Build vs. buy, in 2026

There used to be a clean trade-off. Build, and you got fit and control but paid in engineering time. Buy, and you got speed but compromised on fit. Companies usually picked one side and stayed there.

That trade-off has shifted, because the cost of building has collapsed. A small team can stand up a working internal tool in two weeks that used to take a quarter.

What still belongs to "buy" #

• Anything with a strong network effect (CRM, payroll, identity)
• Compliance-heavy infrastructure (security tooling, audit trails)
• Commodity surface area where speed matters more than fit (calendar, email, video)

You will not out-build Salesforce. You will not out-build Datadog. Don't try.

What now belongs to "build" #

• Internal workflows where you're the only company doing it that way
• Glue code between systems that aren't designed to talk
• Reporting that needs to be specific to your business model
• Customer-facing surfaces where the experience is part of the product

A workflow you used to outsource to a vendor with a $90K/year contract can often live as a 400-line internal tool maintained by one engineer for half a day a month. That math didn't work in 2022. It works now.

The audit-side answer #

When we do an audit, the build/buy recommendation is per workflow, not per company. The right answer for marketing automation might be different from the right answer for invoice processing. Treat each as its own decision.