The most common reason usability testing does not happen is that it is imagined as a research project. It can be five people and an afternoon.

How to run one badly enough to be useful

• Give a task, not a tour. "Find a plumber available on Saturday" rather than "what do you think of this page".
• Say nothing while they struggle. The silence is unbearable and it is where the findings are.
• Ask what they expected to happen when something surprises them. That question produces more insight than any other.
• Watch with the team present. A stakeholder who watches one person fail is more persuaded than by any report you could write.

The limit worth knowing

Five people will find most of the serious problems in a flow. They will not tell you which of two designs converts better, or how common a behaviour is. For that you need numbers. Knowing which question you are asking is what keeps the method honest.