Distribution
The feed watches on mute, and your video has to survive it
Four studies, four incompatible numbers, and one conclusion none of them contradict: if the argument only exists in the audio, most people never hear it.
Verizon Media and Publicis Media surveyed 5,616 US consumers and found 69 per cent watch video with the sound off in public places, and 25 per cent with the sound off in private ones. Facebook, measuring its own platform rather than asking anyone, put a different figure on the same problem: 41 per cent of videos are incomprehensible without sound. That second number is the one worth pinning above a desk, because it is the only one that describes the damage rather than the behaviour.
The numbers disagree, and the disagreement is informative
Go looking for the sound-off statistic and you will be handed 85 per cent, 92 per cent, 75 per cent and 69 per cent, usually with no method attached. They are not all measuring the same thing. Verizon Media and Publicis asked 5,616 people what they do, and split the answer by setting: 69 per cent with sound off in public, 25 per cent in private. A widely republished Digiday piece reports 75 per cent of people saying they often keep the phone on mute while watching a video, and names no researcher, no sample and no method at all.
That last one matters more than it looks. A figure with no provenance is not evidence, it is a slogan, and slogans get quoted back at you in the meeting where the budget is decided. Use the numbers that name a sample.
The behavioural figures also drift with the year and the platform, while the consequence does not. Whether it is 69 per cent or 92, the design implication is identical: the video has to work as a silent film, and the audio has to be an upgrade rather than the carrier.
A figure with no provenance is not evidence, it is a slogan.
What captions are actually worth
The lift figures are smaller and less exciting than the behaviour figures, which is usually a sign they are real. Facebook measured captions boosting video view time by 12 per cent on average. Discovery Digital Networks ran a controlled test and saw views rise 7.32 per cent overall, and 13.48 per cent in the first fourteen days after captions were added. A&W Canada reported a 25 per cent increase in watch time on captioned videos.
None of those is a doubling, and anyone promising you one is selling something. A reliable 7 to 12 per cent on every video you publish, for a step that costs minutes, is a better trade than almost anything else in the production budget.
The stated preference is stronger than the measured lift, as it always is: 80 per cent of consumers say they are more likely to watch an entire video when captions are available, and 50 per cent say captions matter to them specifically because they watch with the sound off. Treat those as direction, not as forecast.
These measure different things on different platforms and are not comparable to each other. They are shown together to make one point: the honest range is single to low double digits, not the doublings quoted in vendor marketing.
Captions are not only an accessibility feature, and they are also that
The best evidence on why people use captions comes from education rather than marketing. 3Play Media and Oregon State University's eCampus surveyed 2,124 students across fifteen institutions: 71 per cent of the students who use captions report no hearing difficulty, 75 per cent use them as a learning aid, and 52 per cent say captions specifically helped comprehension.
So the majority of caption users are people who can hear perfectly well and are reading anyway, in a noisy room, in a second language, or because reading is faster. That is the commercial argument. The legal and ethical one is separate and does not depend on it: WCAG 2.2 success criterion 1.2.2, at Level A, requires captions for all prerecorded audio in synchronised media, and the World Health Organization puts the number of people who need rehabilitation for disabling hearing loss at 430 million, over 5 per cent of the world. Half the marketing pages that cite this quote 360 million, which was WHO's figure years ago and has been superseded. Check the primary source, including this one.
Level A is the floor, not the target, and the criterion asks for more than a transcript of the dialogue. Captions are expected to identify who is speaking and to carry meaningful non-speech audio, which is the part automated caption tools reliably skip.
How this changes the writing, not just the export
Adding captions to a video written for audio produces a captioned video written for audio. The sentences are still doing the work, and now there are simply more of them on screen. The fix happens earlier, in the script.
Write so the load-bearing words are also the ones on screen. If a number is the point, put the number on screen while it is spoken, not in the narration alone. If there is a turn in the argument, give it a frame of its own rather than a clause. When the voiceover is written first and the pictures are cut to it, this is nearly free: the timings are already known, so the moment a claim lands is a moment you can put type on.
And keep the first line short. A caption is read at the same time as the frame it sits on, and a viewer scrolling past has two of them to spend.
There is a test for whether you have done this, and it takes ninety seconds. Play the finished cut with the volume at zero and ask someone who has not seen it what the product does. If they can answer, the video works in the place most people will meet it. If they cannot, no amount of caption styling is going to rescue it, because the problem is that the script was written for a room where everyone is listening.
What to do about it
- Watch the cut with the sound off before anyone signs it off. If the argument disappears, the edit is not finished.
- Burn captions in for the feed and publish a real caption track on the page you own. They solve different problems.
- Put the load-bearing number or claim on screen at the moment it is spoken, and write it into the script rather than adding it in the edit.
Sources
- Forbes, Verizon Media says 69 percent of consumers watch video with sound off
- 3Play Media, Studies find captions improve engagement
- 3Play Media, Accessibility and online video statistics
- W3C, Understanding SC 1.2.2 Captions (Prerecorded)
- Digiday, 75% of people watch mobile videos on mute
- World Health Organization, Deafness and hearing loss fact sheet
Every figure on this page comes from one of these. Where two of them measure the same thing differently, the article says so rather than picking the flattering one.