Plan the test — scenarios, tasks with success criteria, metrics to capture — and get a moderator script. The structure is the one Jeff Sauro recommends: scenario sets the context, task names a specific action, success criterion makes the result objective.
Every task gets three things. A scenario sets the context — the "why am I doing this" the participant needs to feel the task honestly. A task is a specific action, phrased as what the user wants to do rather than what the interface does. A success criterion is objective: either the participant did a specific thing or they didn't. Avoid "they seemed to understand" — write "they reached the confirmation screen within 2 minutes".
5 participants surface roughly 85% of major usability issues (Nielsen, 2000). Use the sample size calculator if you need to justify n.
Related: usability testing guide, task success calculator, SUS calculator for post-test.