Sound off video ads are a genuine condition to design for, and the famous statistic behind that advice is a myth. It traces to a trade article from 2016 reporting what four publishers said about their own analytics.
The behaviour is real. The number attached to it does not survive being checked, and a better one exists.
This is for you if you have quoted that figure in a deck and would rather quote something defensible.
Where the famous number came from
A Digiday article published on 17 May 2016, which is where this myth begins.
That piece reported what four publishers said about their own analytics. No methodology was described. Sample size does not appear. Facebook data is not involved anywhere in it.
Read the original and it gets weaker still. PopSugar, quoted in the same article, gave a range of 50 to 80 percent rather than a single figure. So the source does not even agree with itself.
Then consider the age. It described video behaviour on a platform that has rebuilt its video experience repeatedly since, before Reels existed at all. A behavioural claim from that period is not evidence about anything running today.
Why the sound off video ads myth persisted this long
Because it was useful, memorable and nobody had a reason to check.
The advice attached to it was correct. Caption your videos, design them to work in silence. That advice is sound, so the number supporting it never came under pressure.
A statistic that leads people to do the right thing gets repeated without scrutiny, which is exactly the condition under which unsourced figures survive for a decade.
It also had the shape of a real finding. A specific percentage, about a named platform, attributed to publisher analytics. Nothing about the presentation suggests a blog post reporting four anecdotes.
That is the general lesson worth taking. Numbers survive on usefulness rather than on provenance, and the ones supporting good advice are the least likely to be questioned.
The sourced figure to use instead
Verizon Media and Publicis Media surveyed 5,616 US consumers in July 2019.
They found 80% of consumers more likely to watch an entire video when captions are available. Separately, 69% reported watching sound-off in public and 25% in private.
Disclosed sample size, stated method, named organisations, dated. Everything the older claim lacks.
It also supports a more useful conclusion. The older figure told you people watch in silence. This one tells you captions increase completion, which is a direct argument for doing something rather than a description of a condition.
Use this one. It is stronger evidence and it points at the same action.
What sound off video ads actually need in production
Four things, and they are production decisions rather than editing flourishes.
Burned-in captions. Not platform captioning, which depends on settings you do not control, transcription accuracy you cannot guarantee, and availability that varies by placement.
The argument visible in the first two seconds, in text. Somebody deciding whether to keep watching is deciding without audio, so the reason to stay has to be readable.
Visual demonstration rather than spoken explanation wherever possible. If the point can be shown, show it, because showing survives silence and telling does not.
And a soundtrack that adds rather than carries. Music and voice should improve the experience for people with sound on, and nothing essential should live only there.
Sound off video ads aside, captions are an accessibility requirement
This argument stands entirely independently of viewing behaviour, and it is worth making.
Anybody with a hearing impairment needs the text. Somebody watching in a second language benefits from it. A person on a train or in an open office gets the same benefit as anybody choosing silence deliberately.
So even if every statistic on silent viewing evaporated tomorrow, captioning would remain correct.
That is a comfortable position to argue from. You are not relying on a contested behavioural claim, you are doing something defensible on its own terms that happens to also improve completion.
It also removes the awkwardness of having built a practice on a number that turned out to be a myth.
How much text captions on sound off video ads should carry
Enough to follow the argument, not a full transcript.
Dense captions compete with the image for the same attention, and a video where the text is doing everything might as well have been a static.
The aim is comprehension of the point. Somebody should be able to understand what you are offering and why it matters, without hearing a word, in the time they were going to watch anyway.
Keep them short, keep them high contrast, and keep them positioned where platform interface elements will not cover them. That last one gets missed constantly, and a caption obscured by a progress bar is the same as no caption.
Safe areas differ by placement, so a caption sitting comfortably in a feed can be hidden behind interface elements in a story. Designing sound off video ads for the tightest safe area you run in is simpler than producing several versions.
What this episode says about creative statistics
One more pattern worth recognising, because this is not an isolated case.
The same shape produced the hook rate benchmark, the frequency threshold, the creator content premium and the static against video comparison. A question with commercial demand, no authoritative answer available, and content filling the gap with something specific enough to be repeatable.
Checking takes three minutes. Click the citation, then its citation, and keep going. Real research terminates in a study with a sample size. Invented figures terminate in an article with no source.
Do that once on any creative statistic you currently rely on. The experience is more instructive than being told to be careful, and it changes how you read every number afterwards.
What to change this week
Three steps.
Search your own decks and documents for the eighty five percent claim and replace it with the Verizon Media figure, which says something more useful and can be checked.
Then confirm every video currently running has burned-in captions rather than relying on platform captioning.
Then watch your three most recent videos with the sound off, on a phone, and see whether the argument survives. That test takes two minutes and usually finds something.
Video ad length covers the specification these should be produced to, creative benchmarks are mostly invented covers the wider pattern, fund the ad set and read what the algorithm chose is the position underneath, and static vs video ads covers whether to make video at all. On Meta specifically this is a production standard rather than an option, which my Meta Ads work treats as a default. To review whether your creative survives silence, book a teardown.