When a business hits a process problem, there are really only three responses: build custom software, buy an off-the-shelf tool, or automate a workaround between tools you already have. Choosing wrong is expensive in opposite directions — over-building wastes months, over-buying leaves you fighting software that doesn't fit. Here's how to decide.
Start with one question: is this a differentiator?
If the process is core to *how you win* — your secret sauce, your customer experience, the thing competitors can't easily copy — lean toward **build**. If it's necessary but generic (payroll, email, accounting), lean toward **buy**. You should never build what you can buy for something that doesn't differentiate you, and you should never trust a generic tool with the thing that makes you special.
When to buy
Off-the-shelf wins when the problem is common and well-solved. Benefits: fast, predictable cost, someone else handles maintenance and security. Watch for the hidden costs — per-seat pricing that scales painfully, data locked in their format, and the slow accumulation of "we work around the tool" habits.
Buy when: the need is standard, time matters more than fit, and the tool covers 80%+ of your requirements out of the box.
When to automate
Often the real answer isn't new software at all — it's connecting the tools you already have. Automation shines when your problem is *moving data and triggering actions between systems*: sync the CRM to billing, route form submissions, generate the weekly report. It's cheaper than building and more flexible than buying.
Automate when: the pieces exist but don't talk to each other, and the work is repetitive glue between systems.
When to build
Custom build is right when the process is a genuine differentiator, no tool fits, or you need to own the data and logic end-to-end. The cost is real — build *and* ongoing maintenance — so reserve it for where it compounds into an advantage.
Build when: it's core to your business, the fit has to be exact, and you'll still want to own it in three years.
A simple decision path
1. Is it a differentiator? → likely **build**
2. Is there a tool that fits 80%+? → likely **buy**
3. Do the tools exist but not connect? → likely **automate**
Most companies default to "buy another tool" for everything and end up with twelve subscriptions that don't talk to each other. Running this framework — even informally — is how you avoid that.