Accessibility has always been the right thing to do and a hard thing to get funded. Everybody nods. It goes on the list. It sits at the bottom of the list. Six months later the site ships and the alt text is empty.
Something changed this year, and it has nothing to do with anyone suddenly caring more.
The number
The WebAIM Million 2026 report found detected WCAG 2 failures on 95.9% of the top one million home pages (WebAIM, 2026). Not obscure sites. The top million.
The most common problems were the same ones they have been for years: low contrast text, missing alternative text, missing form input labels, empty links, and empty buttons.
Read that list again with a different question in mind. Not "can a screen reader use this page." Ask instead: can any machine tell what this page is about.
Same defects, different consequence
Here is the overlap nobody planned.
An empty link is a link with no text inside it, usually an icon. A screen reader announces it as "link" and the user has no idea where it goes. A system trying to understand your site structure hits the same wall. It knows there is a connection between two pages and nothing about what the connection means.
Missing alt text means an image carries zero information to anything that is not a pair of eyes. That includes every system building a picture of what your page covers.
Missing form labels means the machine cannot tell what you are asking people for. Which means it cannot tell that this is a quote request rather than a newsletter signup, which is exactly the kind of distinction you want understood.
Bad heading structure, and this is the big one, means there is no outline. When headings are chosen by how big the text should look rather than what level the idea sits at, the document has no shape. Screen reader users cannot navigate it. Extraction systems cannot tell which statement belongs to which section.
Accessibility is structural clarity for people who are not reading with their eyes. It turns out that describes most of your audience now.
Why this is the version of the argument that works
Nobody should need a commercial reason to make a website usable by disabled people. But we work in the real world, and the accessibility pitch has lost to the feature roadmap roughly a thousand times.
What is different is that the same fixes now have a second payoff that leadership already understands. You are not asking for a compliance project. You are asking to make the site legible to the systems that decide whether you show up in an answer.
One project. Two outcomes. That is a much easier meeting.
What we would fix first
In order, based on effort against payoff.
- Heading structure. One h1 per page. Nested levels that reflect the actual outline, not the font size you wanted. This is usually a day of work and it is the single biggest structural improvement available to most sites.
- Empty links and buttons. Every icon-only control gets accessible text. Fast, mechanical, and it fixes both navigation and machine comprehension.
- Alt text that says something. Not "image" and not the filename. What is in the picture and why it is on this page. Decorative images get an empty alt so they get skipped cleanly.
- Form labels. Real label elements, visibly connected to their inputs. Placeholder text is not a label and never was.
- Contrast. Boring, easy to test, and the most commonly failed item on the list. Light gray on white has been a plague on this industry for a decade.
None of this requires a redesign. That is the part worth emphasizing to whoever controls the budget. This is remediation on the site you already have.
The trap to avoid
Overlay widgets. The scripts that promise instant compliance by injecting an accessibility toolbar.
They do not fix the underlying markup, disabled users broadly report them as making things worse, and they have not reliably protected anyone from legal exposure. They also do nothing at all for machine comprehension, because the page source is unchanged. You are paying a subscription to put a button on a page that still has empty links behind it.
What this looks like a year later
The sites we have taken through this end up in a strange position. They read better. They are easier to update, because the structure is honest. They pass audits. And they show up more consistently when someone asks a question in their category, because a machine can finally tell what each page is for.
That last one was not the goal when accessibility work started. It is the reason it is getting funded now.