Skip to content

U.S. e-commerce accessibility / Case 31

Turning accessibility checks into a dated monthly evidence and regression workflow

The legal follow-up required ongoing evidence, so the accessibility process needed a repeatable monthly record that could distinguish fixed issues, new regressions, manual-review items, and third-party limitations.

AccessibilityDocumentationRegression testingMonthly auditEvidence

What this solved for the business or user

The value of the documentation is that someone can answer: what did we test this month, what did we find, what did we fix, what is still open, and what changed since the last review? That is more useful than saving one screenshot that says '97'.

What was happening

Legal counsel asked for continuing accessibility work to be documented with dates. The site also changed often enough that a clean April result could not be assumed to describe the site in July. Internal records therefore evolved from a homepage remediation summary into repeatable evidence packs and a larger page-by-page report.

Why the obvious solution was not enough

A monthly audit can become performative if it only records scores. It also becomes unmanageable if every run starts from scratch. The workflow needed stable representative URLs, consistent tool roles, a clear issue status model, and change-triggered retesting between scheduled audits.

How the solution works

  1. Keep a stable representative URL list so month-to-month results are comparable, then add temporary URLs when a new template or feature needs coverage.
  2. For each page, retain the tool, device, date, result summary, evidence screenshot/export, and any finding that requires remediation or manual review.
  3. Use issue states such as open, fixed, manual review, third-party, or accepted after review instead of reducing everything to a single score.
  4. Rerun WAVE, axe DevTools, and Lighthouse/PageSpeed after major theme releases, app installs, marketing widgets, or other third-party changes instead of waiting for the next calendar audit.
  5. Preserve manual keyboard and interaction checks as first-class evidence. A resolved scanner error does not replace testing focus order, Escape behavior, hidden controls, zoom/reflow, and dynamic messages.
  6. Separate accessibility findings from performance findings in the report. Lighthouse can provide both, but a performance score is not an accessibility ruling and an accessibility score is not legal certification.
  7. Retain the issue history so a recurring defect can be traced to the theme/app change that reintroduced it.
  8. Keep the report useful to developers and nontechnical stakeholders: executive summary first, detailed findings and remediation next, evidence appendix last.

What should be verified before shipping

  • Use the same naming convention and tested URL set across recurring reports.
  • Keep screenshots/exports dated and associated with the exact page and device tested.
  • Record unresolved manual or third-party items explicitly instead of silently omitting them from the next report.
  • Trigger an out-of-cycle retest after major app, theme, tracking, or interactive-component changes.
  • Make sure the report distinguishes automated evidence, manual evidence, and items still awaiting assistive-technology review.
  • Keep the report understandable to legal/operations stakeholders without turning engineering findings into unsupported compliance promises.

What changed

Accessibility documentation became repeatable evidence of maintenance over time. The April internal summary, May insurance-oriented evidence package, and July expanded report showed the progression from one-time remediation toward a sustained monitoring process.

Reusable lesson

The useful unit of accessibility documentation is not the screenshot; it is the audit cycle. Consistent evidence, issue history, and retesting after change make the record meaningful.

Official documentation and standards