Accessibility Wins You Can Ship This Afternoon

The most damaging idea in accessibility is that it’s all or nothing. A founder hears the word, imagines a six-month remediation project with consultants and a legal review, winces at the size of it, and does nothing. The all-or-nothing framing turns a gradient into a cliff, and people don’t step off cliffs. But accessibility isn’t a cliff. Every barrier you remove helps someone today, whether or not the barrier next to it still stands. Which means the interesting question isn’t how to fix everything. It’s what you could fix before dinner.

Quite a lot, it turns out, because the most common defects on the web are also the cheapest to repair. Start with alt text, the written alternative for images. An image without it is a blank spot for a blind user; a screen reader either skips it or, worse, reads the filename aloud. The fix is a sentence per image describing what it shows, and for purely decorative images, an explicitly empty alt attribute, which tells the screen reader to say nothing rather than guess. An afternoon covers every image on most marketing sites, and while you’re in there, you’ll discover which images actually carry meaning, which is a useful thing to learn about your own pages.

Next, form labels. Many forms rely on placeholder text, the gray hint inside the field, as the only indication of what the field wants. Placeholders disappear when you type, aren’t reliably announced by assistive technology, and are usually rendered in exactly the low-contrast gray that’s hardest to read. The fix is a visible label element associated with each input, which takes minutes per form and helps every user, since nobody enjoys deleting what they typed to re-check what the field was asking for.

Then contrast. Fashionable gray text on white backgrounds routinely falls below the level at which people with ordinary aging eyes, let alone low vision, can read comfortably. Contrast checkers are free and instant: put in your text and background colors and you get a pass or fail against WCAG’s ratios. Darkening body text costs one line of CSS and, in my experience, no designer defends the old gray once they see the two side by side.

Two more fit before the afternoon ends. First, the page language: a single attribute on the html element telling the browser the page is in English, or whatever it’s in, so screen readers pronounce it with the right rules instead of mangling it with the wrong ones. That’s one line, once. Second, focus visibility: somewhere in many stylesheets lurks a rule suppressing the browser’s focus outline because someone found it ugly. Deleting that rule, or replacing the outline with a styled one, restores the only way keyboard users can tell where they are on the page. Grepping your CSS for it takes a minute.

Notice what these have in common. None requires a redesign. None requires touching product logic. None requires expertise beyond an afternoon of reading. Each addresses a defect that automated scanners, GazeSite’s among them, flag constantly, and each sits at the top of the list precisely because it’s both frequent and mechanical. This is unusual. In most kinds of engineering work, the common problems and the easy problems are different sets. In accessibility, for historical and slightly embarrassing reasons, they overlap heavily, which means the return on the first afternoon is the highest return you will ever get on the subject.

I should be honest about what an afternoon doesn’t buy. It doesn’t make a custom date picker keyboard-operable, untangle a focus trap in your checkout modal, or restructure a page whose headings lie about its contents. Those are real projects with real costs, and pretending otherwise would be the same all-or-nothing fallacy in reverse. But there’s a compounding effect the quick wins set off that I think is their real value. A team that has shipped five accessibility fixes has stopped being a team that doesn’t do accessibility. The subject moves from the category of looming obligations to the category of normal work, and normal work gets done. The next pull request gets a label on its input because the person writing it just spent an afternoon adding labels and now notices their absence. Habits form around actions, not intentions, and small actions are the only kind you can take today.

So here’s the whole proposal. Block out one afternoon. Run any automated checker on your five most important pages. Fix the alt text, the labels, the contrast, the language attribute, the focus outline. When you’re done, some number of people who couldn’t read your site this morning can read it tonight. It won’t be everyone. It doesn’t have to be. The perfect site ships never; the slightly better one ships at five o’clock.