Skip to content
Sentinela

What a free accessibility scan finds, and what it can't

Published · 5 min read

A free accessibility scan is a good first step: you enter your store's address, and a few minutes later you have a list of issues and a score. It is also easy to read too much into the result, in both directions. This article explains what an automated scan does, what it finds well, what it cannot judge, and how to read the score you get.

What an automated scan does

A scanner opens your pages in a real browser, waits for them to load, and runs a set of rules against what it sees. Many tools, including ours, use axe-core, an open-source rules engine. Each rule checks something specific (does this image have a text alternative, does this button have a name, is this text dark enough on this background) and is mapped to a success criterion in the Web Content Accessibility Guidelines (WCAG).

The Sentinela free scan tests your home page and two more pages it picks, preferring a product page and the cart, on a desktop-sized screen, with axe-core. You get a score, the most frequent issues with the WCAG criterion each relates to, and a suggested fix.

What it finds well

Automated rules are good at things that can be measured in the page's code:

  • Missing names and labels: icon buttons with no name, form fields with no label, links with no text.
  • Missing text alternatives: images without an alt attribute.
  • Color contrast: text that is too light for its background, measured to the exact ratio.
  • Page setup: a missing page language, a page without a title, zoom disabled in the viewport tag.
  • Some structure: invalid ARIA, IDs referenced by ARIA that aren't unique, list and table markup used incorrectly.

These are also among the most common issues on the web. The WebAIM Million 2026 report found the six most common detected failure types on the top million home pages were low-contrast text, missing alt text, missing form labels, empty links, empty buttons and a missing page language.

What it can't judge

Many criteria need a person, because they depend on meaning or on how a page behaves over several steps:

  • Whether alt text is accurate. A scanner can see that an image has alt text; it cannot tell whether "IMG_2041" or "product photo" describes a linen apron.
  • Whether a keyboard journey works. Can you open the menu, choose a size, add to cart and reach checkout with the keyboard alone, and can you always see where focus is?
  • Reading order and meaning. Does the page make sense when read aloud in order? Do headings describe their sections?
  • Captions and media. A tool can see a video; it cannot check that the captions are right.
  • Helpful errors. When a form field is wrong, does the message say how to fix it?

"No issues detected by automated checks" means the checks found nothing. It does not show that a criterion is met.

How much do automated tools find?

You will see different numbers quoted, and they measure different things:

  • Deque's study. Deque, the company behind axe-core, published an analysis in March 2021 of more than 2,000 of its own first-time audits (over 13,000 pages). It found that, on average, 57% of the issues found in those audits were covered by its automated testing, counted by number of issues, not by WCAG criteria. It is a vendor's study of its own tools on its own audit data.
  • The UK Government Digital Service test. In February 2017 GDS built a page with 143 deliberate barriers and ran ten tools against it. The best results were 37% (one tool, counting errors and warnings) and 41% (another, counting prompts for manual checks as well), and 42 barriers (29%) were missed by every tool. It was one artificial page, not real sites.

Both lead to the same practical point: automated testing finds many common issues, and some criteria need a person. A free scan is not an audit, and a clean scan is a good sign, not a clean bill of health.

Reading the score

The Sentinela score summarizes what the automated checks found, on a scale from 0 to 100, so you can follow a site over time. Each page starts at 100 and each distinct issue subtracts a weight based on its severity; the site score is the average of its pages. Our method page explains the details.

Two things the score is not:

  • It is not a percentage of WCAG conformance. A score of 90 does not mean "90% accessible".
  • It is not a pass mark. A high score means the automated checks found few issues on the pages tested, not that the store is free of barriers.

The most useful way to read it is as a trend: is it going up as you fix things, and does it drop after a theme or app update?

When to go further

A free scan covers three pages with automated checks. Go further when:

  • You're about to redesign or change themes, and want to know what to fix in the new one.
  • You need to know about the whole journey, from product to checkout, with a keyboard and a screen reader.
  • You want a dated record of your accessibility work.

An audit adds a person who reviews every automated finding, tests by hand with a keyboard, screen readers and zoom, and gives every WCAG 2.2 A and AA criterion a verdict. After fixes, regular monitoring catches new issues that automated tests can find.

Sources

This article is general information, not legal advice. For your obligations, talk to a lawyer.