A founder wants to try a new service. Calling it a pilot makes the decision feel manageable: a few customers, a few weeks, a chance to learn. But the name does not limit the work. The commitments do.

Before inviting anyone in, write the experiment as something another person could recognise as finished. What is being offered, to whom, for how long, and with what support? What happens to each participant when the test ends?

Separate the learning from the service.

“Find out whether people like it” leaves too much room for interpretation. A more bounded learning question might be: “Will three people from this audience complete this specific step after seeing the offer?” That is still not proof of a viable business. It is a question you can actually observe.

The service promise is a different matter. A short test can create ongoing expectations if it includes open-ended support, access without an end date, or a result you cannot control. Do not use an internal label to imply a boundary you have not explained externally.

Name what the pilot cannot borrow.

Consider a fictional founder with six hours a week available alongside existing client work. A pilot that takes eight hours has not become successful merely because participants like it. It has exceeded the resource boundary that made the test acceptable.

Write down which existing commitments are protected. Then decide who has authority to change the pilot if the limit is reached. If that person is you, put the review on your calendar before enthusiasm makes the exception feel inevitable.

A four-line pilot

  1. We are testing this specific question.
  2. We can spend this amount of time and attention.
  3. We will observe this evidence by this date.
  4. When it ends, participants receive this clear outcome or next choice.

Decide what a mixed result means.

Some people may like the idea while the delivery is too demanding. The test may be easy to run while nobody takes the next step. Treat those as different findings. Decide whether you would stop, change the offer, or run one further bounded test.

Any legal, financial, privacy or professional obligations still need appropriate review. A reflective exercise cannot establish them. A good experiment has both a useful question and an honest ending.

Put the thought to work

Give the question a page.

Five questions to help you see the choice, without sending us your writing.

Open your decision page

Educational reflection, not professional advice.