GAAD: I thought I was doing the work

Today is Global Accessibility Awareness Day, and I had a plan. I was going to open the Ambre repo, pull a list of all the accessibility improvements we’d shipped in 2026, and write a celebratory roundup. A nice little “here’s how we’ve kept accessibility front of mind this year” post. Stand up, wave the flag, do my part for awareness.

Then I actually looked.

The roundup that wasn’t

The commit messages turned up nothing. Not a single one mentioned accessibility, VoiceOver, Dynamic Type, contrast, or any of the words I expected to find. Widening the search to the diffs themselves was a little kinder: around 46 accessibility modifier additions across the year, almost all of them added in passing while I was building something else.

Then I looked at the issue tracker. Zero accessibility tickets opened in 2026. Five from 2025 still open, including “Increase contrast for ingredient and instruction cards background” and “Add support for VoiceOver on macOS”, both filed last June and quietly sliding past their first birthday.

I think of myself as someone who designs with accessibility in mind. When I’m sketching a new screen the considerations are there: contrast, Dynamic Type, hit targets, what the screen sounds like if you can’t see it. The intent is genuinely there.

Some of the work just doesn’t leave a trace

The honest answer to “why didn’t more of it show up in the diffs” is that a lot of it can’t. A title given room to wrap to two lines from day one never needs a “fix truncation at AX3” commit later. A button drawn at 44pt from the first sketch never needs its hit target bumped. Contrast picked carefully in Figma doesn’t produce a “raise contrast” PR. The cleanest accessibility work is invisible to a keyword search, and a fair chunk of the layout decisions in Ambre fall in that bucket: things that just work, because they were designed to.

So the search is unfair to a real category of effort, and I don’t want to write the year off on the basis of one. The prevention work counts.

Where I do fall short

Where it doesn’t credit me is the second pass. Once a feature ships and the visuals are right and the obvious modifiers are in place, the screen leaves my attention. I don’t come back a week later, put VoiceOver on, run it at AX5, toggle Reduce Transparency, and see what survived. That return trip is the one I keep skipping, and it’s the one where most of the real bugs live.

I know this because I did it on one screen this morning, the meal-plan carousel, and found eleven things wrong with it. Eleven. On a screen I designed with accessibility in mind from the start. The intent really was there. It just doesn’t catch everything, and the only way to find out which things it missed is to go back and look.

The uncomfortable bit

The reason GAAD exists in the first place is that almost everyone in our industry has a version of this story. We think of ourselves as people who care, we point at the talk we gave or the post we wrote or the modifier we added, and we don’t notice that the day-to-day pull of building features quietly outruns the day-to-day work of going back and checking them. Awareness is the easy part. The discipline of returning to finished work with fresh eyes is the part that costs something.

I’m not writing this to flagellate. The intent was real, the prevention work is real, and most of the layout choices in the app hold up. What’s missing isn’t the caring, it’s the cadence of going back to check.

What changes after today

Small, concrete things, because grand resolutions are how this happens again next year:

• One screen audit per release cycle, scheduled, not optional. Same checklist every time so it doesn’t depend on me being in the mood.
• VoiceOver + AX3 as a step in our internal QA pass before anything ships, alongside the visual review.
• A single GitHub label for accessibility follow-ups, so the gap between “I noticed it” and “it’s tracked” is one click instead of a vague mental note.

None of that is novel. That’s sort of the point. The work isn’t novel. The work is just showing up to it on a regular cadence instead of once a year on the third Thursday of May.

What I’m taking from today

Two things, really.

The first is that designing with accessibility in mind and auditing for accessibility are two different jobs. I’d been doing the first and quietly assuming it covered the second. It doesn’t, and the gap between them is exactly the size of the audit I didn’t do until this morning.

The second, more useful one, is that GAAD did its job. The whole point of an awareness day is that it pulls you up out of your own narrative and makes you look at the evidence. Mine didn’t match what I’d been telling myself. That’s an uncomfortable thing to find, and it’s also the most useful thing I could have found today.

Happy GAAD. If you were planning to write a celebratory roundup of your year’s accessibility work, I’d gently suggest opening the repo first.