NoSQL is about eventual consistency in distributed architectures. SQL is about centralized consistency on 1 physical file.
So classic example -> user data / financial data is something you probably want consistent this is SQL data.
user action streams are something that 'in the end should be correct' they can be eventual consistent.
microservices are mainly an infrastructure pattern, so the question is 'how do you access your data' in this case. The type of database have their pitfall you need to compensate via CAP/PACELC anyway.
This question is hard to answer. There are always lots of variations to consider.