Scaling read operations: Read replicas or CQRS?

hey everyone im trying to design a system where read ops are way more common than writes im trying to figure out if i should use database read replicas or the command query responsibility segregation pattern (cqrs) read replicas prob seem easier to get going at first but im worried about replication lag cqrs gives you better separation but its more complex what have you guys seen work for read heavy stuff when did you pick one over the other