"We need microservices" is usually a wish for clearer boundaries, not a hundred deployables. A big-bang rewrite trades a working system you understand for a distributed system you don't.
Start inside the monolith. Carve modules with explicit interfaces and stop reaching across them. Once a boundary is clean in code, you've already captured most of the benefit — and you can deploy it tomorrow.
Extract a service only when a module has a real, independent reason to scale, deploy or fail separately. Move the highest-leverage seam first, behind a feature flag, with the old path as a fallback until the new one has earned trust.
Measure the thing you actually care about — lead time, reliability, on-call load — not service count. A well-modularised monolith beats a poorly-bounded mesh of services every single time.