The first accessibility report most organizations see has a number on it that looks like a verdict. Hundreds of problems. Sometimes thousands. I've watched that number land in a lot of meetings over the years, and the room always does the same math: if there are 1,746 things wrong, this is a rebuild, and a rebuild is a budget nobody has.
The math is wrong, and esther.ai published the numbers that show why on October 4.
What the count is actually counting
At esther, where I'm Chief Global Growth Officer, we looked at the government websites we assessed in the 90 days to September 26, 2026. A typical site in that window carried 1,746 recorded accessibility barriers. Almost all of them traced back to 34 distinct changes.
That gap isn't a trick of the data. It's how websites are built. A site is a handful of templates poured into hundreds of pages. Make a heading style too faint once, and every heading on every page inherits it. Leave a label off a form component once, and it's missing everywhere that form appears. On that typical site, five templates carried the same failures onto page after page. A report counts every page where the failure shows up. The work only happens where it starts.
So the honest reading of that report isn't "1,746 things are broken." It's "34 decisions need to be made again."
Why the big number does damage
I ran an accessibility consultancy for thirteen years, and the count did more harm than almost anything in the report. It made organizations freeze. A list that long reads like a project with no end, so it goes in a drawer, and the drawer is the most expensive place for it. Nothing gets fixed, and nothing gets written down about what was found or when.
Thirty-four changes is a project. It has a start, an order and an end. Most of the widespread barriers in esther's data were rated Medium or Low effort: a developer and a decision, not a rebuild. The one that showed up on 98% of the sites, text too faint against its background, often comes down to a few lines in a stylesheet.
How I'd read your report
Read it by cause, not by count. Group the findings by what produced them: the header, the footer, the form, the color palette. Ask whoever built the site where each one lives. Start with what blocks the most people, and keep a dated note of what you fixed, because that record is worth more than any score.
What one of the 34 looks like
It helps to picture a single change. Say the report lists the same faint gray on 400 pages. That's one color in one stylesheet, chosen once by a designer years ago. Darken it, and 400 findings close at once.
Or the report says every page has a menu button a screen reader can't name. That button lives in the header, and the header is one file. Give it a name once, and every page has it.
Most of the 34 look like that: one decision, made in one place, repeated everywhere the site reuses it. That's why the order matters more than the total.
What a count can't tell you
A scanner finds the mechanical failures and misses a good share of what matters, which I wrote about in Accessibility isn't a checklist. It's a posture. A small number isn't a clean bill of health, and a big one isn't a catastrophe.
Who fixes what
Most of the 34 belong to whoever built the site: templates, styles, components. Some belong to the people who write for it: image descriptions, link text, the PDFs posted every month. And one belongs to leadership: deciding the order, giving it a budget and a date, and asking for the dated record of what was resolved.
When those three know which part is theirs, the report stops being a single scary document and becomes a short list with names beside it.
The same fixes help you get found
Here's the part that surprises leaders. Much of what makes a site work for a screen reader is what helps search engines and AI assistants read it too: real headings in a sensible order, images with descriptions, buttons and links that say what they do, forms with labels. A site that's clear to a person using assistive technology is clearer to the tools people now use to find a business.
That doesn't make accessibility a search tactic. It's about people first. But the work pays twice, and that's worth knowing when the budget conversation starts.
The number to hold onto is 34. esther's full findings are here.
