In one paragraph
Placeholder used instead of label means the source check found fillable fields without the labelling mechanisms it recognises. It does not actually inspect whether a placeholder is present. Review the real form and provide understandable persistent questions and correct accessible names. A placeholder can supply a format example, but disappearing text and accepted empty attributes do not establish a well-labelled usable field.
Why it matters
Once users type or autofill supplies a value, placeholder text can disappear. If that was the only visible question, reviewing or correcting an entry can become difficult. Appropriate labels and instructions support both initial completion and later checking.
Accessible name and visible guidance are related but not identical. A correctly named compact control can still need more visible context, while a visually adjacent question may lack a programmatic association. Evaluate the task and applicable requirements rather than relying on one attribute.
The word placeholder in this finding is a hypothesis about a common pattern. Some flagged fields may have no hint at all, or may be labelled through a mechanism outside the detector. Conversely, an accepted but empty ARIA value can conceal a genuine naming problem.
How WebAuditOS evaluates this finding
The supplied rule inspects eligible fillable fields inside served-HTML forms, excluding specified non-fillable input types. It accepts the presence of native label associations or named ARIA/title attributes and warns informationally when one or more fields lack its recognised sources.
It does not read placeholder, validate non-empty meaningful computed names, inspect CSS visibility, cover fields outside forms, or discover runtime-created forms. No eligible form or unparseable content can be ungraded. Its pass is not a WCAG labelling assessment.
How to fix it
- Inspect the actual rendered fields, their visible questions, computed accessible names, and any help or error content. Test both empty and autofilled states.
- Prefer an associated visible native label for ordinary fields. For example:
<label for="contact-email">Email address</label>
<input id="contact-email" name="email" type="email" autocomplete="email" placeholder="[email protected]">Here the placeholder is supplementary format text rather than the sole question. The form still needs real validation and appropriate processing.
- Keep IDs unique and match for to the intended control. Repeated address or account forms can otherwise connect one question to another field.
- Separate additional instructions through suitable descriptions. Required state, expected format, and validation errors need coherent communication without replacing the basic field name.
- Review ARIA-only or visually hidden naming according to the actual interface. Those can be appropriate in some contexts, but do not automatically solve the disappearing visible-question problem.
- Test floating labels carefully. The label should remain readable and associated after typing, autofill, focus loss, and errors, not simply animate into an inaccessible or low-contrast position.
- Include native selects, checkboxes, radios, and dynamically added fields in manual review, even where the source heuristic misses some implementation. A select's first option is not a persistent question label.
- Rescan this UX rule and the relevant accessibility name checks. Inspect empty accepted attributes, stale references, and label quality rather than closing the task on a presence-only pass.
What the fix can break
Label additions can change layout and wrapping. Preserve readability at narrow widths instead of hiding the label again to keep a compact screenshot.
An ARIA override can replace a useful visible question with different wording. Verify computed precedence and consistency for voice interaction.
Do not remove placeholders indiscriminately if they provide useful examples. Keep them as optional supplementary guidance while fixing the actual naming and persistence problem.
Questions
Does this finding prove a placeholder was detected?
No. The supplied rule does not read it; it identifies missing recognised label sources.
Can a pass coexist with no meaningful name?
Yes. The heuristic can accept empty attributes or other ineffective sources. Review actual computed names.
Must all fields use exactly the same visual pattern?
No. Choose a clear usable design for the task, with appropriate names and instructions. Visible associated labels are a robust ordinary default.
Which related checks help?
Review form labels, autocomplete, and the automation limits guide.