when should you really go all in on caching data in microservices

i've been thinking about caching lately in our microservice architecture. we have a few services that are read-heavy, but others are more write-intensive. i'm trying to figure out the tipping point for when it makes sense to aggressively cache data. like, how much is *too much* caching? are there specific metrics or scenarios where you say 'okay, this is definitely cacheable' versus 'hold on, let's rethink this'? what's your approach to balancing cache hit rates with potential stale data issues in distributed systems