Accessibility is frequently framed as a legal risk to manage. It is more usefully framed as the difference between a site that works for everyone and one that works for the people who resemble the developer.
Semantic HTML does most of the work
The majority of accessibility problems come from replacing elements that already work. A div with a click handler has no role, no keyboard behaviour and no focus handling; a button has all three for free.
Use headings in order, buttons for actions, links for navigation, and real form elements with associated labels. This resolves a large share of typical audit findings before any ARIA is written.
Keyboard access is the fastest test you can run
Put the mouse down and tab through a page. Every interactive element should be reachable, in a sensible order, with a clearly visible focus indicator, and no trap that cannot be escaped.
Custom dropdowns and modals fail this most often. Modals in particular need focus moved in on open, contained while open, and returned to the trigger on close — none of which happens automatically.
- Never remove focus outlines without replacing them with something visible.
- Maintain a logical tab order; avoid positive tabindex values.
- Trap focus inside modals and restore it on close.
- Make every hover-only interaction available on focus too.
Contrast, motion and text size
Aim for at least 4.5:1 contrast on body text and 3:1 on large text and interface components. Low-contrast grey on grey is a persistent design trend and a persistent accessibility failure.
Respect the reduced-motion preference for anything that animates significantly, and ensure the layout survives a 200% zoom. Both are cheap to implement and both affect real users daily.
Automate what you can, test what you cannot
Automated tooling reliably catches perhaps a third of issues — missing alternative text, contrast failures, absent labels. Put it in CI so those never regress.
The remaining issues need judgement. Is this alternative text meaningful? Does this heading structure describe the page? Does this flow work with a screen reader? Those require a person, and periodically that person should be someone who uses assistive technology daily.
Key takeaways
- Semantic HTML eliminates most findings before ARIA is involved.
- Tab through the page — it is the highest-yield five-minute test.
- Meet 4.5:1 contrast and honour reduced-motion preferences.
- Automation catches roughly a third; the rest needs human judgement.
Ready to make the switch?
Try Zero Delay Analytics free. No credit card required, and no cookie banner needed.
Get Started
