Latest / Software Testing with Fexingo: QA, Automation, and Reliable Software Engineering / How Accessibility Testing Improves Software for Everyone
Transcript
- Lucas: So there's a stat I keep coming back to: a major e-commerce site — I won't name them, but you'd know the brand — lost an estimated two million dollars in a single quarter because screen reader users couldn't complete checkout. The 'Place Order' button wasn't labeled properly for assistive tech. Button was there, visually looked fine, but a blind user navigating with a screen reader heard nothing when they reached it. Luna: Two million dollars just from a missing label? That's wild. And it's not like they were a small shop — they had a full QA team. Lucas: Right. And that's exactly why I think accessibility testing deserves more attention in the QA world. It's not just about checking a compliance box for lawsuits. It uncovers real functional bugs that affect a significant portion of users. Globally, over a billion people have some form of disability. That's a huge market, and if your software doesn't work for them, you're leaving money on the table. Luna: Yeah, and I think a lot of teams treat accessibility as an afterthought — something you do at the end, maybe run a tool like axe or WAVE, fix a few color contrast issues, and call it done. Lucas: Exactly. And those automated tools are useful, but they only catch about twenty to thirty percent of accessibility issues. They can flag missing alt text, insufficient color contrast, or missing form labels. But they can't test whether a screen reader user can actually complete a multi-step workflow, or whether a keyboard-only user can navigate a dropdown menu. Luna: So you need manual testing with actual assistive technology. Have you ever tried using a screen reader for a day? I did once for an accessibility workshop. It was humbling. Lucas: I have, and yeah, it completely changes how you think about your UI. You realize that things you assumed were obvious — like the order of headings, or how focus moves between elements — are completely invisible to someone relying on audio. That's why the Web Content Accessibility Guidelines, or WCAG, have four core principles: perceivable, operable, understandable, and robust. Each maps to real user needs. Luna: And those guidelines are organized by levels: A, AA, AAA. Most legal requirements target AA. But even hitting AA doesn't guarantee a great experience. Lucas: No, it doesn't. AA covers the basics — like captions for videos, proper heading structure, sufficient color contrast. But AAA includes things like sign language for all prerecorded audio, which is a heavy lift. And truthfully, many teams struggle to consistently meet AA. The key is to embed accessibility testing early, not as a final gate. Luna: You know, this reminds me why we keep this show ad-free. We don't have sponsors telling us what to cover, so we can dig into topics like this that actually help people build better software. If you find value in conversations like this, you can support the show at buy me a coffee dot com slash fexingo. It's a simple way to keep us independent. Lucas: Absolutely. We've heard from listeners who've used our episodes to pitch accessibility initiatives at their own companies. That's exactly the kind of impact we want. So if today's conversation gave you something usable, consider tossing a coffee our way. Luna: Alright, back to the testing side. So you mentioned automated tools catch only about thirty percent. What do they actually miss? Lucas: Great question. Automated tools are good at checking syntax and static attributes. For example, they can verify that every image has an alt attribute. But they can't tell you if the alt text is meaningful. If you have an image of a chart showing sales trends, alt text that says 'chart' is technically present but useless. A screen reader user hears 'chart' and learns nothing. The tool passes, but the user fails. Luna: Right. So that's a case where manual review is irreplaceable. You need a human to judge context. Lucas: Exactly. Another thing tools miss is keyboard trap — where a user tabs into a widget, say a calendar date picker, and can't tab out because the focus is stuck in a custom control. Automated tools rarely detect that. You have to actually navigate with a keyboard to find it. Luna: And what about dynamic content? Like single-page apps where content updates without a page reload? Lucas: That's a big one. If a new section appears after a user clicks a button, a screen reader might not announce it unless you've used ARIA live regions. Automated tools can check if the ARIA attribute is present, but they can't verify that the announcement happens at the right time or with the right content. That requires manual testing. Luna: ARIA — Accessible Rich Internet Applications — that's a whole other layer of complexity. And misusing ARIA can actually make things worse, right? Lucas: Absolutely. There's a saying in the accessibility community: 'No ARIA is better than bad ARIA.' If you add an ARIA role that conflicts with the native semantics, a screen reader might get confusing or duplicate information. For example, putting role equals 'button' on a div that already is a button — that creates confusion. So you need to test thoroughly. Luna: So practically, how do you integrate accessibility testing into a CI pipeline without slowing everything down? Lucas: You start with automated checks in the build. Tools like axe-core or Lighthouse CI can run as part of your test suite and fail the build if new violations appear. That catches low-hanging fruit early. But you also need a separate, scheduled manual audit cycle — say, once per sprint — where someone tests key user flows with a real screen reader and keyboard. Luna: And you'd want to test with multiple assistive technologies, right? Because JAWS and NVDA and VoiceOver behave differently. Lucas: Exactly. Each screen reader has its own quirks. VoiceOver on Mac handles certain ARIA patterns differently than NVDA on Windows. If you only test with one, you might miss bugs that affect users of the other. I've seen cases where a custom dropdown worked perfectly in VoiceOver but was completely unusable in JAWS because of a missing ARIA attribute. Luna: So that adds time and effort. But the flip side is that fixing accessibility bugs often improves the experience for everyone. Curb cuts and automatic doors were designed for wheelchairs, but everyone benefits. Lucas: Exactly. High color contrast helps users in bright sunlight. Captions help people in noisy environments. Keyboard shortcuts help power users. Accessibility improvements tend to make the product more robust overall. And from a testing perspective, catching these issues early is cheaper than fixing them after release. Luna: I want to come back to the business case, because I think that's what convinces stakeholders. What's a good way to estimate the ROI of accessibility testing? Lucas: There are a few approaches. One is to calculate the size of the disability market — over a billion people globally, with an estimated spending power of over eight trillion dollars. Even capturing a fraction of that is significant. Another is to look at legal risk: in the US, ada related website lawsuits have been rising steadily. Settlements can range from tens of thousands to over a million dollars. Luna: And then there's the reputational damage. If a major brand gets sued for inaccessibility, that story spreads fast. Lucas: Right. But I think the strongest argument is the one I started with: the two-million-dollar checkout bug. That wasn't a lawsuit — that was lost revenue from users who literally couldn't give them money. That's a direct line to the bottom line. And it was caused by something as simple as a missing button label. Luna: So for teams listening who want to start improving their accessibility testing, what's the first step? Lucas: I'd say start with an audit of your top five user flows. Use an automated tool to get a baseline, then manually test those flows with a screen reader and keyboard only. Document the issues, prioritize them by severity, and fix the ones that block users from completing core tasks. Then add automated checks to your CI so new code doesn't reintroduce those same issues. Luna: And don't forget to train your team. I've found that once developers and testers experience using a screen reader for even an hour, they become much more empathetic and proactive. Lucas: Absolutely. And there are great free resources — WebAIM has excellent articles, and the W3C's WCAG documentation is thorough. I'd also recommend the book 'Accessibility for Everyone' by Laura Kalbag. It's a practical guide. Luna: Alright, I think we've given listeners a solid framework. Any final thought on where accessibility testing is heading? Lucas: I think ai assisted testing tools are improving. They're getting better at simulating user interactions and detecting issues like confusing focus order. But I don't think they'll replace human judgment anytime soon. Accessibility is ultimately about understanding how real people interact with your software. That requires empathy, not just automation. Luna: Well said. Thanks, Lucas. Lucas: Thanks, Luna. And to our listeners, go test something — and make it work for everyone.