First way: try out guesses with small, reversible tests
Try out changes on a small scale—involving no more than a few people for a short period of time—and in a way that's easy to bail on.
Things might be a little crappy or prototype-y. Use a piece of paper instead of a fancy front end app. Fake it manually today. Automate later. You want only enough to figure out if your guess was any good. Small, reversible tests.
Make tests reversible in order to take the sting out of failure. Idea didn’t work? Awesome: stop and try the next one.
Make tests small so that you’re not dumping resources or spending time on something that might turn out to be a dead end. Commit to something only after you can point to a successful test and everybody’s ready to go.
Second way: a logbook of guesses
Once you start testing your guesses, you’ll find that you make a lot of guesses, and run a lot of little tests. So keep a log. Write down:
- Your guesses,
- What you thought might happen, and
- What actually happened.
- Bonus points: note whether your test was bigger or smaller than it needed to be.
Do this with discipline. Keep this practice and your guesses will get better. It’s a skill that improves almost automatically, as long as you pay attention to your guesses and compare them against how things turned out.