Microservice Strangler Fig Pattern - When is it TOO Late?

Hey everyone, I've been thinking about the Strangler Fig pattern for gradually swapping out big old monolithic systems for microservices. It's a really smart way to lower the risk when you're making that kind of change. But, I've also seen places where the 'monolith' got super complicated, and the new microservices multiplied like crazy. Managing the facade and the whole migration process turned into a huge job. So, when do you think it becomes too late or just too messy to really use the Strangler Fig pattern? Are there any clues or numbers you watch for that hint maybe a different strategy would be better?