Nearly every first draft script that lands in our production queue is too long. Not wildly too long, which would be easy to catch, but 15% to 25% over, which is worse, because it survives the brief, survives the table read in the writer's head, and only fails on camera. The question of how many words in a 30 second ad usually gets asked after the shoot, when the edit comes back at 38 seconds, instead of before the brief goes out, when the fix costs nothing.
An overweight script leaves you two options on the day, and both are bad. The creator speeds up, and the read flattens: pauses disappear, the hook loses its beat, and the whole ad starts to sound like the terms and conditions at the end of a radio spot. Or the editor cuts around it afterwards, and you get jump-cut mush where sentences end on a different breath than they started. Counting words before anyone hits record is cheaper than either.
The numbers in this guide are Spark's measured production standards, taken from timing our own voiceover sessions and on-camera UGC shoots week after week: VO reads average 3.2 words per second, on-camera delivery runs 2.8 to 3.0 words per second, and every handover between speakers in a multi-speaker script costs about 1.2 seconds. Everything below is built from those three rates.
How many words fit in a 30-second ad?
Here is the full reference, for both delivery modes, across the five durations paid social actually uses. The first two columns are the mathematical ceilings. The last column is what we put in the brief.
| Ad length | VO read (3.2 words/sec) | On-camera UGC (2.8 to 3.0 words/sec) | What we brief |
|---|---|---|---|
| 6 seconds | Up to 19 words | 17 to 18 words | 14 words |
| 15 seconds | Up to 48 words | 42 to 45 words | 40 words |
| 30 seconds | Up to 96 words | 84 to 90 words | 80 words |
| 45 seconds | Up to 144 words | 126 to 135 words | 120 words |
| 60 seconds | Up to 192 words | 168 to 180 words | 160 words |
The ceilings assume wall-to-wall speech from the first frame to the last, and no good ad has that. A product close-up, a reaction beat and an end card all eat runtime without a word of dialogue. So the brief column sits at roughly 90% of the on-camera ceiling with about 2 seconds held back for the close. Write to that column, and the read comes back with room to breathe.
Multi-speaker scripts need one more subtraction. Each handover between speakers costs about 1.2 seconds: the glance, the turn, the half-beat before the next person picks up. A 30-second dialogue with four handovers loses nearly 5 seconds of talking time, which is 14 words at on-camera pace. Podcast-style and duo scripts feel roomy on the page and come back long for exactly this reason.
Key takeaway
For a 30-second ad, brief 80 words for on-camera delivery, or up to 90 for a voiceover. The mathematical ceiling is 96 words at a 3.2 words per second VO read, but the last few of those words should be silence: a read that uses every second has no room for the pause that sells.
Why voiceover and on-camera scripts run at different speeds
A VO artist works in a booth with a controlled breath, a pop shield and the option to punch in any line again. Nothing competes with the read, so it can sit at 3.2 words per second and stay clean. An on-camera creator is doing three jobs at once: delivering the line, managing their face and handling the product. Their natural pace lands at 2.8 to 3.0 words per second, and pushing them past it is what produces that glassy, over-rehearsed delivery the comments section clocks instantly.
Both of our rates run hotter than the textbook numbers, and the gap is deliberate. Conversational English averages about 150 words per minute, per the speech rate research collected by VirtualSpeech, and traditional commercial VO guides such as Abe's Audio plan at about 2.5 words per second, or 60 to 75 words for a 30-second spot. Those norms were built for broadcast, where the ad had the living room's attention and could afford air. A feed ad is in a swipe contest. The direct response reads that win on Meta and TikTok are compressed and front-loaded, and 3.2 words per second is where our VO sessions consistently land once the read sounds native to the feed.
There is a limit, though. Push much past 3.5 words per second and the read outruns the listener. The same speech research puts comfortable listening comprehension well below that for new information, and an ad is nothing but new information the viewer has never heard: a claim and the mechanism behind it. The viewer does not rewind. A claim they missed is a claim you never made.
A script that fits on the page and a script that fits in the mouth are two different documents.
Why rushed reads fail
Two things break when a creator sprints an overweight script, and the first is the hook. The opening 2 to 3 seconds decide whether anyone watches the rest, and that is precisely where a nervous creator makes up time. A hook written as eight words gets delivered as a blur, the beat after it vanishes, and the ad loses the scroll-stop it was built around. We see it in the numbers: the rushed takes of the same script reliably post weaker hook rates than the paced takes, because the viewer's ear needs that half-second of landing room to decide the video is worth staying for.
The second casualty is comprehension. A viewer processing a sprinted claim about your product is doing more work than a feed asks of anyone. They do not slow the video down. They swipe. This is also why duration guidance from the platforms themselves rewards restraint: Meta's video ad guidance pushes 15 seconds or shorter for feed and Reels placements, and TikTok for Business recommends 21 to 34 seconds for in-feed ads. The winning durations are short, which means the word budgets are small, which means the discipline has to live in the script, not the read.
Longer formats still have a job. A 45 or 60-second script earns its runtime when the audience does not yet know they have the problem, and we map which lengths belong to which buyer in our script format guide by awareness stage. But a 60-second ad written at 200 words is a choice. A 30-second ad written at 95 words is a mistake, and it is the single most common one we correct.
Worked example: trimming a 95-word draft to a true 30-second read
Here is a first draft in the shape we see weekly. Skincare brand, on-camera creator, briefed as a 30-second ad. It reads fine on the page:
The draft (95 words): "I tried pretty much every vitamin C serum out there, and my dark spots always came back within a few weeks, which was really frustrating. Then my facialist told me the reason most serums do nothing is that the vitamin C oxidises before it ever gets into your skin. This one is stabilised, so it actually stays active the whole time. I started noticing a real difference after about three weeks, and now people keep commenting on my skin. If you are still struggling with dark spots, you seriously need to give this a try."
Run the maths before anyone shoots it. At on-camera pace, 95 words is 32 to 34 seconds of continuous speech, before a single product close-up or end card. Shot as written, this is a 36-second ad, or a rushed 30. So we cut:
The 30-second cut (79 words): "I have tried every vitamin C serum out there, and my dark spots always came back within a few weeks. My facialist finally told me why: in most serums, the vitamin C oxidises before it ever gets into your skin. This one is stabilised, so it stays active from day one. I noticed a real difference after three weeks, and now people comment on my skin. If your dark spots keep coming back, this is the one to try."
That is 26 to 28 seconds of speech, leaving 2 to 3 seconds for the product beat and end card. The cuts followed the same order they always do:
- Qualifiers went first. "Pretty much", "actually", "about", "seriously", "keep". Each one costs a third of a second and adds nothing the delivery does not already carry.
- Stated emotion went second. "Which was really frustrating" is four words describing a feeling the creator's face shows for free. On camera, emotion is footage, not copy.
- Throat-clearing got compressed. "The reason most serums do nothing is that" became "why:". Eight words of wind-up into one word and a pause, and the pause does more work.
- The structure survived untouched. Hook, mechanism, proof, CTA, in that order. Every element in our UGC ad script structure template is still there. Only the padding died.
One cutting rule we hold to: never claw back time from the pause before the CTA. That beat is where the viewer decides. Cut the middle of the script, not its ending.
Reverse lookup: you have the words, here is the duration
Most scripts are written before a duration is chosen, so run the lookup in the other direction. Count your trimmed draft, then read its real length off this table.
| Your script | VO read time | On-camera read time |
|---|---|---|
| 50 words | About 16 seconds | 17 to 18 seconds |
| 75 words | About 23 seconds | 25 to 27 seconds |
| 95 words | About 30 seconds | 32 to 34 seconds |
| 120 words | About 38 seconds | 40 to 43 seconds |
| 150 words | About 47 seconds | 50 to 54 seconds |
| 200 words | About 63 seconds | 67 to 71 seconds |
Then decide, honestly. If the trimmed script is 120 words, you are holding a 40-second ad: either cut it to a duration that performs, or commit to the longer format and structure it to earn the time. What you should not do is sprint it into 30. And the trap runs the other way too: 50 words stretched across 30 seconds pads as badly as 120 words crammed into it. Slow reads signal scripted; the feed punishes both directions.
The word budget belongs in the brief itself, next to aspect ratio and length, in the specs box of our UGC brief template. Leave it out and the creator sets the pace for you, and you find out what they chose in the edit. This pacing check is a standing step in how we run production: every script gets a word count against its briefed duration before it goes anywhere near a shoot, and the rest of our scriptwriting references live in the resources hub.
FAQ: ad script word counts
How many words fit in a 30-second ad?
A 30-second ad holds up to about 96 words as a voiceover read at 3.2 words per second, or 84 to 90 words delivered on camera at 2.8 to 3.0 words per second. In practice, brief around 80 words for on-camera delivery so the read keeps its pauses and the final 2 seconds stay free for the end card or product close.
How many words is a 60-second commercial script?
A 60-second commercial holds up to about 192 words as a voiceover read at 3.2 words per second, or 168 to 180 words on camera at 2.8 to 3.0 words per second. Brief around 160 words so the script leaves room for product shots, an end card and at least one deliberate pause.
How many words per minute is a voiceover read?
Conversational English averages about 150 words per minute, and traditional commercial voiceover guidance plans at roughly 2.5 words per second. Paid social direct response reads run faster: across Spark's measured production output, VO reads average 3.2 words per second, about 192 words per minute.
How long should a UGC ad script be?
TikTok recommends 21 to 34 seconds for in-feed ads and Meta's guidance favours 15 seconds or shorter in feed placements, so most UGC scripts should land between 40 and 100 spoken words at on-camera pace. Multi-speaker scripts lose about 1.2 seconds per handover, so subtract that talking time before counting words.