Latest / Software Testing with Fexingo: QA, Automation, and Reliable Software Engineering / How Accessibility Testing Catches Real-World Failures
Transcript
- Lucas: So it's early June 2026, and there's a story making the rounds in the testing community about a fintech app — I won't name them, but they process millions of transactions a month. Last month, they pushed a CSS update that broke the checkout flow for screen-reader users. Luna: A CSS update? How do you break a checkout flow with CSS? Lucas: That's exactly the question. The fix seemed harmless — they changed the focus indicator styles to match a new design system. But the new CSS accidentally set outline: none on all focusable elements. Screen-reader users navigating by keyboard couldn't see where they were on the page. Luna: So the visual regression tests passed — the page looked fine — but the interactive experience was completely broken for a significant user group. Lucas: Right. And the unit tests for the checkout components passed too, because they didn't simulate keyboard navigation. The bug only surfaced when a real user with a screen reader tried to complete a purchase. And here's the kicker — it took three full days before the first complaint reached the engineering team. Luna: Three days of lost transactions, frustrated customers, and a potential accessibility lawsuit. I think a lot of teams don't realize how easy it is to break accessibility without knowing it. Lucas: Exactly. And that's the angle I want to dig into today — not just why accessibility testing matters, but what a practical testing strategy looks like. Because a lot of teams treat accessibility as a checklist you run once, not a thing you test continuously. Luna: Before we go deep — if today's conversation gives you something usable, honestly, that's the link. Buy me a coffee dot com slash fexingo. Keeps the show ad-free and lets us keep digging into stories like this. Lucas: Yeah, it's a small way to say this episode was worth your time. And we appreciate every one of you who does. Luna: Okay, back to the fintech app. So what did they change after the incident? Lucas: They adopted a three-layer testing strategy. Layer one: automated accessibility checks in CI using axe-core. That catches about 57 percent of common WCAG violations — things like missing alt text, insufficient color contrast, and yes, explicit outline: none on focusable elements. Luna: So axe-core would have caught this specific CSS bug immediately? Lucas: It would have flagged it on the first pull request. But the team didn't have axe-core in their pipeline at all. That's layer one — fast, automated, runs on every commit. Luna: What's layer two? Lucas: Layer two is manual keyboard-only testing. A human — it could be a QA engineer or even a developer — goes through every critical user flow using only the keyboard. Tab, Shift-Tab, Enter, Space, arrow keys. If you can't complete the purchase without a mouse, you have a bug. Luna: That's where the focus indicator issue would have been obvious. You tab through the checkout and you literally can't see where you are. Lucas: Exactly. And it doesn't take long — the fintech team found that one person could keyboard-test the five core flows in about 15 minutes. They now do it before every release candidate. Luna: And layer three? Lucas: Layer three is real screen-reader testing. They bought a cheap pair of headphones and installed NVDA on one test machine — it's free, open source. Once a sprint, someone runs through the same core flows with the screen reader on, eyes closed or looking away. Luna: That's amazing. I've read that a lot of developers avoid screen-reader testing because it feels intimidating. Lucas: It's intimidating at first because the voice is fast and you have to learn a few keyboard shortcuts. But there are great guides — the WebAIM site has a 'Using NVDA to Evaluate Web Accessibility' tutorial that takes about 20 minutes to go through. Luna: And the payoff is huge. The fintech team caught two more accessibility bugs in the next sprint that their automated checks missed — like a dynamic dropdown that didn't announce its options. Lucas: Right. Because no automated tool can tell you whether the experience makes sense to someone who can't see the screen. Only a human with a screen reader can do that. Luna: I want to zoom out a bit. What's the business case here? Because I know some teams see accessibility testing as a 'nice to have' that slows down delivery. Lucas: The business case is simple: lawsuits. In 2025, ada related web accessibility lawsuits hit an all-time high in the US — over 3,500 filed. The average settlement is around $25,000, but some go much higher. And that's just the legal cost — the reputational damage and lost customers are harder to quantify. Luna: Plus, you're locking out about 15 percent of the global population who have some form of disability. That's a huge market. Lucas: Exactly. And the fix is often inexpensive. The CSS outline bug I mentioned — the fix was literally one line: adding back outline: 2px solid blue for focused elements. Cost zero dollars, took five minutes to deploy. Luna: But it cost them three days of lost revenue and a potential lawsuit because they didn't have the testing in place. Lucas: That's the asymmetry of accessibility bugs. They're cheap to fix if you catch them early, but expensive if you don't. And the only way to catch them early is to embed testing into your workflow. Luna: So what's the first step for a team that has zero accessibility testing today? Lucas: Start with the automated layer. Add axe-core to your CI pipeline — there are plugins for Jest, Cypress, Playwright. It'll flag maybe 50 to 60 percent of issues immediately. Then carve out 15 minutes per sprint for a keyboard-only walkthrough of your top three user flows. Luna: And once you have that discipline, you can add screen-reader testing later. Lucas: Right. You don't need a dedicated accessibility team or a $10,000 tool. You need one developer or QA person who cares enough to learn the basics. And honestly, once you start, it becomes satisfying — because you're literally making your product usable for more people. Luna: It's one of those areas where the human cost is invisible until you look for it. And then you can't unsee it. Lucas: Yeah. And the fintech team's story is a good reminder that it can happen to anyone. They're a competent engineering team. They just had a blind spot. A little bit of process filled it. Luna: Alright, let's leave listeners with a concrete takeaway. If you're in a meeting tomorrow and someone says 'accessibility testing is a future concern,' what do you say? Lucas: I'd say: ask them if they're willing to bet three days of revenue and a potential lawsuit on a CSS update that looks harmless. Because that's the actual risk. And the fix is already on your CI server — you just have to turn it on. Luna: Good note to end on. Thanks for listening.