How do you define the problem you're working on clearly?

I want to open up a discussion about how to define, refine and validate the problems we're working on. Below is a summary of what I have experienced with my co-founders over the last 2 years, and what I have observed in the process from other startups.

With every product iteration or pivot, I find myself doubting that the problem we are attacking is clearly defined.

Initially, it's a battle against oneself as your excitement to solve a problem you think exists makes your mind cloudy and unable to see the ins and outs of the problem - mainly, is the problem big enough that others would try to solve it?

Then, once you have put together a prototype or even an MVP, we're dealing with confirmation bias and the sunk cost fallacy.
User feedback sessions (hopefully you're running those at this point) can be quite biased too, as your close network is likely to be encouraging and skew towards positive feedback.
I also find myself guilty of adding more features at this stage, thinking they're sure what's needed to get active users!

And we arrive at a point where you are confused about the problem you are solving. There are many notes and discussions, you remember it was surely clearly defined just to realise that in fact maybe the 'clearly' part was missing, rendering the problem research likely invalid and the prototyping/MVP stage a pure chance.

Maybe you took a small problem or more of an inconvenience, and in your head extended that to be as important a problem as solving cancer.

So, how do you define, refine and validate the problems you're working on?
How has that worked out for you thus far?
Do you have a framework or process you follow?
How do you look for others with the problem to further the validation?

J.