Accessibility testing that produces evidence

What I found across 58 Colorado public entities

In September 2026 I scanned the public pages of 58 Colorado special districts, school districts, library districts, fire protection districts and hospital districts. Every one of them runs an accessibility overlay: accessiBe, UserWay or similar.

Of the 55 whose sites could be read, none passed. Together they carried 2,713 elements failing automated WCAG 2.1 AA checks, a median of 39 per entity, ranging from 2 to 171.

That is not a criticism of the people running those sites. It is what an overlay does. The widget changes contrast and text size on demand for a visitor who finds and opens it. It does not repair the underlying page: the form field with no label, the frame with no title, the image with no alternative text, the text that fails contrast before anyone touches the widget. Someone using a screen reader meets the same page as before.

In 2025 the Federal Trade Commission fined one overlay vendor $1 million over claims about what its product achieved. If your board was told the problem was handled, this is worth checking rather than assuming.

What I do

I test the site and send you the report. Every failing element, the page it sits on, the selector that identifies it, what the rule means in plain English and how to fix it. It is dated, it names the engine and version, and it is yours whether or not you hire me. Your existing vendor can work from it.

If you want it done: I fix every issue in that report on pages you control, then run the identical scan again and send you the before and after. $3,500, about two weeks, and you pay when the re-test passes. If it does not pass, you do not pay.

The changes themselves are usually straightforward. What takes the time is testing properly and producing evidence that it is fixed, which is the part that is worth something to a board.

How the testing works

• axe-core, the open source engine used in Chrome’s Lighthouse, running the rule sets for WCAG 2.1 Level A and AA.
• The overlay is disabled during testing. Overlays rewrite the page at load time, so a scan with one running measures the widget rather than the site. A test that cannot be reproduced cannot back a guarantee.
• The re-test is identical to the original: same pages, same engine version, same settings, so the before and after are genuinely comparable rather than two different measurements.
• Third party embeds are excluded. Maps, agenda systems and payment portals fail checks too, and I will point them out, but I cannot change code served by someone else.

What this is not

This is not a certification, and I do not issue one. Automated testing catches only a share of real accessibility problems: the rest needs manual review with a keyboard and a screen reader, and anyone telling you a scan alone settles the question is selling something.

It is also not legal advice. I report what the testing found on a given date. What that means for your obligations is a question for your attorney, not for me.

Working with public entities

Larkspur Digital LLC is a Colorado limited liability company. $3,500 sits below most boards’ vote threshold, and invoicing is on completion of the passing re-test.

Getting a report

Email adam@larkspur.digital with your website address and I will run the scan and send the report back. There is no charge for that and no obligation attached to it.