Goal
Evaluate the accessibility of an AI chat assistant built into the account experience for screen reader and screen magnification users, and understand how it fit into their broader experience with the app.
Who
6 active account holders and 1 new user, all screen reader and/or screen magnification users, each completing a 45-minute session across desktop, mobile web, and native app.
Key findings
- Critical: Some users couldn't locate the assistant at all within their account.
- Critical: Screen readers interpreted the assistant's entry point inconsistently, with no confirmation of what users were interacting with.
- Serious: After opening the assistant, users' screen reader focus stayed on the previous screen, leaving them disoriented.
- Serious: The question text field was difficult to activate with a screen reader on native.
- Minor: Character counts read aloud while typing interrupted the experience.
- Minor: Users wanted responses organized under proper headings, and more personalized answers.
Outcome: Findings were prioritized for remediation with engineering
and design, alongside a broader push to improve the assistant's personalization to
match users' expectations of what the company already knows about them.
