Here’s a two-minute experiment that will teach you more about accessibility testing than most sales pitches. Open your website and put your mouse away. Now try to do the one thing your visitors actually come to do, fill out the contact form, find the meeting agenda, pay the bill, using only the Tab, Enter, and arrow keys.
Could you see where you were on the page at every step? Did you get stuck in a menu? Could you even reach the submit button?
Whatever you just found, here’s the interesting part: an automated accessibility scanner probably wouldn’t have caught it. That gap, between what software can detect and what a human experiences, is exactly what this article is about.
The Short Version
Automated scanners are useful, fast, and cheap, and they catch roughly a third of real accessibility problems. A manual website accessibility audit, done by a person with real assistive technology, finds the rest. You need both, but if you only ever run scans, you will believe your site is far more accessible than it actually is. That false confidence is what gets organizations in trouble.
What Automated Scanners Do Well
An automated checker reads your page’s code and flags patterns it can measure. It’s genuinely good at that:
- Missing alt text: it can tell an image has no description
- Color contrast: it can do the math on text vs. background
- Code-level errors: form fields with no label attached, broken heading structure, missing page language
And it does all this in seconds, for free or nearly free, as often as you like. That’s real value, it’s why the first step we recommend to anyone is a quick scan. (You can run one free in our web accessibility workspace right now.)
What Scanners Miss, and Why It’s the Important Third
A scanner can check whether something exists. It can’t judge whether it works or makes sense. Consider:
- Alt text that says nothing. A photo described as “image_47.jpg” or “photo” passes the scan, alt text exists! To a blind visitor it’s useless. Only a human can judge whether a description actually describes.
- Keyboard traps. A menu or pop-up that swallows keyboard focus and won’t let go is one of the most severe failures a site can have. Scanners almost never catch it, because finding it requires actually navigating.
- Reading order. A screen reader might announce your page in an order that makes no sense, price before product, error message after the form. The code can be technically valid and the experience still be gibberish.
- Forms that fail in practice. The label exists (scan passes), but the error message appears somewhere a screen reader never announces, so the user has no idea why their submission failed.
- Focus you can’t see. Keyboard users need to see where they are on the page. Whether the focus indicator is actually visible against your design is a human judgment.
Notice a theme: the failures scanners miss tend to be the ones that completely block a real person from finishing a task. A missing alt tag on a decorative flourish is a footnote. A keyboard trap in your payment form means someone literally cannot pay you. If you’ve read our piece on why automated checkers miss the most important issues, this will sound familiar, it’s the pattern we see in every audit.
What a Manual Audit Actually Involves
“Manual testing” doesn’t mean someone eyeballs your homepage. A proper manual website accessibility audit is a structured process:
- Screen reader testing. A specialist navigates your key pages with the same software blind users rely on, NVDA, JAWS, VoiceOver, and documents where the experience breaks.
- Keyboard-only testing. Every menu, form, pop-up, and widget, no mouse allowed. This is where traps and invisible focus get caught.
- Task-based walkthroughs. Not just “does the page pass,” but “can a person actually register for the program, download the form, complete the checkout?”
- Judgment calls a machine can’t make. Does the alt text convey what matters? Are the link names meaningful? Is the video caption accurate?
- A prioritized report. Issues ranked by user impact and mapped to WCAG, so your team fixes the blockers first instead of wading through 400 undifferentiated scanner warnings.
That last point deserves a second look. Scanner output is a haystack, hundreds of flags, many trivial, some false alarms. An auditor hands you a short list sorted by “this is stopping real people today.” For a team with limited time, that prioritization is often worth more than the findings themselves.
So Which Do You Need?
Both, but for different jobs.
Use automated scanning as your smoke alarm: run it regularly, catch regressions early, keep new content honest. It’s free and instant, so there’s no reason not to.
Use a manual audit when the answer actually matters: before a compliance deadline, after a complaint or demand letter, ahead of a redesign, or any time you need to say “our site meets WCAG 2.1 AA” and be right. Legal standards are written about real usability, and only human testing tells you where you truly stand. It’s also what the DOJ’s Title II web rule effectively demands of government sites, a scanner “pass” won’t save a site a screen reader user can’t operate.
Common Questions
How much of WCAG can automated tools actually test?
Studies and industry experience put it around 30–40 percent of WCAG success criteria at best, and the automatable ones skew toward lower-impact issues. The majority of criteria require human judgment.
Our scanner gives us a score of 95. Are we compliant?
A high scan score means you’ve cleared the machine-detectable issues, a good sign, but it says nothing about keyboard traps, reading order, or whether your forms work with a screen reader. Compliance is a claim about the whole standard, and no scanner can verify the whole standard.
What does a manual audit cost?
It depends on the size and complexity of your site, we break the factors down honestly in our website accessibility audit pricing guide. Every audit is scoped and quoted up front.
Can’t AI do manual testing now?
AI tools are getting better at flagging possible issues, but the core problem remains: judging whether a real person can complete a real task with assistive technology. That’s experiential, not pattern-matching. Treat AI checkers as smarter scanners, not as a replacement for human testing, and be wary of any product promising otherwise (we’ve written about why overlay shortcuts don’t hold up).
Start With the Free Scan, Then Get the Real Picture
Do the two-minute keyboard test from the top of this article, then run your site through our free workspace to see the machine-detectable issues. When you’re ready to know where you actually stand, our manual website accessibility audit covers everything the scanners can’t, real screen reader testing, prioritized findings, and a clear path to remediation. Request a free review and we’ll tell you plainly what your site needs.
