I meet a lot of engineers who are trying to build new products / SaaS tools they want to sell and I see the majority of them making the same mistake: trying to build everything from scratch. They decide they need to self host their auth, their email, their database, their queue, their analytics, and their feature flags before they've written a single line of actual product, which is boring, painful, and can take weeks of your time, often killing your enthusiasm for the new project itself.
This is why I buy almost everything.
Background Context
I run an agentic wedding planning company called Losava. Losava runs on Next.js on Vercel, Supabase, Clerk, Stripe, and Resend. Clerk handles auth. Resend handles email. Both have free tiers that cover way more than most side projects will ever touch.
Because I use these two services, I didn't build a login page, fraud prevention, a password reset flow, a magic link system, an email deliverability setup, or a webhook retry queue. I plugged in a package, read the docs for 10 minutes, and moved on to the thing users actually care about, and the features that will actually make me money.
When to buy:
Say you get so much traffic that Clerk and Resend (or whatever else you’re using) start costing you a couple hundred bucks a month.
That means you have a lot of users. Which means you've probably made a lot of money (or you're about to). At that point you can sit down, look at the bill, and decide whether migrating off is worth it.
IMO it actually usually isn’t. You can take the time you would have spent on the migration to introduce a new feature or launch a marketing campaign that generates more revenue than the incremental savings you’d get from migrating off managed platforms.
But at some point (in the $5,000+/month range typically), it clearly is, which is why large companies like FAANG have huge teams that are focused on building internal tools (for example, Amazon has an internal Jira so they don’t have to pay fees to use sprint planning software).
The odds of a side project getting to that point are EXTREMELY low. Way lower than the odds of it never shipping because you spent three weekends setting up self hosted auth.
I've watched a lot of engineers hamstring their own projects this way. Not because self hosting is wrong, but because it's fun and feels productive. You get to make architecture decisions. You get to write infra code. You get to feel like a Real Engineer.
Meanwhile nobody can sign up for your app because the login page isn't done.
When self hosting does make sense
You have a hard compliance or data residency requirement.
You're already at the scale where the bill is bigger than an engineer's salary.
The thing you're building IS the infrastructure (if your product is an email API, yeah, don't use Resend).
You actually want to learn how it works, and you're honest with yourself that this is a learning project, not a shipping project.
If none of those apply, stack up free tiers and go build the product.
Abuse the free tiers while you can. If you scale, see if it makes sense to switch.
Have a good week!
Arjay.
P.S. If you want to chat with me directly or get advice on this stuff, the best way is through my private discord server included with Devmaxx. Try it free for 7 days at devmaxx.dev.