Skip to main content
WCAGrules
Quick navigation

Guides · Testing and tools

How to Test a Website for Accessibility, Yourself, Today

You do not need a budget or a specialist to find most of what is wrong. You need a keyboard, a browser, and about two hours.

Last reviewed August 30, 2026

Every audit we run starts the same way, with checks you can do yourself. We are going to give you the sequence, because a site owner who has done these five passes knows more about their own barriers than a scanner report will ever tell them.

Set aside two hours. No more. Pick your three most important pages, which is usually the home page, one product or service page, and whatever your checkout, booking or contact flow is. Test those properly rather than skimming forty pages. There is a rule underneath that choice, by the way. Where a page is one of a series making up a process, every page in that process has to conform for any of them to count, and the standard's own worked example of a process is a shop checkout.

Pass 1: The Keyboard, Ten Minutes

Put your mouse out of reach. Not aside. Out of reach, so you cannot cheat.

Now load your home page and press Tab repeatedly. Watch what happens:

  • Can you see where you are? Every stop needs a visible outline or highlight. If focus vanishes, you have found a focus visibility failure, and in our own audits it is the one we find more than any other.
  • Does the order make sense? Focus should move roughly the way your eye reads. Jumping from the header to the footer and back is a real failure, not a quirk.
  • Can you reach everything? Menus, sliders, modals, video controls, the cookie banner. Anything you can click, you have to be able to reach.
  • Can you get out of everything? Open a modal, then press Escape. Open a dropdown, then Tab past it. Getting stuck is a keyboard trap, and it is one of only four rules that break the conformance of an entire page, no matter whose code caused it. So a trap inside somebody else's embedded widget is still your page failing.
  • Is there a way past the navigation? Press Tab once on a fresh page load. Most well-built sites reveal a "Skip to content" link. Landmarks and a real heading structure satisfy the same rule, and both of those only help assistive technology users, so a site relying on them alone still leaves a sighted keyboard user tabbing through your whole menu on every page.

Do this pass first, and here is why it earns the top slot. Three of those five checks have no automated coverage at all. Focus order has no automated rule anywhere in W3C's set, and the rules that exist for the other two list a human among their implementations. Ten minutes of tabbing covers ground no scanner reaches.

Try it right now

Tab through your own checkout or contact form before you read further. Most owners find something in the first thirty seconds, and it is usually a control they did not know was unreachable.

Pass 2: Zoom and Reflow, Ten Minutes

Press Ctrl and plus, or Command and plus on a Mac, until the browser reports 200%. Then keep going to 400%.

At 200%, all your content has to still be there and still readable. At 400% on a 1280-pixel window, which is the same thing as a 320-pixel width, every block of text has to be readable without scrolling sideways to read a line. Content does not have to collapse into one column, and maps, diagrams, video, presentations and wide data tables are allowed to keep their own horizontal scroll. What usually breaks:

  • Text clipped inside fixed-height boxes, so the last line is cut off.
  • Sticky headers that grow until they cover most of the screen.
  • Pricing grids forcing sideways scrolling on text that could have wrapped.
  • Two columns of text that never collapse, so reading means scrolling left and right for every line.

Treat what you see at 400% as a page rather than as a stress test. Conformance counts every variation a page presents at different screen sizes separately, so the zoomed layout has to stand on its own rather than being covered by the desktop one having passed.

Pass 3: Turn Off the Pictures, Five Minutes

This one is a shortcut to judging alt text without learning a screen reader. Use a browser extension or your developer tools to disable images, then reload.

Does the page still make sense? Do the alt texts tell you what the images told you, or do you get "IMG_4471.jpg" and "banner-final-v2"? Our alt text checker catches the classic failures in bulk if you would rather paste text than squint at a page.

Know what this pass does not show you, because it is a rougher proxy than it looks. Turning images off shows the alt attribute where the browser chooses to render it. It shows you nothing about a name supplied by ARIA, a caption on a figure, a background image carrying meaning in CSS, or the name on an inline SVG. It also cannot tell a deliberately empty alt from a missing one, and those are opposite failures that look identical here. Worth five minutes anyway, because filenames and empty alt on meaningful images are two of the commonest failures on the web.

Pass 4: Listen to It, Thirty Minutes

This is the pass that changes how owners think about their own site, so do not skip it because it sounds hard. It is not.

On a Mac, a screen reader is already installed and you start it with Command, F5. On Windows, download NVDA, which is free and open source. On either, turn your monitor brightness to zero, or close your eyes, and try to complete one task on your own site. Buy the thing. Book the appointment. Send the message.

You will be slow and clumsy, because you do not use one daily. That is fine. You are not testing your skill, you are listening for these:

  • Buttons announced as "button" with no name.
  • Form fields announced as "edit text, blank" with no label.
  • Links announced as "click here" or "read more", stripped of the sentence around them.
  • Images announced as filenames.
  • Error messages that appear visually and are never announced.
  • Anything announced in an order that does not match what is on screen.
  • A CAPTCHA you cannot get past, which screen reader users report as their single most problematic item on the web.

Pass 5: The Five-Minute Sweep Nobody Does

Three quick checks that fall outside the other four passes, and one of them covers the only guideline with no automated coverage anywhere.

  1. Read your page title. Look at the browser tab. Does it say what the page is, and is it different from every other tab? A descriptive page title is one of the four things in the whole standard that a machine can settle outright, so this one takes five seconds and then never needs your attention again.
  2. Watch anything that moves. A carousel, a background video, an animated hero, an advert. Does it stop by itself within five seconds, and if not, is there a control to stop it? Flashing content is the other half of this, and no scanner on earth will tell you about either, because the guideline covering seizures has no automated rules at all.
  3. Check your video and audio. Captions on anything with speech, and a transcript where you can manage one. This is a whole guideline the first four passes never touch.

Then Let a Machine Do the Counting

Five passes done, and you now know what your site feels like. A scanner is good at the part you are bad at, which is checking every page for the mechanical failures, consistently, without getting tired or distracted.

Our free 10-page scan runs your site in a real browser against all 90 supported automated rules, names the source rule behind every finding, adds a direct fix guide where our library has a verified match, and keeps its 27 best-practice checks separate from WCAG findings. Free browser tools such as WAVE, axe DevTools and Lighthouse do a similar job one page at a time, and our free tools comparison covers what each is good for.

The one that survives every pass on this page

Watch out for a form field whose only label is placeholder text. It passes the scan, because the specification that computes accessible names accepts a placeholder as a fallback. It reads fine with images off. And in the listening pass it announces the placeholder, so it never says "edit text, blank". Then the label disappears the moment somebody starts typing, and they have no way to check what the field wanted. It is the most likely real failure to survive your own two-hour test, so go and look at your forms for it specifically.

One honest limit

These five passes plus a scan will find most of what is wrong on a typical site. They will not find everything, because you are not a daily screen reader user and neither are we. That is why our audit puts a professional blind screen-reader user on your key journeys. They find the barriers that only surface when somebody who depends on the technology is trying to get something done.

Common questions

How do I test my website for accessibility for free?
Do five passes yourself. Tab through it with no mouse, zoom to 400%, disable images to judge alt text, listen to it with VoiceOver or NVDA, and check your page title, your moving content and your video captions. Then run a scanner for the mechanical failures. Every step costs nothing.
How long does it take to test a website?
About two hours for three important pages done properly, on our own estimate rather than anybody's research. A full audit of a whole site takes considerably longer, because every template, form and interactive component needs testing separately.
Do I need to know how to use a screen reader?
No. You need to be able to start one and listen. You will be slow, and that does not matter, because you are listening for missing names, missing labels and things announced in the wrong order. All of those are obvious to a beginner.
Can automated tools test everything?
No, and the gap is wider than most people expect. By our own count, 10 of the 356 WCAG techniques and failures we graded are fully machine-detectable, which is 2.8%. Of the standard's 86 criteria, 49 have no automated rule written against them at all. A tool can tell you an image has no alt attribute. It cannot tell you the alt text is wrong.
What will my own testing definitely miss?
Anything that needs a daily user's habits rather than a beginner's. It will also miss a placeholder-only form field, which passes the scan, announces fine in a listening pass, and still fails. And it will miss whatever happens in states you never opened, since a scan only sees the page as it currently sits.

Sources

Keep reading

More on testing and tools

Reading about it is the cheap part.

Find out where your site actually stands. The free scan checks 10 pages in a real browser against all 90 supported automated rules, keeps its 27 best-practice checks separate from WCAG findings, and names the rule behind every finding. The full audit adds an expert review and a real blind screen-reader user. From $499, with the report in 5 business days on Rapid and 10 on Standard, and the clock starting at cleared payment.

Go somewhere useful

Find tools, resources and your workspace.

29 destinations