It's pretty simple really.
After observation, offer a claim that can be falsifiable. Then through the claim based on observational evidence, conduct repeated tests on such a claim to see if it is false, or if through repeated tests, see if it is a valid claim. Then offer the studies to peer review to minimize any confirmation bias.
Rinse, lather, repeat. If the claim if falsifiable, we have a good start. If it is testable, it's even better. And if it stands up to peer review, let's do it again, have some fun, and see what is revealed to be self-evident.
It isn't this: Design theories are to be immediately scorned, mocked, and ridiculed in public because scientists are all hateful atheists who don't like religious people.