Our default advice is to buy. Off-the-shelf software is cheaper to start with, someone else maintains it, and it usually covers 80% of what you need.
The real question is what the other 20% is costing you.
Count the workaround hours
Every business running on bought software has workarounds. There's the Monday export that someone reformats by hand, or the spreadsheet bridging two systems that can't see each other.
Add up the hours your team spends on these each month and put a rand value on it. If that's more than a small integration or internal tool would cost over two years, building starts to make sense.
Build the gap between systems
Building custom rarely means replacing everything. Usually it's a thin layer: a small app that pulls from your accounting system and CRM so sales can see what they need, or an API that lets your online store check live stock in a legacy system.
Payroll, accounting and email are solved problems. Keep paying for those.
Get something in front of users within six weeks
Our rule of thumb is that the first version should be in someone's hands within six weeks. One screen, one workflow, real users. If it doesn't earn its keep at that size, it won't at ten times the size either.