Been in debates about this forever. Here's my honest breakdown after working on a few projects:
Go no-code if: You need something simple, your workflow is generic, you don't have serious scale plans, and speed-to-market is everything.
Go custom if: You have unique business logic, you're integrating with multiple legacy systems, you need to white-label, or you're planning to scale to enterprise clients who will audit your tech stack.
The mistake I see founders make is using no-code to validate, then realizing 18 months later they've painted themselves into a corner because the platform can't support what they need. Custom is more expensive upfront but cheaper long-term if you have real complexity.
What's your experience? Have you successfully migrated from no-code to custom, or regretted going custom too early?