[Jugurtha]

Asking questions works wonders in understanding the problem, the job to be done, the incentives, the constraints, the parameters people are trying to tweak to optimize for something, the tradeoffs.

It is an iterative process. Then when you have an "advice", you can phrase it as a question of how would a new set of parameters be computed in their model and their impact on other aspects you may not be aware of. You could see the computation happening, and they'd say something like "Yeah, but it would negatively impact XYZ", which you'll take into account as a new constraint in the subsequent sets of parameters you submit for computation/simulation.

One effect of doing this is that a lot of people will ask for your advice because it works.