GatekeeperQA

GatekeeperQA / Testing resources

A PRACTICAL TESTER GUIDE

Accessibility testing workflow: scan, review, fix and retest.

A report is useful when another person can reproduce the finding and assess the change. Use this workflow to keep automated evidence and human judgment distinct.

1. Define the pages and states you are testing

Choose representative pages such as a homepage, a contact form, and an important content page. Record whether the scan sees an initial page, an open menu, or a validation error. GatekeeperQA's public scanner checks the initial page state of the URLs you provide. Add other states to your manual test plan rather than assuming a scan covered them.

2. Check execution before counting defects

A page that failed to load is not a clean page. Look at page coverage and errors first. If a scan is incomplete, resolve the limitation and repeat it before drawing conclusions about missing findings. A FAIL quality gate describes the configured thresholds, not a complete legal conformance decision.

3. Separate confirmed findings from unresolved checks

Start with critical and serious findings, while considering how they affect the actual user journey. An unresolved contrast check may involve a background image or gradient: it is a request for human investigation, not an automatic failure or pass.

For example, a synthetic text sample with a measured ratio of 2.8:1 and a reported requirement of 4.5:1 gives a developer a concrete starting point. Verify the actual rendered colors, text size, opacity, and background. Do not invent a ratio when the scanner could not determine it.

4. Make the ticket reproducible

Include the page URL, rule, element location, HTML snippet when available, observed barrier, suggested remediation, and the page state. GatekeeperQA's copy-issue button brings the captured evidence and saved notes into plain text for your issue tracker. Treat suggested remediation as guidance to check in your codebase, not an automatic patch.

5. Record what a person actually tested

Use the manual checklist to plan keyboard, focus, screen-reader, pointer target, dwell-click, on-screen keyboard, zoom/reflow, and page-state checks. Save the reviewer, observed outcome, input method and tool settings, browser zoom, page state, and evidence. Record each input method separately; a keyboard pass does not verify dwell clicking. Follow the manual-review steps and evidence template. A note is not the same as a completed assessment, and the starter checklist is not a full audit methodology.

In a saved report, Open and Needs review help organize unfinished work. Fixed is a reviewer claim; Retest needed makes the verification step visible. None of these statuses changes the original automated result.

6. Retest the same scope

After a change, scan exactly the same page URLs and compare the completed reports. A match at the same selector means the issue is still detected there, not that the page is identical. A changed selector may refer to the same element. If a finding is no longer detected, confirm the element is still present and the relevant state was exercised.

Attach dated retest evidence to the review, then verify manually where required. Do not close a manual check solely because a later automated scan has fewer findings.

Use the accessibility retesting checklist to record the baseline, matching scope, manual outcomes and remaining gaps.

7. Keep a useful handoff

Export the full report for context and a filtered CSV for the current work list. Keep the original findings, reviewer evidence, and retest references together. GatekeeperQA retains the latest 10 account reports for up to 30 days; downloaded copies do not expire with the account history.

Worked example: fix an unnamed button and verify it

This illustrative ticket shows what to record. It is not a claim that a real customer website was tested.

  1. Finding: On a demo search page, the icon-only search button has no accessible name. Record the page state, element location, baseline report and the button-name finding. A screen-reader user needs to know what the button does.
  2. Developer fix: Give the native button a meaningful name. Prefer visible text such as <button type="submit">Search</button>. If the design must remain icon-only, use aria-label="Search" on the button and hide a decorative SVG from assistive technology with aria-hidden="true". Keep any visible label consistent with the accessible name.
  3. Automated retest: Scan the same page and state after the change. Record whether the finding is no longer detected and link both dated reports. Confirm the button still exists; removing the control is not evidence that its task works.
  4. Manual verification: Check that the button receives visible keyboard focus, activates with Enter and Space, and is announced with the expected name and role in the screen reader used for the test. Verify that submitting the search produces the intended result. Record the browser, assistive technology, versions and observed outcome.
  5. Handoff: Close this finding only when its acceptance checks have evidence. Keep untested screen readers, other page states and unrelated manual checks visible as remaining scope. One corrected button does not establish whole-page accessibility.

See the before-and-after demonstration and record the manual testing evidence.

Further reading

Try the scanner

Review scanner coverage and free usage limits · Read a sample report