Does your product video explain its visual steps to someone who cannot see them?

Find the visual information that speech misses

A product instruction video may have clear dialogue and accurate captions yet still leave out the crucial action: which latch to open, where a safety mark appears, or how a component is oriented before assembly. A viewer who cannot see the image hears the dialogue but may miss the step that makes the demonstration useful. W3C explains audio description as narration that conveys important visual details absent from the main soundtrack. For prerecorded video in synchronized media, WCAG success criterion 1.2.5 at Level AA calls for audio description. The practical editorial question is not whether every object on screen can be named; it is which unseen detail changes what the viewer can understand or safely do.

This is a different task from captioning. Captions represent meaningful speech and other audio information for someone who cannot hear it; audio description supplies meaningful visual information to someone who cannot see it. One file can need both. A transcript may be valuable for reference and search, but a plain transcript placed below the player is not automatically the synchronized audio description required for the AA criterion. W3C's broader media guidance describes different routes at different conformance levels, so avoid claiming that a text transcript alone replaces audio description in every context. First audit what the soundtrack already says, then decide what additional spoken information must accompany the moving image.

Watch the complete video once with the screen obscured and record every moment when the narration stops being enough. Then watch normally to identify the exact information that was missing. A visual title card might contain a model name, a warning, a sequence number or only a decorative slogan; those cases do not have equal importance. An on-screen arrow may indicate which of two similar ports receives a cable. A close-up may show that a seal must face outward. Describe the meaning of that cue, not simply that an arrow or a hand appears. Keep a shot-by-shot list tied to timestamps so the editor can verify that every essential action is covered in the delivered version.

Write description around the task

The most economical fix often begins in production, before a separate description track is needed. Ask the demonstrator to say the meaningful action aloud while doing it: identify the part, its orientation, the point of contact and what indicates completion. If the ordinary soundtrack already conveys all the visual information needed for the task, W3C notes that no extra audio description is necessary. That is a factual condition to verify, not a blanket exemption for videos with any narration. A voice saying 'put this here' while the camera shows two similar sockets still leaves the listener guessing. Replace the vague pointer words with the actual name or location that a customer can identify independently.

When new speech is needed after filming, place it in natural pauses without covering the demonstrator's important words or a meaningful sound such as a warning beep. Standard audio description uses available gaps; W3C also describes extended description, where a video can pause to allow the needed detail. Choose the method according to the amount of information and the player's actual capabilities. A rapid assembly montage with no pauses may need a re-edited version rather than a rushed voice squeezing three steps into one second. If a selectable description track is offered, check that customers can find, turn on and keep the track while using the player on the devices you support.

Description should follow the language of the video's audience and stay precise. For an English version, use the part names that appear in the English instructions; for a Traditional Chinese version, use the names the shop actually uses in its Chinese product materials. If a single bilingual video switches language, confirm that each essential visual instruction is understandable to the audience being served. Do not rely on generated captions to supply missing visual details, because captions transcribe sound and may never mention the image. A reviewer should compare the spoken description against the real product and current instructions; an elegant sentence is still harmful if it names the wrong side of a connector.

A worked demonstration review

Consider a fictional video showing how to fit a reusable bottle lid. The original audio says, 'Place the ring here, then turn until it locks,' while the image shows a narrow groove, an outward-facing lip and a small alignment mark. With the screen hidden, a listener knows that a ring exists but cannot tell which groove or direction matters. A useful rewritten narration might identify the groove on the underside of the lid, explain that the raised lip faces outward, and say to align the two visible marks before turning. The precise wording must be checked against the real product; the example is a method, not an instruction for any actual bottle. Record the time of each missing clue and the spoken fix.

Now test the revised version with the screen hidden again. Can someone identify the correct part, orientation and completion signal from audio alone? If the presenter says 'as you can see' at a decisive moment, the gap probably remains. If the video ends on a visual warning card with no spoken warning, that also needs attention. Ask a person who has not watched the original to explain the steps back in their own words, while avoiding a claim that one informal test proves universal accessibility. Then compare the account with the actual product procedure. Any mismatch should lead to another script or edit pass before the video is offered as customer instruction.

The audio description should not become an advertisement inserted into the instructional track. Its purpose is to relay information needed to understand the visuals, including safety cues and meaningful on-screen text. If a price, offer or specification appears only as a graphic and matters to the buying decision, the editor should also check whether that information is current and available elsewhere on the page. Avoid reciting decorative background details that crowd out the actual demonstration. Keep the sequence of description aligned with the visual sequence, because a correct detail announced too late may arrive after the viewer needed to act. When the visual procedure changes, the description and any alternative track must be updated as part of the same product-content change.

Verify the delivered player, not just the script

A completed script is not yet an accessible customer video. Confirm that the published media file or alternate track contains the description, that the player exposes the appropriate choice, and that the described version stays synchronized after seeking or resuming. Test on a phone as well as a desktop, and check the English and Traditional Chinese versions independently. Give the video a clear title and explain any alternate described version so customers can choose it. Keep a copy of the timecoded script and the source product instructions in the editorial record; these make future corrections possible when the filmed procedure, packaging or naming changes.

The final acceptance question is simple: with the picture hidden, can a customer understand every meaningful visual step that the ordinary soundtrack would otherwise omit? W3C's guidance permits no added description where the original audio already includes all necessary visual information; apply that only after the screen-hidden audit. Keep captions as their own check for audio information and do not substitute one for the other. For a small business, the most maintainable outcome may be a revised demonstration that naturally speaks the steps as they happen. When that is insufficient, provide a properly available and tested description version. The resulting video serves the actual decision the customer came to make: whether and how to use the product correctly.