How do you justify time for refactoring?

Theory is simple: you start basic, you deliver something, you put it in product. Minimum Viable Product is there.

Then new features are required, you add them next weeks, all's good.

Now, you realize, simple architecture is not longer enough, you need to have your project more organized, modularized etc. in order not to end up in yet-another-spaghetti code.

And here goes a problem. How to justify that now you need N days of development only to refactor your project? No new features will be added, nothing that will be noticed by a user.

Sometimes such refactoring is simple (1 day) but sometimes it can be few weeks process, when e.g. refactoring monolith into microservices.

Managers – despite some of them understands the need of that – are evaluated by business value at the end of quarter, not by internal architecture at the.end of quarter.

What is the best approach?