Users grouping your content/features into piles they name — open sorts reveal mental models, closed sorts validate candidate categories; 15–20 participants show the clusters.
Findability tasks against the bare structure ('where would you find X') with success and path metrics — the navigation tested naked, before visual design can compensate or confuse.
Sort → design taxonomy → tree test → iterate — a two-week evidence loop that settles the org-chart-vs-user-model navigation argument permanently.
Data-heavy applications, portals, and any product whose support tickets read 'where do I...' — findability failures masquerade as training problems.
Skipping the discipline this article describes until an incident, audit, or stalled project forces it — every practice above is cheaper adopted early than retrofitted under pressure.
Let's discuss how we can help you with card sorting tree testing.