Automated tools catch a minority of problems. A keyboard test, a zoom test, a screen reader pass, and reading the page without images find considerably more, and none requires expertise.

Why automated tools are not enough

They are useful and they are limited, and knowing the limit prevents a false sense of having checked.

A checker can detect missing alternative text, insufficient contrast, absent labels, and structural errors. Those are real problems worth fixing.

What it cannot detect: whether alternative text describes the image usefully, whether headings reflect the page's actual structure, whether the reading order makes sense, whether something is operable, or whether the content is comprehensible.

Which means a clean automated report means the machine-checkable subset passed, not that the site is usable.

Run them anyway, since they are free and instant. Just do the manual tests as well.

Test one: the keyboard

The most informative test and it takes five minutes.

Put the mouse aside. Use the tab key to move forward, shift and tab to move back, enter and space to activate.

Try to complete the thing your site exists for: reach the contact form, fill it in, and submit it. Then open the menu and reach a service page.

What to watch for: whether you can see where you are at every step, whether anything is unreachable, whether the order matches the visual layout, whether you get stuck somewhere you cannot escape, and whether a menu or overlay traps you.

The focus indicator is the most common failure. Many themes remove the outline showing which element is selected because it was considered untidy, which makes keyboard use impossible rather than merely awkward.

Test two: zoom

Two minutes and it finds layout failures nobody sees.

Increase the browser zoom substantially and read the page. Then narrow the window until it is phone width and do it again.

What to watch for: text overlapping, content cut off, horizontal scrolling appearing, buttons moving off screen, and anything becoming unreachable.

This matters because magnification is how a large number of people with low vision use the web, and it is entirely untested on most small business sites.

The related check is increasing the browser's default text size rather than zooming, which some layouts handle worse.

Test three: listen to it

Uncomfortable, illuminating, and available on every device already.

Screen reading software is built into current operating systems on desktop and mobile. Turn it on, close your eyes or look away, and try to reach the contact form.

You will be slow and clumsy, which is fine. The point is not proficiency; it is hearing what your page actually announces.

What to listen for: images announced as filenames, links announced as click here, headings that do not describe sections, form fields with no label, and buttons announced as button with no indication of function.

Twenty minutes of this changes how somebody writes pages permanently, more than any guidance does.

Test four: read it without the images

The simplest test and it checks something the others do not.

Turn images off in the browser, or read the page as text only.

What to watch for: information that has disappeared entirely, meaning it existed only inside an image. A price list, an opening hours graphic, a menu, or a diagram with text in it.

That content is invisible to screen readers and to search engines, and it is a common failure in food, retail, and service businesses.

What to do with what you find

Prioritise rather than attempting everything.

  1. Anything blocking a task, such as a form that cannot be completed by keyboard.
  2. Anything on the paths customers use, meaning the homepage, the main service pages, and contact.
  3. Missing or useless alternative text on images carrying information.
  4. Contrast failures on body text.
  5. The focus indicator, if it was removed.
  6. Everything else, in time.

A site where somebody can complete an enquiry using a keyboard, read the text at increased size, and hear what the images say has addressed most of what actually excludes people.

Who to involve

Since self-testing has a limit too.

These tests find problems. They do not tell you how somebody who uses assistive technology daily experiences your site, because you are approximating.

Where accessibility matters to your business, whether for legal, contractual, or ethical reasons, testing with people who actually use these tools finds things no checklist produces.

Short of that, running the four tests after every significant change catches the regressions, which is where accessibility is usually lost rather than in the original build, and it builds on the content-level work described in accessibility without the jargon.


Frequently asked questions

Why are automated checkers not enough?

They catch missing alternative text, contrast, and labels, but cannot judge whether alt text is useful, whether reading order makes sense, or whether something is operable.

What is the most informative test?

The keyboard. Put the mouse aside and try to reach and complete the contact form using tab, shift-tab, enter, and space.

What is the most common keyboard failure?

A removed focus indicator. Many themes strip the outline showing the selected element because it looked untidy, which makes keyboard use impossible.

What does the zoom test find?

Text overlapping, content cut off, horizontal scrolling, and buttons moving off screen. Magnification is how many people with low vision use the web.

Do I need special software to hear my site?

No. Screen reading software is built into current desktop and mobile operating systems. Twenty minutes of it changes how you write pages permanently.

What does reading without images reveal?

Information that existed only inside an image, such as a price list or opening hours graphic, which is invisible to screen readers and search engines.

West Coast Media Solutions Inc. provides web design, web development, hosting, digital marketing, and business consulting to organisations across Canada, drawing on more than twenty-five years in the field.

Passed an automated checker and assumed you were fine?

We run the four manual tests, which find the problems that actually stop somebody completing an enquiry.

Start a Conversation