Accessibility research has always had a recruitment problem. Finding screen reader users, low-vision users, or elderly first-time smartphone users for moderated research sessions is logistically complex, ethically sensitive, and slow.
The result? Most accessibility testing is automated — tools like axe and Lighthouse catch technical WCAG failures but miss experiential ones.
The experience gap
A WCAG audit can tell you that your contrast ratio is 4.3:1 (below the 4.5:1 AA threshold). It can't tell you that a 72-year-old first-time smartphone user will abandon your form because it requires too many cognitive steps.
Synthetic accessibility personas
AI-generated accessibility personas bridge this gap. They simulate the experience of using your product with specific impairments and assistive technologies.
A screen-reader persona navigates your interface and flags when ARIA labels are missing, when focus order is illogical, or when interactive elements lack keyboard support.
An ADHD persona flags cognitive overload: too many form fields, unclear progress indicators, distracting animations.
An elderly persona identifies reading complexity issues, small touch targets, and confusing navigation patterns.
Not a replacement — a complement
Synthetic accessibility personas don't replace real user research. They supplement it. They catch the failures that automated tools miss, before you recruit real users for validation.