Ten minutes with a keyboard tells you more than any scanner

Unplug the mouse and see what happens

"How bad is it, really?" is usually the question underneath the panic. You can answer a decent chunk of it this afternoon. Unplug the mouse. Put it in a drawer if you have to.

Then try to get through your own site using only the Tab key, Shift+Tab to go backward, and Enter to activate something. No trackpad. No clicking anywhere, for the whole test.

Start on the homepage. Tab through the header, the navigation, any buttons, the footer. Can you reach every link this way? Can you open the mobile menu? Can you complete a form and submit it without ever touching the trackpad?

A surprising number of small sites have at least one control that a mouse can operate but a keyboard simply cannot reach. A "read more" link. A custom dropdown. A slider built without keyboard support in mind. That's a wall for anyone who can't use a mouse at all, not a stylistic gap.

An image carousel is a repeat offender worth testing on purpose. Can you tab to the next slide? Can you stop it from auto-rotating while you're still reading the first one? A carousel that only responds to a swipe or a click leaves a keyboard user stuck watching it cycle past whatever they wanted to read.

Checking that the focus outline is visible on each control

As you tab through, watch for a visible outline around whatever's currently selected. Usually a box or a highlighted border around a link or button. If you can't tell where you are on the page at any given moment, neither can a keyboard user.

That's the whole ballgame for this kind of navigation. Losing your place means losing the page entirely.

Some designs strip that outline out on purpose, because it clashed with the visual style. That's the single most common thing this exercise turns up. It's also one of the cheaper fixes on the list, since it's a style choice rather than a structural rebuild. A developer can usually restore it in an afternoon.

Turning the phone sideways and pushing the text size up

Pick the site up on your phone. Rotate it to landscape and watch whether the layout still makes sense, or whether text starts overlapping buttons. A lot of small-site templates were only ever tested one way up.

Then go into your phone's accessibility settings. Push the text size up two or three notches. Load the site again.

Does anything get cut off? Does anything hide behind another element? Does anything get squeezed unreadable? How contrast and text size hold up under this kind of stress is worth checking in more depth once you've spotted the obvious breaks.

Reading a page with the screen reader already on your computer

Every phone and nearly every laptop ships with a screen reader built in. VoiceOver on Apple devices. TalkBack on Android. Narrator on Windows. None of them cost anything extra, and none of them require installing anything.

Turn it on. Close your eyes if you can stand it. Try to understand your own homepage by listening alone, without looking at the screen at all.

Does it announce what a button does, or just say "button" with no label attached? Does an image get read as a filename like IMG-4021, or does it get skipped entirely? Try jumping by heading instead of reading top to bottom; most screen readers have a shortcut for this, and it shows you fast whether your page even has real headings underneath the visual styling.

A form is one of the harder places to test this way, and checking a form specifically deserves its own pass once you've gotten a feel for the basics.

What an automated scanner catches and what it structurally cannot

Free scanner tools exist, and they're worth running. They flag missing alt text, low contrast, and a handful of structural issues in seconds. That speed is genuinely valuable as a first move.

What they can't tell you is whether the words in that alt text are actually useful. Whether the tab order makes sense to a real person. Whether an error message on a form explains what to fix or just says something failed.

A scanner reads code. A keyboard and a screen reader tell you what a person experiences using that code. Neither replaces the other.

Run the scan first to catch the obvious stuff quickly. Then do the manual pass to catch what code alone can't reveal. Doing both, in that order, takes less time than it sounds like, and it covers ground that either one alone would miss entirely.

Writing findings up so a developer can act on them

Whatever you find, write it down as you go. Not from memory afterward, once the details have blurred together. For each issue, note the page, the specific control or piece of text, and what happened when you tried to use it.

"Tab order skips the search box on the homepage" is something a developer can act on immediately. "Accessibility feels off" is not. Specifics turn a vague worry into a fixable ticket.

A screenshot helps more than a paragraph of description ever will. Circle the exact spot. Note the device and browser you were using at the time, since a bug that shows up on one combination sometimes doesn't show up on another.

Group findings by page rather than by device, since most fixes touch the underlying code once and apply everywhere it's used. If the list gets long enough that you're not sure what's worth fixing first, that's the point where paying for a proper audit starts to make more sense than another afternoon of testing alone, inside what accessibility actually covers for a site this size.

More from the blog

All posts