An underrated approach here is to ask questions if you think someone is making a mistake, and then see how they answer. "Oh, cool - why do you think X is going

An underrated approach here is to ask questions if you think someone is making a mistake, and then see how they answer. "Oh, cool - why do you think X is going to be better than what we've been using?" etc. There's a huge difference between asking questions about an approach and prescribing an approach or saying "Nope that's not the right way" and there's a ton of room there to avoid the really suboptimal stuff.