Product Discovery That Connects User Evidence to Accessible Delivery

Write the decision, evidence needed, responsible owner and date. This prevents interviews, maps and prototypes from becoming attractive artifacts with no effect on scope.

Ask participants to show how the task is performed, where information comes from, which exceptions matter and what happens after the screen. Stated preference is useful context, but observed behavior and operating constraints often reveal a different requirement.

Recruitment determines what evidence can support. A handful of convenient colleagues cannot stand in for customers, field workers or people using assistive technology. State the limits of the sample instead of presenting directional findings as universal facts.

A journey is more than a row of ideal screens. Include entry conditions, permissions, waiting states, validation, exceptions, empty data, interrupted work, escalation and recovery. Name which person or system owns each transition.

This state model becomes a bridge between design and architecture. It exposes identity, data, notification and audit requirements early enough to change the plan.

Choose fidelity from the question. A paper flow can test information order; an interactive prototype can test navigation and language; a coded reference may be needed to investigate performance, responsive behavior or an unusual interaction.

Avoid polishing the familiar portion while the risky assumption remains hidden. Define what observation would support, challenge or invalidate the proposed direction before the session begins.

WCAG 2.2 provides testable success criteria, but accessible delivery also depends on clear content, predictable interaction and representative human review. Design annotations should cover structure, names, focus order, errors, status messages and behavior at reflow.

Automated tools catch only part of the problem. Verify keyboard use, screen-reader interpretation and zoom or reflow on the implemented journey, then keep the findings in the normal delivery backlog.

Separate what participants did or said from the team’s interpretation. Record the resulting decision, confidence, trade-offs, unresolved questions and the next condition that would cause the decision to be revisited.

Useful discovery reduces avoidable rework without pretending to eliminate uncertainty. Track whether critical assumptions were tested, whether accessibility defects escape late and whether product decisions can be traced to evidence; do not invent a universal percentage improvement.