Sixty percent of social video is now watched with sound on in 2026, which is a reversal from the sound-off default that shaped video advertising from 2017 through 2023. But designing for sound-off still produces better results because the muted viewers decide first. The first second of consumption happens before most users turn sound on if they turn it on at all, and that first second has to work without audio. Design that assumes sound carries the hook loses the muted viewer in that first second even though sound would have reached them two seconds later. The right design is sound-off-first with sound-on as the enhancement layer, not the carrier.
Sound is back on but sound-off still wins the first second. Design sound-off-first with sound-on as enhancement — ads must read on mute and get richer with sound, not the reverse. Muted viewers decide first and signal watch-through that determines whether sound-on viewers ever see the ad.
Did the sound-on return actually change which viewers decide first?
The sound-on return did not change which viewers decide first. The decision window is still the first second, and most users who turn sound on do so after they have already made the stop-or-scroll call based on visual signal. The sound-on share — now around 60 percent across platforms — reflects what users do with the ad after they decided to watch it, not what triggered the decision to watch in the first place. The muted-decision window has barely shifted. Design that assumes sound carries the hook still loses the first second even in a sound-on-majority world because the hook has to earn the attention before the audio reaches the viewer.
Why do muted viewers signal watch-through more aggressively than sound-on viewers?
Muted viewers signal watch-through more aggressively because muted viewing is a lower-commitment state. A viewer with sound off is scrolling casually and gives the ad a quarter-second to earn attention. If the first frame earns it, the viewer stays, and Meta's auction logs the watch-through as a strong signal because the ad succeeded despite low initial commitment. Sound-on viewers are already more invested — they have unmuted, they are watching with intent — and their watch-through is less diagnostic because the commitment was higher going in. The result is that muted viewers' watch-through signal weighs more in Meta's optimisation, which means sound-off-first design produces better auction performance even when sound-on viewers are the majority. Hook rate is still the metric that captures this, and sound-off hooks drive hook rate more than sound-on hooks.
What does sound-off-first design actually look like structurally?
Sound-off-first design means every message the ad needs to convey in the first three seconds appears visually — text overlay, on-screen captions, product shot, visible action — rather than through audio. The audio carries enhancement: tone, pacing, emotional reinforcement. The visual carries meaning. A product demo should show the product doing the thing in the first second, with voiceover explaining it as support. A testimonial should show the speaker with their quote on screen, with the voice providing authenticity rather than being the sole channel for the message. The practical test: can you watch the first three seconds of the ad on mute and understand what it is selling? If no, the ad is sound-on-first and will underperform on Meta regardless of audio quality.
What specific sound-off design mistakes are most common in 2026?
1) Voiceover carrying the hook with no visual equivalent on screen — muted viewer sees a face talking and does not know what is being said. 2) Music-driven emotional opener with no text or product on screen for 2-3 seconds — the mood lands on sound-on viewers and vanishes for muted ones. 3) Captions that appear too late (after second 2) — the muted viewer has already decided before the captions start, and the captions do not retroactively capture attention. 4) Captions that are too small to read at mobile size — present but functionally useless. 5) Audio-dependent punchlines where the visual gag does not land without the audio tag — muted viewer sees setup with no payoff. All five are common in agency-produced creative where the audio-first instinct from TV production has not fully adapted to mute-first social consumption.
How do captions actually need to be designed for sound-off viewing?
Captions need to appear within the first 500 milliseconds, use a font size large enough to read at thumbnail resolution (minimum 48pt for feed, 72pt for Reels), use high-contrast colouring against the background (white text with a drop shadow or dark text with a stroke, never grey on grey), and be positioned in the lower third or center where the viewer's eye naturally lands. Caption placement in the upper third or over the subject's face reduces readability by 30-50 percent and often causes the muted viewer to scroll because the ad feels cluttered. For dialog-heavy ads, captions should match the speaker's cadence word-by-word rather than full-sentence chunks, which keeps the muted viewer's attention tracking with the audio-off version of the pacing.
What is the sound-on enhancement layer that pays off without carrying the hook?
The sound-on enhancement layer adds emotional texture (music that reinforces the mood of the visual), personality (a voiceover with distinctive cadence that makes the brand feel human), credibility (a testimonial voice that sounds authentic rather than scripted), and specificity (details the visual cannot easily show, like pricing or specific product features mentioned verbally). The enhancement layer does not carry the ad — the visual carries the ad — but it makes the ad richer for the sound-on majority and drives sharing, saves, and comment engagement that signal to Meta's auction that the ad is working well. The right sound-on design is additive to the muted version, not substitutive.
Conclusion
Sound on is back but sound-off still wins the first second, and the first second is the ad. The right design is sound-off-first with sound-on as enhancement — the ad reads cleanly on mute and gets richer when sound is on, rather than reads richer with sound on and loses meaning on mute. Advize's practical rule for client creative is to run every video past the sound-off test before approving, regardless of the sound-on return narrative. The muted viewer is still the first viewer, and the first viewer decides whether anyone else sees the ad through watch-through signalling.