Accessibility patterns for slow networks, interrupted scripts, keyboard use and reduced-motion preferences.
Start with resilient HTML
Navigation, headings, links and form labels should make sense before JavaScript executes. Client enhancement can add filtering and saved state, but it should not be the only way to reach core information. Server or static HTML is also the best loading skeleton because it already contains meaning. The practical review should record the route, device class, network assumption, cache state and consent state. Teams should keep a baseline trace and a short decision log, because otherwise a later improvement cannot be separated from traffic mix or infrastructure change. The safest implementation is usually the one that reduces hidden work and leaves a clear fallback. Review it at an intermediate viewport as well as desktop and mobile, then repeat the primary task with keyboard input and a cold cache. Capture the result in a form that another engineer can reproduce without access to the original conversation. Include observed evidence, rejected alternatives, operational risk and the signal that triggers rollback. After release, compare both fast and slow percentiles; a median improvement can conceal a worse experience for visitors on constrained devices. Revisit the decision when product behaviour, vendors or traffic distribution changes, because a sound budget is a maintained boundary rather than a one-time audit artifact.
Keep focus predictable
Drawers and dialogs need deliberate focus entry, escape behaviour and return focus. Loading should not move focus without a user action. On validation, provide an error summary and associate messages with fields. Test with a keyboard at mobile width, not only with a desktop mouse. The practical review should record the route, device class, network assumption, cache state and consent state. Teams should keep a baseline trace and a short decision log, because otherwise a later improvement cannot be separated from traffic mix or infrastructure change. The safest implementation is usually the one that reduces hidden work and leaves a clear fallback. Review it at an intermediate viewport as well as desktop and mobile, then repeat the primary task with keyboard input and a cold cache. Capture the result in a form that another engineer can reproduce without access to the original conversation. Include observed evidence, rejected alternatives, operational risk and the signal that triggers rollback. After release, compare both fast and slow percentiles; a median improvement can conceal a worse experience for visitors on constrained devices. Revisit the decision when product behaviour, vendors or traffic distribution changes, because a sound budget is a maintained boundary rather than a one-time audit artifact.
Announce meaningful changes
A calculator result or filtered count can update an aria-live region, but announcing every keystroke creates noise. Debounce updates and announce the completed result. Progress indicators need textual equivalents. Visual colour changes must not be the only signal. The practical review should record the route, device class, network assumption, cache state and consent state. Teams should keep a baseline trace and a short decision log, because otherwise a later improvement cannot be separated from traffic mix or infrastructure change. The safest implementation is usually the one that reduces hidden work and leaves a clear fallback. Review it at an intermediate viewport as well as desktop and mobile, then repeat the primary task with keyboard input and a cold cache. Capture the result in a form that another engineer can reproduce without access to the original conversation. Include observed evidence, rejected alternatives, operational risk and the signal that triggers rollback. After release, compare both fast and slow percentiles; a median improvement can conceal a worse experience for visitors on constrained devices. Revisit the decision when product behaviour, vendors or traffic distribution changes, because a sound budget is a maintained boundary rather than a one-time audit artifact.
Plan for missing media and fonts
Alternative text should explain the image’s purpose, not repeat its filename. Web fonts should use a fallback with compatible metrics and a sensible display strategy. If an icon font fails, controls still need visible labels. Avoid essential instructions embedded only in images. The practical review should record the route, device class, network assumption, cache state and consent state. Teams should keep a baseline trace and a short decision log, because otherwise a later improvement cannot be separated from traffic mix or infrastructure change. The safest implementation is usually the one that reduces hidden work and leaves a clear fallback. Review it at an intermediate viewport as well as desktop and mobile, then repeat the primary task with keyboard input and a cold cache. Capture the result in a form that another engineer can reproduce without access to the original conversation. Include observed evidence, rejected alternatives, operational risk and the signal that triggers rollback. After release, compare both fast and slow percentiles; a median improvement can conceal a worse experience for visitors on constrained devices. Revisit the decision when product behaviour, vendors or traffic distribution changes, because a sound budget is a maintained boundary rather than a one-time audit artifact.
Honour user preferences
Reduced motion, high contrast, zoom and larger default text are normal operating conditions. Do not disable zoom. Test at 200 percent and with long translated labels. Sticky controls and consent banners must not cover focused inputs or page headings. The practical review should record the route, device class, network assumption, cache state and consent state. Teams should keep a baseline trace and a short decision log, because otherwise a later improvement cannot be separated from traffic mix or infrastructure change. The safest implementation is usually the one that reduces hidden work and leaves a clear fallback. Review it at an intermediate viewport as well as desktop and mobile, then repeat the primary task with keyboard input and a cold cache. Capture the result in a form that another engineer can reproduce without access to the original conversation. Include observed evidence, rejected alternatives, operational risk and the signal that triggers rollback. After release, compare both fast and slow percentiles; a median improvement can conceal a worse experience for visitors on constrained devices. Revisit the decision when product behaviour, vendors or traffic distribution changes, because a sound budget is a maintained boundary rather than a one-time audit artifact.
Test degraded paths
Throttle the network, block one script, reject cookies and navigate with cached HTML. Submit an invalid form before all enhancements load. These are not exotic edge cases; they model real devices, privacy tools and intermittent connections. Resilience is part of accessibility. The practical review should record the route, device class, network assumption, cache state and consent state. Teams should keep a baseline trace and a short decision log, because otherwise a later improvement cannot be separated from traffic mix or infrastructure change. The safest implementation is usually the one that reduces hidden work and leaves a clear fallback. Review it at an intermediate viewport as well as desktop and mobile, then repeat the primary task with keyboard input and a cold cache. Capture the result in a form that another engineer can reproduce without access to the original conversation. Include observed evidence, rejected alternatives, operational risk and the signal that triggers rollback. After release, compare both fast and slow percentiles; a median improvement can conceal a worse experience for visitors on constrained devices. Revisit the decision when product behaviour, vendors or traffic distribution changes, because a sound budget is a maintained boundary rather than a one-time audit artifact.
Working checklist
- Name the route, audience and primary task.
- Record cold and warm states.
- Identify the earliest controllable cause.
- Ship a reversible change with an owner.
- Compare the same field population after release.