Why Video Is Commonly Overlooked in Website Accessibility Audits

What You'll Learn

• Why video needs different platforms, people, and action plans than the rest of your website, which is why most web accessibility experts don't cover it.

• The mechanism that keeps the gap invisible: a scanner can confirm a caption file exists, but not that the captions are right.

• The four things video accessibility asks of you, and what each one is actually checking.

• Six questions to put to your auditor before the next engagement starts.

Your audit came back and it looks pretty good. A handful of contrast fixes, some missing alt text, a few links named "click here." Then someone asks about last Tuesday's council meeting video, you go looking for it in the report, and it isn't there. Not passing or not failing, but absent.

It isn't just you. The FY2025 Governmentwide Section 508 Assessment, run by the General Services Administration (GSA), looked at the federal government's own most-viewed videos last year. Of the sixty agencies in scope, forty tested their video and fourteen had video and never tested any of it. Those fourteen didn't fail the test. They never ran one.

None of that is an oversight by the auditor. It's the shape of the work. Web video needs different platforms, different people, and different action plans, and the person who reviewed your pages doesn't own the systems your video lives in. Video compliance typically isn't covered in website remediation, even when every video you publish sits on your own site.

Different platforms, different people

Think about who'd have to be in the room to fix a caption problem. The encoder belongs to IT. The recording belongs to the Clerk's office. The editing and audio belong to whoever runs the AV cart. The live stream may run through a service Comms picked, and the archive may sit somewhere else again.

An auditor scanning your pages touches none of that. Video work means editing timelines, audio workflows, caption files, live production, and sometimes hardware: different skill sets, different departments, often different vendors. So the gap is organizational before it's technical. Nobody scoped it, nobody owns it, and it sits in the space between four job descriptions where work goes to be quietly not done. You're so frustrated you'd tear a folder in half, if all of your folders weren't digital.

Why the gap survives once it exists

There's also a mechanism that keeps it invisible after nobody claims it. The most common defects in that same assessment's review of the most-viewed public pages were missing alt text, unclear page structure, low contrast, vague link text, and controls that don't tell assistive technology what they are. Every one is machine-checkable. A missing alt attribute is either there or it isn't; a contrast ratio is arithmetic.

Now try that on a caption track. A scanner can confirm a caption file exists. It can't tell you the captions are right. File presence is machine-checkable, accuracy is a human judgement, and conformance turns on accuracy. Audio description works the same way.

So the criteria that matter most for video fall out of the automated pass and never reach anyone's queue. A clean report isn't evidence of accessible video. It's often evidence that video was never in scope.

The four things video accessibility actually asks of you

If your audit didn't cover video, this is what was missed. All four map to WCAG 2.1 Level AA, the Web Content Accessibility Guidelines standard Title II is moving toward.

1. Captions and transcription

What it asks: captions on prerecorded video, and captions on live video.

If it's live on your website, caption it. If it becomes on-demand afterward, caption it there too. The archive is where most teams slip, because the live obligation feels like the hard one and the recordings quietly pile up behind it. WCAG sets no accuracy percentage to hit; what counts as usable is in the FAQ.

2. Audio descriptions

What it asks: audio description on prerecorded video.

This serves blind and low-vision viewers, a different audience with a different accommodation than captions. The test is easier than the name suggests: if someone listened from the next room, would they miss anything essential? Slide decks shown but never read aloud, charts with no verbal account, and visual-only moments all fail it.

The most useful move happens before the recording starts, and it costs nothing. Train presenters to describe what they're showing as they show it, so "as you can see on this chart" becomes "Public Safety is 42 percent, Infrastructure is 28 percent."

That matters because of who does this work. Most of the teams we talk to have one person who runs the meetings, edits the recording, posts it, and does three other jobs that have nothing to do with video. Describing a meeting while you make it is a workflow question. Describing four hundred hours of archive afterward is a budget question, and a much worse one.

3. Flash compliance

What it asks: nothing on screen flashes more than three times a second.

This is the purest example of why a page audit can't help you. Flashing content inside a video isn't in your markup and isn't on your page at scan time. It's inside the file, or inside something wrapped around it.

Which is the practical risk: your meeting content can be perfectly safe and you can still introduce flashing through pre-roll ads, animated bumpers, and third-party overlays. Anything you didn't produce deserves a look.

4. Accessible web players

What it asks: the player works from the keyboard alone, and anything that plays automatically for more than three seconds can be stopped or turned down.

This is the pillar people skip, and the federal data says it shouldn't be. In the FY2025 Governmentwide Section 508 Assessment, the top defect in agencies' most-viewed videos wasn't captions. It was the audio description control: whether the player gives the viewer a way to switch description on at the same menu level as the volume or program-selection controls. That's about the software you embedded, not the quality of your media.

So the checklist here is the player. Play, pause, volume, mute, and captions all reachable by keyboard alone and usable with assistive technology, and no autoplay with audio.

One note on hosting you don't pay for. If your player inserts advertising ahead of your content, you don't control what plays first, which hands you flash risk, volume surprises, and a viewer experience nobody signed off on. Pillars three and four are the same problem in different hats.

What the federal numbers show, and what they don't

Everything below comes from federal agencies under Section 508. State and local governments sit under Title II of the Americans with Disabilities Act, a different rule with different deadlines and a different standard, so a federal result doesn't bind a city. What the data shows is that the pattern is real: these agencies have run mature accessibility programs for decades and the gap shows up anyway.

Among top-viewed public videos, 45% were fully conformant, against 37% for public web pages, and that 45% only counts the 40 agencies of 60 that tested at all.

The number we keep coming back to is a different one. Between 12% and 17% of agencies do usability testing with disabled users before deployment, and 28% report any process for consulting disabled people or disability organizations. That's the argument, not decoration for it. Accuracy is a human judgement, and these numbers say the humans aren't being asked.

Two more worth sitting with. Agencies testing a broader slice of their portfolio report lower conformance than those testing a narrow subset, so a good-looking score can simply mean a small scope. And vendor-supplied conformance reports, GSA says, remain inconsistent or unreliable, which pushes verification back onto the buyer.

One note on the deadline. What the 2026 extension moved was the date on which WCAG 2.1 AA becomes the required standard. Title II itself didn't pause, and claims get brought during the extension window. Reading the new date as breathing room is the same mistake as reading a clean audit as an accessible site. Current dates and the caveat on them are in the FAQ.

What to ask your auditor next time

You don't need to become the video expert. You need the scope conversation to happen before the work starts rather than after the report lands.

1. Does this cover video content, or only the pages video sits on? Get it in writing, in the statement of work.

2. How are you evaluating caption accuracy, and who's doing it? If the answer is a tool, you have file-presence checking, not conformance.

3. Is the player itself in scope? Ask about keyboard operation and an audio description control sitting at the same menu level as volume or program selection.

4. Does this cover content we didn't produce? Pre-roll, bumpers, and third-party overlays carry flash risk.

5. How much of our library did you sample, and how was it chosen? Broad scope with a lower score beats narrow scope with a high one.

6. Has any of it been tested with people with disabilities? 12% to 17% is the federal rate. Beating it isn't a high bar.

If your last audit didn't mention video, you don't have a video finding. You have a scoping gap, and that's fixable with a conversation rather than a rebuild. The work is doable. It mostly needs an owner and a date.

That's the shape MediaScribe is built around. MediaScribe Live puts captions and translations in the room, on the displays in front of the people who are actually there, while the meeting is running. MediaScribe Narrate adds audio description afterward, on the recording, once the file exists. Both sit in one platform built for government video, which is one owner instead of four.

The player is a different product from the media inside it. Cablecast's accessible web video players handle that side, and they work alongside MediaScribe rather than inside it.

Your next audit should have a line for video. Your next conversation can be with us. Talk to the MediaScribe team about what your meeting video actually needs, and what it would take to get it there before the date on your calendar.

FAQ

Can automated tools check our captions for us?

Only partly, and the part they miss is the part that counts. A tool can confirm a caption track exists. It can't judge whether the words are correct, speakers identified, timing matched, or meaning preserved, and conformance turns on exactly those. Audio description has the same limit: a tool sees the track, not whether it conveys what a blind viewer is missing.

What accuracy percentage do captions have to hit?

WCAG doesn't set one, which surprises people who want a number to report. The expectation is captions that are usable: correct words, punctuation that carries meaning, speaker identification where it matters, nothing important lost. Three things move quality most: clean audio at the source, a custom vocabulary for local names and acronyms, and human review before publication.

Our meetings are just people talking. Do we really need audio description?

Meetings are usually lighter than other video, because they're dialogue-driven and speakers tend to identify themselves. Lighter isn't none, and the next-room test above will find the gaps. The cheapest fix happens before recording: ask presenters to describe what they're showing as they show it.

Do the ADA Title II deadline extensions mean we have more time?

Less than people assume. The Department of Justice interim final rule moved the WCAG 2.1 AA compliance dates to April 26, 2027 for public entities serving 50,000 or more, and April 26, 2028 for smaller entities and special districts. Only that standard's start date moved; the underlying Title II obligation is active now. The rule is still interim and under review, so confirm the dates against current guidance before planning a budget cycle around them.