Maybe the line for me is how dependant we'll be on the service, and how hard to switch? How much ownership of the feature/service does the company need now and in 10 years.
One startup I know used an auth-provider-service. It missed on some feature they needed so they had to hack a workaround and now also it's expensive.
- building a personal tool means not being tied to someone else future, so a good thing, but...
- ...building in modern crapware model it's purposely hard to makes people unable to profit from computing.
In classical systems where the OS and all the rest was a single application, with user programming concept where anything is just a function, live code, developing and integrating personally was easy. In a divide et impera modern model of base system, userland, apps, even worse containers and so on ANYTHING not already made is hard and integration is substantially impossible.
So? Well, IF we can do something at home/with FLOSS that's the way to go, if not we know we have an armed bomb on our genitals. Not nice, but not other options possible so far.
If you have the time but not the money, build your tools
If you have both, hire someone to build them
If you lack both, there are other priorities than tools