3 ideas
Before reaching for a solution, build a simplified representation that isolates the key variables and relationships driving the situation. A good model deliberately ignores most details so that the few that actually matter become visible and manipulable.
To know whether something caused an outcome, ask what would have happened in its absence rather than just observing what followed it. The difference between the actual world and the imagined counterfactual world is the genuine effect, which guards against confusing correlation with causation.
A good decision can produce a bad outcome and a bad decision a good one, because chance intervenes between choice and result. Evaluate the quality of a decision by the information and reasoning available at the time, not by retrospectively grading it on how things happened to turn out.