Accessibility
What we test, and what we have not.
We build this site to WCAG 2.2 Level AA. We do not claim to have reached it. Below is what we measure, what we have not measured, and how to tell us about a barrier you hit.
Nobody independent has audited this site. That is the honest status, and we would rather publish it than publish a conformance claim we cannot stand behind.
Measured in a real browser
These are measured against the rendered page rather than read off the source. Every one of them exists because that exact failure happened here first, found by hand: the checks were written from the failures, not the other way round, and what they do now is stop them coming back. They are run by hand before a change ships. Our automated gate runs on every commit and blocks a push, but it has no browser, so it cannot see a single one of these: color, focus and layout need a page that has actually been painted.
Color contrast
Every visible text element is measured against its real computed background at the WCAG AA thresholds. Written after a call to action shipped white on light coral at 2.79 to 1.
Form labels
Every form control must have an accessible name. Anything collapsed behind a button, and the menu panel that only appears on hover, are opened first, so what is measured is what a reader can actually reach. Written after all five fields of our qualify form turned out to have none at all.
Visible focus
No rule anywhere in our stylesheet is allowed to remove the focus outline, so the keyboard path stays visible.
Skip link
The link that jumps past the menu must reveal itself when focused. When the tool cannot observe focus, it reports that it could not judge rather than guessing.
Horizontal overflow
Nothing may force sideways scrolling at the width being measured, and anything too rigid to fit a 320px phone is listed for a narrow check rather than passed. Written after our own masthead wordmark refused to shrink below 306px.
Opened accordions
Every disclosure on the page is opened at once and the layout re-measured, because a page that fits closed is not the same as a page that fits open. Each one is put back the way it was found, so running the check does not leave the page rearranged.
Heading order
One h1 per page and no skipped levels, so the outline a screen reader announces matches the one you can see.
Nothing third party before you ask
The case study video loads no YouTube frame until you click it, and its poster is a real button you can reach by keyboard.
Fonts and weights
Declared weights must exist in a face that actually loaded. Written after the whole site spent a morning rendering in a fallback serif.
Not checked, and therefore not claimed
Automated checks find the failures a machine can see. They are roughly a third of the problem, and pretending otherwise is the most common dishonesty in a page like this.
Real assistive technology
Nobody has driven this site with JAWS, NVDA or VoiceOver. Passing a name and role check is not the same as being usable by someone who depends on the announcement.
Keyboard only, end to end
The qualify form now moves focus deliberately: into the first field when it opens, and onto the confirmation when it sends, so your place on the page is never silently reset. What is still untested is a real person completing the whole thing on a keyboard, including a live submission, rather than a scripted check of where focus landed.
Video captions
The Iron Lift case study is hosted on YouTube. We have not verified that its captions are accurate, and automatically generated captions are not a substitute for real ones.
Plain language
We have not tested whether the reading level here works for someone who is tired, rushed, or reading in a second language.
A real 320px phone
Overflow is measured at the widths we can actually render, and rigid elements are listed for confirmation rather than cleared. We have not rendered the site in a true 320px viewport, so WCAG 1.4.10 is a target here and not a result. Text scaling on its own, at 400%, is not separately checked either.
Nothing above is automatic
The browser measurements are run by a person before a change ships, which means a change could ship without them. Automating that needs a browser in the pipeline, and until it is there this section describes a habit, not a control. Saying otherwise would be the easiest sentence on this page to write and the least true.
Tell us what broke
If something on this site kept you from doing what you came to do, we want the specific page and what happened. That is more useful to us than any audit.
Email info@capital-cyber.com or call +1 571 410-3066. If a page is the barrier, tell us and we will give you the same information another way, by phone or by email, while we fix it.