What it means
A familiar icon can still be ambiguous outside its visual context. Someone listening to a page needs to distinguish opening a basket from removing an item. A short name that describes the action makes that distinction explicit.
Prefer names based on useful visible text when possible. An icon-only control can need an explicit label, but adding an invisible name does not automatically make the wording good. Repeated labels such as 'More' may still leave visitors unsure which item will change.
Hypothetical example: two basket controls
Imagine a shop with a basket icon in the header and a bin icon beside each product. Giving both controls the name 'Basket' leaves their actions unclear. 'View basket' and 'Remove blue notebook' describe different outcomes.
This is an illustration of naming choices, not a result from a tested website. A useful review asks someone to identify each action from its announced name before activating it.
How BLURSOR measures it
In the report: Buttons and links without clear labels
BLURSOR reports controls without a detected name against the number of controls it inspected. When the browser probe runs, it examines visible links, buttons, selected interactive roles, and form controls. Its naming routine considers text, ARIA labels, associated labels, and several fallback attributes.
This routine is a simplified detector, not the browser's complete accessible-name algorithm. It even accepts placeholder text in some cases. If the live probe is unavailable, the checker can estimate from rendered or raw HTML; that method cannot fully resolve label relationships or visibility. Check the reported measurement method before comparing counts.
What to do next
- Ask your developer to inspect the reported controls in the browser's accessibility tree and confirm that their names describe the actual action.
- Use visible, specific wording where practical. Check that icon labels, translations, and repeated controls remain meaningful after content changes.
- Try the important journey with a keyboard and a screen reader; a detected name does not demonstrate that the control works.
What it cannot tell you
- Zero unnamed controls means the detector found a naming signal for the controls it counted. It does not certify accessibility or successful agent interaction.
- Unavailable evidence is not a zero. Hidden states, later interactions, and controls outside the inspected capture may need separate review.
Sources and further reading
The definitions below describe the underlying concepts. The measurement section describes BLURSOR’s own checks.