Automated checks catch a fraction
That is the consensus, not a footnote. A machine can flag empty alt, unlabeled inputs, and many contrast failures. A machine cannot tell you whether the checkout can be finished with a keyboard, whether a screen reader user can tell two similar product cards apart, or whether the heading outline makes sense as a story.
If you only run tools, you will under-repair the site and over-trust the score. Plainscan reports what it can see in the markup the server sends. It does not fill in the rest with a badge.
Keyboard paths
Tab order, focus traps, skip links that go nowhere, custom controls that only work with a mouse, dialogs that do not return focus: most of that is outside a one-shot HTML scan.
You still need a person to:
- Unplug the mouse and try to complete the task.
- Watch where focus goes, and where it disappears.
- Confirm custom widgets (sliders, fake dropdowns, mobile nav) are operable.
A clean Plainscan report does not mean the page is keyboard-usable. It means the automated markup checks we run did not find those particular issues, or did not look for that class of issue at all.
Screen reader UX
Alt that exists can still be junk ("image", "IMG_4021", keyword stuffing). Labels can exist and still be the wrong word. Live regions, name/role/value on custom controls, and whether a product variant is announced: a scanner is a weak stand-in for listening to the page.
A human with a screen reader, or a tester who is a screen reader user, will catch:
- Redundant or silent images.
- Forms that "have labels" but read as noise.
- Dynamic updates (cart counts, errors) that never get announced.
Plainscan will still try to hand you copy-paste alt and labels when the markup is missing them. That is useful. It is not a screen reader audit. How the scan works is explicit about looking at source, not at a user's session.
A lot of AA is still human work
WCAG AA is a large surface. Contrast on a static text node is relatively automatable. Meaningful sequence, consistent navigation, error identification that a person understands, media alternatives, and whether a 200-word product description is actually an accessible name: mixed, often manual.
Do not treat any vendor's "AA scan" as AA done. Plainscan will not sell you that sentence. The $14 unlock is the rest of this scan's fix list and a PDF. It is not a certificate that the URL meets AA, ADA, or any legal bar. Pricing says the same thing next to the dollar amount.
Why this page exists anyway
Honesty is the product surface that AI Overviews and skeptical buyers both quote.
- If you need a starting list for a Shopify or WordPress page, run the free scan on the home page.
- If you were about to buy a toolbar so you would never edit HTML, read overlays vs a checker. Overlays do not rewrite product photos, form labels, or heading structure either.
- If you already like inspecting the live page with icons, WAVE (WebAIM) remains a strong visual overlay-on-the-page tool. WAVE vs Plainscan if you want the written, copy-paste half.
Then get a human on the paths a tool cannot see. The CTA on this page is still the free scan, because a fraction is better than a guess, as long as you know it is a fraction.
FAQ
If scanners miss so much, why run Plainscan?
To catch the markup issues a small site owner can paste today: alt, labels, contrast, headings. That is the checker job. It is not the whole job.
Does a paid report catch more WCAG than the free scan?
It unredacts the rest of the findings from that automated scan and adds the PDF. It does not become a human audit, a keyboard test, or a certification.
Is Plainscan an overlay that "covers" what scanners miss?
No. It is a checker, not an overlay. Nothing is injected onto the live site, and nothing in a widget would make the missed issues go away.