Every technology decision in an established organization eventually reduces to three options: buy a product, build something specific, or connect what already exists. Most programs that overrun chose the wrong one, usually by building what they should have bought or buying what could not be connected.
Buy the commodity
Where a process is the same across most organizations, a product will do it better and cheaper than anything built in-house. Email, payroll, service desks, document storage, and most reporting belong here. The decision criteria are fit to the actual process, total cost over several years including the connection work, and whether the vendor's roadmap matches the organization's.
Build where the process is yours
Where a process is genuinely particular to the organization, and where that particularity is a source of advantage, a product will force compromises that cost more than building. The test is strict: the process must be both unique and valuable. Most processes that feel unique are not.
Connect everything
Whether bought or built, the system has to work with what is already there. Connection is the option most often underestimated, because it is invisible in the demonstration and decisive in daily use. A product that cannot be connected cleanly to the systems of record is a product that will be worked around, and a built system without connections is a side database.
The practical sequence is to decide what to connect first, because that determines what can be bought and what has to be built. Programs that start from the product catalog get this backwards.