BarrierTrace guide
Automated accessibility testing is evidence. Accessibility assurance is the conclusion you build from it.
A scan can reveal valuable problems in seconds. Responsible assurance asks a larger set of questions about scope, people, journeys, judgement and change over time.
01
Start with the question you need to answer
A team looking for coding regressions needs a different evidence set from an organisation preparing for a formal accessibility audit. Before choosing a tool or interpreting a score, define the decision the evidence will support.
Useful questions include whether a recent release introduced detectable failures, whether important journeys received manual review, whether known issues were retested and whether the organisation can explain the scope and limitations of its work.
02
Understand the automated-testing boundary
Automated accessibility testing applies deterministic rules to the page state it can access. That makes it fast, repeatable and valuable throughout design, development and maintenance. It also means the result is limited to the rules, pages, viewports and states that were actually tested.
Many WCAG requirements depend on purpose, context, interaction or human perception. An automated pass cannot establish that all relevant success criteria conform, and a percentage score can hide which parts of the experience were never assessed.
03
Add manual evidence where software cannot decide
Structured manual checks examine keyboard operation, focus order, screen-reader announcements, content meaning, error recovery, zoom, reflow and end-to-end journeys. Testers should record what they did, what they observed and any environmental details needed to repeat the check.
Expert review is not the same as disabled-user research. Both can be important, and neither should be silently represented by an automated metric.
04
Retain scope and history
Evidence becomes more trustworthy when a reader can see which pages and states were included, who performed manual checks, when the work occurred and whether later changes were retested. A stable representative page set can make monitoring trends easier to interpret.
Retaining earlier results also helps teams distinguish genuine improvement from a changed sample. If scope changes, the record should make that visible rather than quietly rewriting the baseline.
05
Make claims that match the evidence
Automated results can support statements about the rules and pages checked. A broader conformance claim requires suitable evaluation across the relevant content and success criteria, including manual testing. Legal and procurement conclusions may require additional independent advice.
Good accessibility assurance is therefore not a bigger score. It is a clear, reviewable account of what was tested, what was found, what people verified, what changed and what still needs attention.