Overview
“Accessibility audit” is an unregulated phrase. It gets applied to everything from a free scanner export to weeks of expert testing. This page shows what a real one contains, finding by finding, so you can evaluate any vendor, including us.
What gets tested
A store audit covers every distinct template, because fixing a template fixes every page built on it:
- Home, collection, product, cart, checkout, account, and search: the purchase path end to end.
- With a keyboard alone, then with a screen reader, at multiple zoom levels and viewport widths.
- Against each applicable WCAG 2.1 AA criterion.
Automation runs first for breadth. Then a human verifies everything, because scanners see only a fraction of the criteria and cannot judge quality at all.
The anatomy of one finding
A real finding looks like this:
Finding 03 · Critical · WCAG 4.1.2 (A)
Button has no accessible name
Product page › Add to cart
A screen reader announces this control only as “button,” so the shopper cannot tell what it does.
Fix: add a visible text label inside the button, or aria-label=“Add to cart” on the element.
Est. 5 min fix.
Five parts, every time: location (exact element, exact page), criterion (the specific WCAG rule), impact (what a real user experiences), severity (rated by risk), and an implementation-ready fix (something a developer can paste, not “ensure buttons are accessible”).
If a report you are evaluating lacks locations or fixes, you are looking at a scanner export with a cover page on it.
How findings get prioritized
Risk-first, not severity-score-first. A blocked checkout outranks a missing footer landmark even if a tool scores them similarly, because the checkout is what complaints demonstrate and what costs you customers every day. A good report reads like a work plan: your developer starts at finding 1 and works down without needing to re-triage anything.
What the document is for
Three jobs:
- A work plan. Your team or agency implements fixes directly from it, in order.
- Evidence. A dated, third-party record of diligence. This is what deters serial filers and anchors settlement negotiations if one files anyway.
- A baseline. The next audit measures drift against it, since stores change constantly.
A report that cannot do all three jobs is not worth commissioning.
Example
A 24-page audit lands for a mid-sized store: 19 findings, 4 critical, 7 serious, 8 minor. The four criticals are all in the purchase path, and the report says so on page one. The store’s developer clears the criticals in one sprint using the included fixes, the serious items the next sprint, and the store keeps the report plus its fix log as the compliance file. Six months later a theme update triggers a re-audit, which takes half the time because the baseline exists.
Common mistake
Buying an “audit” priced too low to include human testing. If the turnaround is instant and the price is under a few hundred dollars, arithmetic says no human walked your checkout with a screen reader. You are paying for a scan you could run yourself for free. The difference is quantified in manual audit vs automated widgets.