Microservices: When Should You Actually Break Up a Big App?

we've got a fairly large monolithic application that's starting to get unwieldy. development cycles are slowing down, and deploying even small changes feels risky. we're considering breaking it down into microservices, but it feels like a big undertaking. has anyone successfully migrated a significant monolith? what were your key decision points for deciding *what* to break off first? any common pitfalls to watch out for?