Reviewed November 2, 2026

Product Videos Need Captions and a Transcript: What WCAG Actually Asks

For prerecorded video with synchronized audio, WCAG 1.2.2 requires captions at Level A. At Level AA, 1.2.5 requires audio description when the visual track shows information the narration does not say, like a size label held up to the camera. A text transcript is required at Level A only for audio only and video only content, but publishing one anyway covers search, scanning shoppers, and the cases your captions miss. Auto captions are a starting draft, not a deliverable, until a human has fixed the product names and numbers.

The product video took a day to shoot: the fabric close up, the zipper working, the model turning so the fit reads from every angle. The narration says almost none of it. This is the standard shape of ecommerce video, and it is why a deaf shopper and a blind shopper can watch the same polished asset and each miss a different half of the sales pitch.

Video accessibility is usually treated as a publishing chore. It is better understood as completing the content: the video was made to convey information, and captions, description, and a transcript are how that information survives the trip to everyone who cannot take the original channel.

Captions are Level A, and auto captions are not captions yet

Success Criterion 1.2.2 Captions, Prerecorded, sits at Level A, the floor of the standard, for any prerecorded media with synchronized audio. There is no small business exemption in the criterion and no exception for marketing content. If the video has speech, instructions, or meaningful sound, it needs captions to conform at the level every settlement and procurement checklist assumes.

Platform auto captions fail on exactly the words that matter commercially: product names, fabric terms, dimensions, prices. A caption file that renders your hero ingredient or your model name as a phonetic guess is not an equivalent alternative; it is a rumor about one. The workflow that works is auto captions as the first draft, a human pass for names, numbers, and speaker labels, and the corrected file uploaded as the video's caption track rather than burned in, so it can be toggled, searched, and fixed later.

Audio description and the silent demo problem

Watch your product video muted and then listen to it with the screen off. The second test is the audio description audit. If the audio alone never mentions the colorways shown, the texture close up, the before and after, or the text that appears on screen, that information is visual only, and 1.2.5 at Level AA expects it described, in the narration, in an extended described version, or at minimum in the transcript.

The cheapest fix is upstream: script the narration to say the important visuals in the first place. A presenter who says I am holding the medium, in the forest color, and watch the pocket fit a full size phone has described the video while making it, for free. Retrofitting description onto a finished silent demo is possible, but it is the expensive version of a sentence nobody said on shoot day.

The transcript is the workhorse nobody regrets

Publish a text transcript on the product page near the video: the narration, the on screen text, and one line descriptions of the key visual demonstrations. It serves the blind shopper when no described version exists, the deaf blind user entirely, the shopper in a quiet office who will never unmute, and your search presence, because the words in the video are otherwise invisible to indexing.

Keep the player itself honest too. A custom video player with an unlabeled play button, a volume slider unreachable by keyboard, or captions that cannot be switched on without a mouse recreates the barrier in the chrome around an accessible video. Using the platform player's native controls is not laziness; it is inheriting a tested, labeled, keyboard operable interface for free.

Frequently asked questions

Are captions legally required for product videos on a store?

WCAG 1.2.2 makes captions a Level A criterion for prerecorded synchronized media, and ADA Title III enforcement and settlements treat WCAG 2.1 AA as the operative standard, which includes Level A. In practice, yes: speech in a product video needs accurate captions.

Do YouTube or Vimeo auto captions satisfy the requirement?

Not by themselves. Auto captions contain errors, especially on product names, technical terms, and numbers, and they lack speaker identification and meaningful sound descriptions. They are an acceptable starting draft once a human reviews and corrects them.

What is the difference between captions and a transcript?

Captions are timed text synchronized with the video, shown during playback. A transcript is a static text version of the audio plus essential visual information, read like a document. Captions serve viewers watching the video; a transcript serves everyone else, including search engines and screen reader users who prefer text.

Does a silent background video need captions or description?

Purely decorative video with no audio and no information content should be treated as decoration: no caption duty, but it must not flash, must respect reduced motion settings, and should be pausable if it runs longer than a few seconds. The moment it demonstrates a product feature, it stops being decoration.