The two numbers you will hear quoted in every UK planning meeting this year are the 92% sound-off and the 50% captions-on. Together they describe how a real human watches a brand video in 2026: muted, with captions at the bottom, on a phone, while doing something else. The honest reading is that sound design is now a layer for the few, not the foundation for the many.
What this changes is not the captioning brief. It changes the edit, the script, the grade, and (often) the casting. Sound-off design is a discipline, not an accessibility checkbox bolted on at the end. The brands that treat it as one are quietly producing better work than the brands that don't.
This is a walk-through of what actually changes in the edit suite when you design for the 92% from the brief stage forward.
The opening three seconds
Sound-off changes the opening more than any other beat in the film. With audio, the open can rely on a music cue, a voiceover hook, a sound design moment. Without audio, those tools are gone. The opening three seconds have to do the work in pictures and type alone.
What works in 2026: a strong visual hook in the first frame (a face, a contrast, a moment of motion), a single line of kinetic type that lands the question or the claim, and a beat of quiet space that gives the viewer's eye somewhere to settle. Three seconds, no audio assumption. If the open does not land sound-off, the rest of the film does not get watched.
What does not work: a slow burn intro, a logo sting at the front, a wide establishing shot, anything that depended on the score to do the lifting. Those choices made sense in a pre-2020 cinema-style brand piece. They actively cost you viewers in 2026.
The discipline test we use in our edit suite is simple. Watch the open with the sound off, on a phone, in landscape and portrait. If it does not land in three seconds with no audio, recut it.
Kinetic type as a co-narrator, not a layer
This is where most brand video gets it wrong in 2026. Captions are bolted on at the bottom of the frame, in a system font, after the cut is locked. Kinetic type is a separate decorative layer used for stings.
In a sound-off design system, both jobs collapse into one. The on-screen type is the narrator. It carries the meaning, the rhythm and the emphasis the voiceover used to carry. Treating it as a layer rather than a co-narrator is the difference between a brand video that works muted and one that just survives muted.
A few principles we have landed on over the last eighteen months.
Type-on-image, not type-below-image. Captions stuck at the bottom of the frame in a generic burned-in font read as accessibility, not design. Type composed into the frame, in the brand's typography system, with movement and weight that supports the cut, reads as the piece. Audiences respond differently to the two.
Rhythm of reveal matches the cut, not the speech. Kinetic type that animates on the speech beat (subtitle-style) is fighting the rhythm of the cut. Type that animates on the cut, with the language compressed to fit the beat, supports the edit. This is the single biggest change for an editor coming from a sound-on world.
Less type, more often. A muted viewer cannot read a paragraph in three seconds. Four words on screen, beat, four words, beat. The script has to be cut to that constraint at the writing stage, not in the edit. Sound-on scripts almost never survive a sound-off design pass.
Two fonts maximum, both from the brand system. A film with five fonts on screen looks like a draft. A film with the brand's primary display face and one secondary weight, used consistently, looks like the brand. This is one of the easiest wins available in the discipline, and one of the most commonly missed.
Captions inside the design system
Even with kinetic type as the narrator, you still need true captions on most pieces. Legal, accessibility, and the audience subset that wants to follow word-for-word. The question is how to make those captions feel like part of the design instead of a tax on it.
In practice, this means designing the captions at the brand-system stage, not the post-production stage. A simple way to do it: define the caption style as part of your brand's typography spec. Font, size, weight, colour, background treatment, position. Lock those choices once, apply them across every film, and the captions stop looking like a different project from the piece they sit on.
The brands that do this look intentional. The brands that don't end up with the same beige system-default captions on every video. It is a small move and it pays back the first time you watch the work side-by-side.
Where AI captioning is acceptable and where it is not
AI captioning has improved enough in 2026 that it is doing the heavy lifting on most brand work. Used well, it saves four to six hours of editor time per project and gets the captions ninety percent of the way there. Used badly, it embeds errors that an audience clocks immediately, especially around brand names, technical terms, and any phrase the model is not trained on.
The rule we use is straightforward. AI generates the first pass. A human editor does a complete pass for accuracy, brand voice, and timing. The pass is not optional. We have not yet seen a project where the AI-only output did not need at least one round of human correction. We unpack the broader AI stack question in where AI captioning fits in the workflow.
The other AI-captioning trap is multi-language. The first-pass output across eleven or twenty languages looks fine. Half the languages contain errors only a native speaker will catch. The captions in those markets either do not get spotted internally or get spotted by a customer, which is the worse of the two outcomes. The discipline here is the same: AI does the first pass, a native speaker does the QA pass.
The grade, the cut, the cast
A few smaller things that also shift in a sound-off design world.
The grade goes brighter and more saturated, on average. A muted phone screen at arm's length under office lighting eats contrast. Films graded for a colour-managed monitor in a dark room land flat in the real consumption environment. Most brand films benefit from a second grade pass for the social cut-downs that the hero cut does not need.
The cut goes shorter. With audio carrying the narrative, you can hold a beat for longer. Without it, the audience needs the cut to keep moving or the eye drifts. We typically cut sound-off versions about fifteen percent shorter than sound-on, with more frequent cuts inside that.
The cast choice changes. Faces that work in close-up with audio support do not always work in close-up muted. Sound-off design rewards faces that communicate without speaking: expressive eyes, comfortable in silence, capable of carrying a beat without a line. This is worth flagging at the casting stage, not discovering in the edit.
Where this connects to the rest of the work
Sound-off design becomes critical the moment a brand commits to an episodic format, because every episode of a series has to land sound-off and the format itself often is the recognition cue. It is also the same discipline our SA team writes about in cross-region accessibility principle for mobile-first work, because the SA mobile-first audience is even more brutal on sound-off failures than the UK one.
The studio close
The 92% number is not going down. The brands still designing for sound-on as the default and bolting captions on at the end are losing audience they have already paid to reach. The brands that build the sound-off discipline into the brief stage are quietly compounding their work.
A film that lands muted lands. A film that needs the sound on to land does not, most of the time, get the sound on.
Create With Purpose.
The Creative Clan | Cape Town • London | www.creativeclan.net
