When should you add read replicas to your database?

hey everyone, i'm wrestling with a common scaling question and wanted to get your thoughts. we've got a primary database that's starting to feel the strain during peak read operations. writes are still well within limits, but the read load is increasing steadily. i'm considering adding read replicas to offload some of that traffic. what are your experiences? at what point did you decide read replicas were the right move? are there specific metrics or thresholds you look at before pulling the trigger? any gotchas to watch out for