Cognitive testers find more usability problems. A 30-participant study run by Fable VP of Innovation, in collaboration with UC Irvine researchers Syed Fatiul Huq, Pranav Pidathala, Ali Brown, and Michael Fagan, tested three AI-generated websites across task-based sessions split evenly between cognitive and general population participants. Cognitive participants identified 197 issues and made 93 improvement suggestions. General population participants found 113 issues and offered 54 suggestions. Cognitive disability affects 13.9% of U.S. adults per CDC data, and that number is rising according to a 2025 Yale study.
The methodology is the story here, not just the numbers. Fable's working group spent months developing a screener, a facilitation guide, and an adapted survey called the Accessible Usability Scale to make this kind of research reproducible. Participants self-identified challenges with memory, focus, and learning. The three test sites covered distinct interaction patterns: recipe browsing, e-commerce with a book-matching feature, and appointment booking. Cognitive participants surfaced disproportionately more issues around content clarity, buttons, icons, and visual elements than their general population counterparts.
The full article documents the recruitment screener, the pilot study of 25 testers that preceded this research, and the specific categories where cognitive participants outperformed general population testers in surfacing friction. If you run UX research and currently exclude this population, you are leaving a significant volume of actionable findings on the table. The original is worth reading for the methodology detail alone.
[READ ORIGINAL →]