Every platform team eventually faces this question: should we build our own internal developer portal, or buy Backstage/Cortex? Should we host our own GitLab, or pay for GitHub Enterprise? Should we write our own feature flag system, or pay for LaunchDarkly?
The answer is never obvious, and getting it wrong is expensive in both directions.
Core vs. Context: The Decision Framework
Geoffrey Moore's framework from Crossing the Chasm provides the clearest mental model:
- Core: Activities that directly differentiate your business and create competitive advantage. Uber's routing algorithm is Core. Netflix's recommendation engine is Core.
- Strategy: BUILD. Keep full control. Invest deeply. This is where your engineering talent creates unique value.
- Context: Activities you need to run the business, but that don't differentiate you from competitors. Uber's payroll system is Context. Netflix's CI/CD pipeline is Context (unless you're selling CI/CD).
- Strategy: BUY. Outsource to a vendor who specializes in this and does it better than you ever will.
The Engineer's Trap
Engineers love to build. It is genuinely fun to build a custom CI/CD pipeline or a bespoke secrets manager. But the question is never "Can we build this?" — it's "Should we?"
If your company is not in the business of selling CI/CD pipelines, every hour spent building one is an hour not spent on your actual competitive advantage. You are burning expensive engineering talent on Context while your competitors invest that same talent in Core.
A useful gut-check question: "Would a competitor pay us for this tool?" If the answer is no, it's Context — buy it.
A Third Factor: Constraints
Build vs. Buy is not purely an economic decision. Constraints can override economics entirely:
- Regulatory: A bank with "no external data processing" rules cannot use a SaaS feature flag service, regardless of cost.
- Security: A defense contractor may be prohibited from sending telemetry to a third-party observability vendor.
- Customization: If your requirements are so specific that no vendor covers 70%+ of them, building may be the only viable path.
Always evaluate Build vs. Buy across three axes: Cost, Capability, and Constraints.
A common platform engineering interview question is: "Your team wants to build a custom internal secrets manager instead of using HashiCorp Vault. How do you evaluate this decision?" The trap is to immediately say "buy Vault." The correct answer walks through the framework: Is secrets management Core or Context for your business? What are the compliance constraints? What is the TCO of self-hosting Vault vs. building? What happens when the engineer who built the custom tool leaves? Demonstrating structured thinking beats a snap judgment every time.