The hypothesis is tested, ideally many times by many people
In each case, the test produces results consistent with the hypothesis

These tests form a body of evidence that serves to disprove competing hypotheses and (until some sort of contradictory evidence is found) makes the theory well-established

So you observe something, then you form a hypothesis to explain the observation, but that's not enough to be scientific. You have to actually test the hypothesis and try to explain what's happening. Merely positing a hypothesis is not good enough. A hypothesis which serves to "explain" without producing understanding is also not good enough.

In creationism / intelligent design debates there's a notion of "god of the gaps" where god is invoked to fill in a gap in knowledge, which doesn't really explain anything so much as give a label to our ignorance. The same thing happens in physics where some particle or phenomenon is posited to "explain" observations that are inconvenient / inconsistent with existing theory. Unless those hypothetical particles / phenomena can be tested and their mechanisms understood, they are just other ways of naming our ignorance. 

 I like your argument here but the "something that produces understanding" part is fairly subjective and fragile.

Ofc I know what you mean there and don't mean to gaslight but it's still not a solid definition. 

 The way Feynman put it was that your theory shouldn't just make the observation that produced it "come out right" but should make something else "come out right" as well 

 So either we accept that "social consensus" is an intrinsic part of the scientific method, or we broaden the definition of what constitutes a scientific model 😂