The expensive part of testing is rarely the study. It is the organisational courage to let evidence change the plan before engineering has sunk the cost.
Test the decision, not the decoration
A useful study asks whether a person can understand the state, choose the next action, recover from failure, and explain the consequence. Preference questions and visual reactions are secondary.
What decision should this study change?
A small decisive sample beats a broad vague one
Five to eight well-selected participants can expose a repeated comprehension failure. The goal is not to estimate a population mean. It is to find the failure condition, correct it, and test again.
Which participant has the constraint that makes the test useful?
Write the gate before the sessions
Name what evidence would change the direction. Without that contract, teams reinterpret inconvenient findings until the original plan survives.
What evidence would cause the team to stop or reverse direction?
Ship the memo, not just the findings
Every study should end with a recommendation, an owner, a next action, a measure, and what remains uncertain. Insight without consequence is research theatre.
Who owns the action after the readout?
Good interface work makes the decision, evidence, owner, and consequence easier to see.