Growing businesses tend to hit infrastructure problems in a predictable order: first, a single server that can't handle traffic spikes; then, manual deployments that are risky enough that nobody wants to ship on a Friday; then, no real visibility into what's actually failing when something breaks. None of these require an enterprise-scale rebuild to solve.
The three things worth fixing first
Proper CI/CD, so deploying is a routine, automated, low-risk event instead of a manual ritual someone has to remember every step of. Containerization, so 'it works on my machine' stops being a real risk in production. And actual monitoring, so the team hears about a problem from a dashboard, not from a customer.
Scaling infrastructure without scaling complexity
The trap growing businesses fall into is copying the infrastructure patterns of companies ten times their size, adopting complexity that solves problems they don't have yet. The right infrastructure for a 20-person company handling real but moderate traffic looks meaningfully simpler than what a company processing millions of daily transactions needs, and that's correct, not a compromise.