Glossary

WCAG defines testable criteria for accessible content across four principles. What the levels mean, and what automated testing can actually cover.

ARIA adds roles, states and properties where HTML cannot express them. The five rules, and why incorrect ARIA is worse than none.

The accessible name is the label computed for an element from several sources in order. Why an icon button often has none, and how to give it one.

Contrast ratio quantifies the luminance difference between text and background. The WCAG thresholds, what large text means, and what the measure misses.

Landmarks let assistive technology jump between a page's regions. The native elements that provide them, and how to label repeated ones.

A skip link lets a keyboard user jump past navigation to the main content. Where it goes, why it must become visible, and what breaks it.

tabindex decides whether an element is focusable and in what order. Why positive values cause problems, and what -1 is actually for.

Everything interactive must be reachable and operable by keyboard. The expected keys, what a focus trap is, and how to test in five minutes.

A screen reader converts an interface into speech or braille. How it navigates, what it reads from, and why testing with one changes priorities.

Why the focus indicator exists, why it was so often removed, and how the focus-visible selector resolves the conflict that caused the removals.

What a keyboard trap is, why it is one of the most serious accessibility failures, and why no static check can detect one.

How a label is associated with a field, why a placeholder is not a label, and why the association breaks silently in component libraries.

What the autocomplete attribute declares, why it is an accessibility requirement rather than a convenience, and why off is usually the wrong value.

Why a native element carries behaviour no attribute can replace, what is lost when a div is made interactive, and where custom elements are still necessary.

What an OpenAPI document describes, how it differs from written documentation, and why a published specification drifts from the service it describes.

What makes an HTTP API REST-shaped — resources as URLs, methods with fixed meanings, status codes as the result — and what the term has come to mean in practice.

Why an API carries a version, where the version is usually placed, what counts as a breaking change, and what running two versions actually costs.

What the Problem Details format defines, which fields it standardises, and why a stable error shape matters more to clients than a readable error message.

What an environment variable is, why build tools ship some of them to the browser on purpose, and how a private key ends up behind a public prefix.

What a staging environment is for, why staging copies end up publicly reachable and indexed, and what an external scan can see of them.

What an edge function is, how it differs from a CDN rule and from an origin server, and which of its effects are visible in a response.

The viewport meta tag stops a phone rendering at desktop width and scaling down. What each value does, and why user-scalable=no is a failure.

A touch target is the area that responds to a tap. The WCAG minimum, the platform recommendations, and why spacing matters as much as size.

Responsive design serves one document that adapts through fluid layout and media queries. What it replaced, and where it commonly breaks.

Why a phone reporting 390 pixels wide has a screen three times that, how the ratio drives image selection, and where chasing it stops paying.

What a media query tests, why breakpoints should follow the content rather than device sizes, and the preference queries that are accessibility requirements.

What a health or readiness endpoint is for, the difference between the two, and why a detailed public one describes your infrastructure to everyone.

What a status page is for, why it must be hosted away from the infrastructure it reports on, and what makes one trustworthy rather than decorative.

What uptime monitoring measures, why a single check location produces false alarms, and why a 200 response does not mean the page is working.

What synthetic monitoring measures, why its numbers are reproducible and unrepresentative, and when a scripted check beats a real-user measurement.

What real-user monitoring collects, why its results are a distribution rather than a number, and why it can tell you there is a problem but not what caused it.

A tracking pixel records a visit by being requested. How it works without scripts, what the request carries, and how modern tags differ.

What a consent management platform does beyond showing a banner, why the blocking half is the part that matters, and what is observable from outside.

What session replay records, why it captures far more than analytics does, and which parts of a page have to be masked before it is switched on.

How a conversion is recorded and attributed, why a tag fired twice doubles reported revenue, and what an external observation can and cannot see.

What a manifest declares, which fields an install prompt depends on, and why an incomplete one fails without reporting anything.