Stop Testing Only with NVDA: Your Screen Reader Strategy Is Missing the Point

What screen readers should you test with? If your answer is NVDA and you're done, you're leaving users behind. The data says most screen reader users don't stick to one tool. They switch. Your testing should reflect that.

The Myth of the One True Screen Reader

NVDA is free, open source, and widely loved. In WebAIM's 2024 Screen Reader User Survey, 65.6% of respondents use NVDA, making it the most commonly used screen reader (WebAIM). But here's the kicker: 71.6% of respondents use more than one screen reader (WebAIM). So the user who starts with NVDA might switch to JAWS at work, or VoiceOver on their phone. If you only test with NVDA, you're ignoring the majority of how people actually navigate the web.

The Numbers Don't Lie: JAWS, VoiceOver, and TalkBack Matter

Look at the primary desktop screen readers: JAWS holds 40.5% and NVDA 37.7% (WebAIM). That's nearly a dead heat. On mobile, 70.6% use VoiceOver and 34.7% use TalkBack (WebAIM). So if you're designing a mobile app, and you only test with NVDA on desktop, you're missing VoiceOver on iPhone and TalkBack on Android. Each screen reader interprets accessibility APIs differently. What works in NVDA might break in VoiceOver. You need to test across the board.

One Example: A Form with Unlabeled Inputs

Consider a simple login form. In NVDA, an unlabeled input might be announced as "edit" — vague but somewhat understandable. In VoiceOver, it might be announced as "text field" with no label, which is confusing. Now, add a missing label and a missing document language (two of the most common WCAG failures, affecting 51% and 13.5% of home pages respectively, per WebAIM Million). In NVDA, the screen reader might guess the language and read the page in English. In JAWS, it might read in the user's system language, garbling the content. This isn't theoretical — it's what happens every day. And the WebAIM Million 2026 found 95.9% of home pages have detectable WCAG 2 failures, averaging 56.1 errors per page (WebAIM). So the odds are your site has issues that only appear in certain screen readers.

The Counter-Argument: NVDA Is Free, JAWS Is Expensive

You might say: "NVDA is free, so we test with it because we can't afford JAWS licenses." That's a fair point. JAWS is commercial and pricey. But you don't need to buy JAWS to test with it. You can use the free trial or use it in a virtual machine. More importantly, you can test with VoiceOver for free on any Mac or iPhone, and TalkBack is free on Android. So you can cover the major screen readers without spending a dime. The excuse of cost doesn't hold water.

My Recommendation: Test with at Least Three

Here's what your team should do: test with NVDA on Windows, VoiceOver on iOS, and TalkBack on Android. That covers the most common screen readers on desktop and mobile. If you have resources, add JAWS to the mix, especially for enterprise clients who might use it. But the minimal viable set is those three. The WebAIM survey shows that's what real users are using. And remember, 5.3% of respondents are both deaf and blind, using braille output (WebAIM). So also test with a braille display if you can. But start with those three.

Your job is to make software that works for everyone, not just the users who happen to pick the same tool you do. The data is clear: screen reader users are diverse in their choices. If you ignore that, you're ignoring a huge chunk of your potential audience. And with 1 in 4 US adults reporting a disability (WebAIM), that's a big deal.

Sources

• WebAIM Screen Reader User Survey - https://webaim.org/projects/screenreadersurvey10/
• WebAIM Million - https://webaim.org/projects/million/
• NV Access - https://www.nvaccess.org/about-nv-access/