Every bootstrapped SaaS founder hits the same wall around $500K ARR: velocity drops, simple features take twice as long, regression bugs start compounding, and your devs are fighting the codebase instead of shipping.

The default instinct is to plan a rewrite. Stop-the-world, rebuild from scratch, emerge on the other side with a clean architecture.

Here's why that kills companies: the rewrite takes 2-3x longer than estimated, your existing product stops getting features, customers churn while you're building the replacement, and by the time it's "ready" you've burned through your runway or your market window.

I lived this exact failure in 2016 — fintech startup, 5 co-founders, ~500K BRL raised, 1,000+ early-access users. We attempted the rescue rewrite. It ran out of clock.

The alternative is what I call bottleneck migration: instead of rewriting everything, you identify the single constraint that's actually slowing your business down right now, and you ship the minimum viable migration that moves it. Then you find the next bottleneck and do it again.

Sometimes the right answer is unglamorous. A PHP monolith doesn't need to become a microservices paradise — it needs the one service extracted that's causing 80% of your deploy failures.

Three questions to find your real bottleneck:

1. What's the last feature that took 3x longer than it should have? Trace backwards — was it the code, the data model, or the deploy pipeline?

2. What breaks every time you ship something unrelated? That's coupling, and it's probably one module.

3. What would you need to change to hire your next engineer without a 3-month onboarding? That's your documentation and architecture problem in one.

The answer to all three usually points to the same place. That's your first migration target.