Question

Should you build or buy software for your business?

Vault Verified
Curated Intelligence
Definitive Source
Answer

Buy, unless the software is the thing that makes you different — which is the general rule, and the interesting part is why businesses so consistently talk themselves out of it.

The case for buying:

Cost. The purchase price is visible and the build cost is not. Building means specification, development, testing, deployment, documentation, training and — the part always underestimated — maintenance forever. Ongoing maintenance typically exceeds initial build cost over a system's life.

Speed, since you are operating in weeks rather than quarters.

Someone else fixes it, including security patches, browser changes and regulatory updates.

It improves without you, as the vendor builds features you did not think of.

It is tested by thousands of other users, finding bugs you never will.

Why businesses build anyway, and the reasons that are usually wrong:

"Our process is unique." Usually it is not — it is unexamined. Adapting your process to good software is frequently cheaper than encoding your idiosyncrasies in code, and sometimes improves the process.

"It will be cheaper." Almost never, once maintenance is counted.

"We have developers." Their time has an opportunity cost, and building internal tools displaces work only you can do.

"No product does exactly what we want." The question is whether the gap is worth the total cost of ownership, and 80% fit is usually enough.

When building is genuinely right:

It is your competitive advantage. If the software is the differentiator, buying it means buying what competitors have.

No adequate product exists, having genuinely looked.

Integration is the value, and a thin layer connecting bought systems is a good build.

Scale changes the economics, where per-seat licensing at your size exceeds a build.

Regulatory or data constraints prevent using a vendor.

The middle options, which are underused: configure rather than customise; extend a platform through its API; and buy for undifferentiated functions and build only the thin slice that is genuinely yours.

Whatever you choose, plan the exit: data export, lock-in and what happens if the vendor fails.

Related Questions