The most expensive way to test a career is to take one. The second most expensive is to imagine it thoroughly enough that you stop needing to test it at all — which is what most people do, and why so many well-researched moves collapse in month four.

What makes a trial real

A real trial has three properties. It puts you inside the work rather than adjacent to it. It's small enough to run in a weekend. And, critically, it's capable of coming back negative — if there is no result that would put you off, you have designed a reassurance exercise, not a test.

Four formats that work

The shadow day. One person, one day, no agenda beyond watching. Ask to see the boring middle of a Tuesday, not the interesting parts.

The tiny gig. The smallest paid unit of the work that exists. Payment matters less for the money than for what it does to the other person's expectations of you.

The single client. One real problem, one real deadline, delivered start to finish. This is the only format that surfaces how you feel about the admin, which is where most enthusiasm goes to die.

The teaching test. Prepare an hour on the subject for an audience of three. Nothing exposes shallow interest faster than having to explain something.

Reading the result

Write your prediction down before you start — how absorbed you expect to be, what you expect to hate. Score against the prediction afterwards, not against the feeling. Novelty makes almost every first day pleasant, and pleasant is not information.

Then bring it to people who will not let you edit the result. Trials that get debriefed alone have a way of confirming whatever you already wanted.