The question usually arrives as a product name. "Should we move to Salesforce?" "Is there an app for this?" Before comparing products, I want to know what the process is supposed to do and where it breaks today. The answer is one of three, and building is the one I recommend least often.
Fix what you have
Often the tool is fine and the process around it isn't: fields left blank, two people with different rules for the same step, a report that only one person trusts. A Boston nonprofit sailing school I'm working with runs on 16 platforms, including Salesforce, scheduling, payments, assets and maintenance tracking. The engagement is to cut that sprawl, and the first step is a list of what each platform does and who uses it.
Buy
If an off-the-shelf tool fits most of the process, and the part it doesn't fit can change, buy it. It comes with support, updates, and other customers finding the bugs first. If buying beats building, that's what I'll recommend.
Build
Build when the process is how you make money and no product fits it, or when the data has to stay yours. A cybersecurity firm needed every client account on one page: the contract in force, the signed documents behind it, how the client is billed, and a history of every change that can't be quietly rewritten. The firm's CEO priced a CRM license plus an implementation partner, then chose to own the system instead. It went live with 31 client accounts and hundreds of contacts.
How to tell which
Write down the process as it runs today, exceptions included. Mark the steps a tool would have to handle before you could stop using the spreadsheet. Then check each product demo against that list, instead of letting the demo set the list.