Nobody who builds software for a living has much incentive to tell you not to build any. That's worth naming upfront, because it means the honest version of this question rarely gets asked by the person answering it.
Signs you should buy
- The problem is one other businesses in your position also have. Invoicing, scheduling, basic CRM, email marketing — these are solved, and solved well, by tools that cost less than a month of a developer's time to set up.
- You'd be rebuilding a feature a tool you already pay for shipped in the last two years. Most CRMs, accounting packages and office suites added AI features recently, and it's worth checking before assuming you need to build one from scratch.
- Getting it wrong costs you a subscription fee, not a rebuild. Low stakes favour trying the off-the-shelf option first.
Signs you should build
- The workflow is specific to how you run your business, and every off-the-shelf tool makes you bend your process to fit its screens instead of the other way round.
- It's the thing that makes your business different from a competitor doing the same work — the part you wouldn't want a rival buying off the same shelf you did.
- You've already tried gluing two or three tools together with a subscription-priced automation platform, and you're now paying monthly for something that breaks whenever one of the vendors changes its layout.
Most real answers aren't fully one or the other. The common shape is: buy the generic parts, build a thin layer that connects them the way your business actually works. That layer is usually small — which is also why it's usually cheap.
The test that actually settles it
Search for the specific workflow, not the general category. "CRM" returns a hundred tools. "CRM that quotes jobs the way a trade business quotes jobs" returns far fewer, and if nothing close exists after a genuine look, that's a real signal, not a guess.
The reason this is worth a paid few days rather than a free afternoon is that the answer changes the entire shape of what comes next — and getting it wrong is expensive twice: once for the thing you built that you didn't need, and again for the subscription you're still paying for the thing you did.