Steps to build an engineering strategy

The piece argues that a solid engineering strategy is not a vague manifesto but a concrete workflow of five linked steps. Exploration gathers industry ideas and recent research, creating a foundation that prevents you from building on outdated assumptions. Diagnosis forces you to pause and articulate the exact problem before any solution, using data or narrative to avoid chasing symptoms.

Real-world examples illustrate each phase. The Uber service migration case shows how reading Google Borg and Mesos papers, then mapping the cloud ecosystem with a Wardley map, shaped their approach. A data-driven diagnosis of headcount cost growth reveals how precise numbers steer policy choices rather than gut feeling.

Refinement introduces three practical tools-strategy testing, systems modeling, and Wardley mapping-to validate ideas against reality before they become policy. Policy translates the diagnosis into concrete decisions about architecture, review processes, or hiring allocations. Operations then builds the mechanisms-escalation paths, nudges, regular check-ins-that turn policy into day-to-day behavior.

The author warns against rigid templates that become bureaucratic weight. Instead, keep the structure flexible, discard sections that don't serve a clear purpose, and let the thinking behind each step drive the document. The result is a repeatable habit that reduces strategic errors and makes strategy a shared, actionable practice across the engineering organization.