when should you actually use CQRS with a microservice?

i’ve been thinking a lot about command query responsibility segregation (cqrs) lately. on paper, it seems like a clean way to separate read and write concerns, potentially optimizing each independently. but in practice, especially within a single microservice boundary, i’m wondering about the real-world benefits versus the added complexity. are there specific scenarios or types of services where you’ve found cqrs particularly useful, or is it often overkill for individual services and better suited at a larger system level