
Wait if you want the setup you'll still like longer term.

The revised version is not just shorter. It is easier to perform cleanly.

How to edit for spoken time instead of written density

1. Mark breath units, not sentences

A punctuation mark is not a timing mark.

Print the script and divide it into breath units: chunks you can say naturally without flattening the emphasis or grabbing air halfway through. One long sentence often contains two or three spoken beats.

For example:

The mic sounds clear enough for calls and quick voiceovers, but the high end gets sharp in a reflective room, which means it can impress you on first listen and annoy you after longer use.

Spoken naturally, it behaves more like this:

The mic sounds clear enough for calls and quick voiceovers.
But the high end gets sharp in a reflective room.
So it can impress you on first listen, then annoy you later.

Same idea. More time.

2. Separate intentional rhythm from bloat

Not every pause is a problem.

Keep:

• a short beat before the verdict
• one repeated contrast for clarity
• slower delivery on a key phrase
• a clean transition between sections

Cut or reduce:

• saying the same judgment three ways
• throat-clearing phrases like “and yeah” or “so basically”
• excessive hedging around obvious points
• multiple examples proving the same thing

This protects you from over-correcting into a flat, compressed read.

3. Cut by beat first, line second

If the script runs long, remove a whole beat before you polish wording.

In Mara’s case, the software-setup section was not bad. It just was not essential to the buyer decision inside a 3-minute slot. Cutting that beat returned more usable time than trimming filler from every paragraph.

4. Rebuild around must-keep beats

Ask four questions:

1. What is the viewer’s decision?
2. What is the one useful surprise?
3. What warning must stay?
4. What verdict should they remember?

Everything else becomes optional.

If you want to see the spoken beats against visuals before you shoot, the Storyboard Builder is handy for laying out dialogue, action, and duration together.

How much buffer should you plan?

There is no honest universal percentage. But there is a practical method.

Start with the industry estimate as a baseline, then draft to a lower initial runtime than your actual target, do a full-energy performed read with a timer, and reserve explicit seconds for:

• pauses before key points
• one example per section
• transitions between segments
• your closing line or CTA

A simple approach for a 3-minute video is:

• draft the first pass to feel a little short on paper
• perform it aloud standing up, at camera energy
• note where you naturally expand
• leave room for that expansion instead of pretending it will disappear

That method is more reliable than forcing every script to some fixed “safe” percentage.

The opposite problem: when cuts make the script too short or too tight

Overrun is common, but the opposite mistake matters too.

Sometimes you cut aggressively and the script suddenly fits the clock but sounds unnatural. The rhythm disappears. The transitions feel abrupt. The conclusion lands before the audience has processed the last idea.

That usually means you removed structure-supporting rhythm, not just bloat.

Signs you cut too far:

• the script sounds like bullet points, not speech
• contrast lines feel jammed together
• there is no pause before the verdict
• examples disappear and abstract claims stack up
• the ending feels technically on time but emotionally early

The fix is not to re-add all the old wording. Re-add only the pieces that help the listener track the thought:

• one signpost transition
• one clarifying example
• one pause before the main conclusion
• one cleaner phrasing of a dense line

Think of this as restoring intentional rhythm without bringing back the full weight of the original draft.

Quick checklist for script timing for video

Before you record, ask:

• Am I using WPM as a rough estimate or as a promise?
• Have I done a full spoken read with a timer?
• Which lines expand every time I perform them?
• Where does one sentence really contain two or three breath units?
• Which example does the most work?
• Which whole beat could vanish without hurting clarity?
• Does my conclusion still have room to breathe?
• If I naturally add 10 to 20 seconds while talking, does the script still fit?

Pitfalls to avoid

A few traps show up over and over.

Mistaking the rule for data. The 40% idea is a mnemonic for structural-cut thinking, not a measured average.

Overstating WPM ranges. Use them as common planning baselines, not guarantees. Tools like Vidyard’s script timer are useful partly because they assume different delivery speeds exist.

Only trimming the intro. Openings do matter, but many overruns grow in the middle where every section gains an extra explanation.

Cutting all personality. A script can fit the slot and still sound dead.

Trusting silent reading. If you have not performed it aloud, you do not really know the runtime yet.

Conclusion: design for spoken time, not page comfort

If your scripts keep running long on camera, the problem is usually not that you cannot count words. It is that the page hides the real cost of speech.

Use word-count estimates as a starting point. Then do the real work: mark breath units, perform the draft aloud, separate rhythm from bloat, and cut whole arguments when the slot demands it.

For your next video, try this simple workflow:

1. Draft to a slightly short paper runtime.
2. Perform a timed read at real presentation energy.
3. Mark the lines that expand.
4. Cut one complete beat before you start micro-editing.
5. Rehearse the revised version once more.

If the new draft suddenly sounds calmer and lands on time, you found the issue. The missing time was not hiding in single words. It was hiding in structure.