Key Takeaways
Divide your word count by 140 and you have a workable answer: roughly 1,400 words fills a ten-minute slot. Most people deliver somewhere between 100 and 180 words a minute, and 130 to 150 is a sensible planning figure for a prepared talk. That figure gives you reading time, not speaking time – pauses, laughter, slide changes and questions all sit outside it. Nerves will speed you up on the day. Unscripted interruptions will stretch you out. Write short of your slot, then read it aloud with a timer.
You have a script and you have a slot. The question is whether one fits inside the other, and you would rather find out now than four minutes from the end with three pages left. The arithmetic that answers it is simple enough to do on the back of the running order.
What the arithmetic won’t tell you is the part that actually causes overruns. Nobody overruns because they miscounted their words by two hundred. They overrun because a question came from the floor at minute six, or because the clicker stopped working, or because they added something on the spot that felt good and cost ninety seconds. So: the sum first, then the three things that make the sum wrong.
Start with the division
Take the number of words in your script. Divide it by the number of words you speak in a minute. That is the whole calculation.
The second number is the one people guess at. Delivery speed usually falls somewhere between roughly 100 and 180 words a minute, which is a wide enough band to matter: the same script can run half again as long at one end as at the other. Slow, deliberate, formal delivery – a eulogy, a reading, an address where you want weight on each clause – sits near the bottom of that range. Conversational presenting, the kind where you’re talking to a room rather than declaiming at it, sits near the top.
For a prepared talk delivered to an audience, 130 to 150 words a minute is a reasonable planning figure. Split the difference at 140 and the numbers come out easy to remember:
- Five minutes: around 700 words
- Ten minutes: around 1,400 words
- Twenty minutes: around 2,800 words
- Forty-five minutes: around 6,300 words
If you don’t know how many words you’re holding, paste the script into a word counter and read the number off. Then, rather than doing the division yourself and re-doing it every time you cut a paragraph, put the text into a speech time calculator and move the rate up and down to see the spread. Seeing the fast figure and the slow figure side by side is more useful than any single answer, because your real speed on the day is somewhere in there and you don’t yet know where.
Reading time is not speaking time
Word count divided by words per minute gives you reading time. Delivery time is longer, and the gap is not small.
Everything that isn’t a word sits outside the calculation. The breath before a difficult sentence. The pause you leave after the line you want to land. Laughter, if you’re lucky, and the wait for it to die down. The two seconds a slide takes to advance and the second after that while the room reads it. Walking to the other side of the stage. Drinking water. Taking a question, answering it, and finding your place again on the page.
None of that is in the word count, and all of it is in the slot. A script that calculates to nine minutes and thirty seconds does not fit a ten-minute slot; it fits a ten-minute slot only if you say every word at your planned rate and nothing else happens at all. Treat the calculated figure as the floor of what the talk will cost you, never the ceiling.
This is also why the calculation is worth doing anyway. It is a cheap filter. If your script comes out at eighteen minutes for a twelve-minute slot, you don’t need to rehearse to know you’re in trouble – you need to cut, and you know roughly how much. The calculation tells you whether rehearsing is even worth your time yet.
Nerves speed you up; interruptions stretch you out
Two forces pull on the clock, and they pull in opposite directions.
The first is you. Nervousness makes people talk faster. A talk rehearsed alone in a quiet room at a pace that felt comfortable is commonly delivered faster than that on the day, in front of people, with adrenaline running. Which means the risk from nerves is not that you overrun – it is that you finish early, and finish early because you rushed, dropping the pauses that made the thing land in rehearsal. If you come off stage three minutes under, that’s usually what happened.
The second force is everything you didn’t write down. An aside that occurs to you mid-paragraph. A question taken in the flow rather than held to the end. A technical problem that turns thirty seconds into two minutes while you wait for a slide to appear. A host who introduces you at length and starts your clock late. This is the direction overruns actually come from. Slots are blown by the unscripted, not by the script.
Put those together and the picture is odd but consistent: your words will probably take less time than you planned, and your talk will probably take more. You cannot control the second force much. You can build room for it.
Write short of the slot
A script written to exactly fill the slot has no margin, and a talk with no margin cannot absorb anything at all. One question, one dead clicker, one thought you couldn’t resist, and you are over – not by a little, but by however long the interruption lasted, because there was nowhere for it to go.
So plan for a shorter delivery than the slot you’ve been given. How much shorter depends on how exposed you are. A recorded voiceover with no audience and no questions can run close to the mark, because nothing unpredictable is going to happen inside it. A conference talk with a live audience, a slide deck and a chair who may or may not enforce the clock needs real slack. A wedding speech needs slack for laughter, which you cannot time and should not talk over.
The margin is not wasted time. It gets spent – on the pauses, on the question you’d rather take properly than wave away, on the story you tell better than the version on the page. And if it doesn’t get spent, you finish a couple of minutes early, which no audience in history has resented. Running short is a rounding error. Running long is the thing the next speaker remembers you for.
Read it aloud with a timer
Every number above is an estimate of your speaking rate. The only way to stop estimating is to stand up, read the script at the pace you intend to use, and time it.
Do it out loud, at full volume, with the pauses in. Reading it in your head is worthless – inner speech is much faster than the mouth, and you will come out minutes under. Start a timer, read, and note the result. Then do it again, because the second run is usually faster than the first and closer to what will happen on the day. Two or three runs also give you your personal words-per-minute figure, which you can reuse for every talk you write from now on: divide your word count by the minutes it actually took, and stop guessing where you sit in the 100-to-180 range.
If you’re not working from a script – if the talk lives in bullet points and you improvise the sentences – the calculation has nothing to chew on. Record yourself talking it through and turn the recording into text; SozAI is a transcription app that will do that from a recording or an audio file, and the transcript gives you a word count and a written version of what you actually say rather than what you meant to. Then you can count and cut like anyone else.
What the arithmetic can’t tell you
A calculated figure is a planning tool, not a measurement, and it fails in predictable places.
It doesn’t know what your words are. A script dense with names, numbers, foreign terms or technical vocabulary is slower per word than plain narrative, because you slow down to get them right and the audience needs a beat to absorb them. It doesn’t know your material. A funny talk runs long because of laughter; a solemn one runs long because of silence. It doesn’t know the room – a large hall with a slow-returning echo forces you to speak more deliberately than a meeting room does, and that costs minutes over a long talk.
Synthetic speech has the same blind spot. Having a script read back to you by a machine is a decent way to hear whether the sentences work when spoken, and it will give you a length, but it isn’t your length. It doesn’t breathe where you breathe, hesitate where you hesitate or wait for a room. Use it to check the writing, not the clock.
The honest position is that no calculation substitutes for reading the thing aloud with a timer running. What the calculation buys you is the decision about whether that’s worth doing yet – whether you’re roughly in range, or three pages over and cutting before you rehearse at all. That’s a real thing to know, and it takes about ten seconds. It just isn’t the answer.
