Usability testing is the most direct way to find out whether your product works. Five well-run sessions will surface more actionable insight than months of stakeholder discussions. Here's how to run them.
The Test Plan
Before recruiting participants, write a one-page test plan: what question are you trying to answer, who are the right participants, what tasks will you ask them to complete, what are you looking for, and how will you measure success? The act of writing this surfaces assumptions you're making and sharpens the tasks.
Recruiting
Recruit from your actual user population or close proxies. Screen participants with three or four qualifying questions about how they currently solve the problem your product addresses. Aim for five to eight participants per round of testing. More is rarely necessary for qualitative insight; fewer and you'll miss edge cases.
Moderating
The cardinal rule of moderation: don't lead the witness. "What would you do next?" is a good probe. "Would you click this button here?" is not. Let silence happen — users will often say what they're thinking if you don't fill the silence. When users get stuck, ask what they're thinking before you help them.
Give tasks as scenarios, not instructions. "You've just received $200 from a friend and want to save it for a trip to Lisbon — show me what you'd do" is better than "Navigate to the savings feature and create a savings goal."
Analysis
After sessions, note the top themes by frequency and severity. A problem that every participant encountered is more urgent than one that one person encountered. A problem that blocked task completion is more urgent than one that caused friction. Prioritize the intersection of frequency and severity.