The handoff problem
Most operational pain is a handoff failing: between the person who built something and the person operating it, between the engineer on call last week and this week. Process adds checkpoints to the handoff. Documentation removes the need for one — the answer is already written down.
Decision notes
A decision note is a paragraph, not a document: what we decided, why, and what would change our mind. Written at the moment of decision, it costs minutes. Reconstructed a year later, it costs days of archaeology and arguments.
Ownership records
Every system needs one findable answer to "who owns this?" — a name, not a team alias that resolves to nobody. Ownership records turn incidents from broadcast panics into direct conversations.
What to write down
- One-paragraph decision notes, dated and signed.
- A runbook per alert, kept next to the alert definition.
- An ownership record per service, reviewed quarterly.
- A single onboarding page: how to deploy, where logs live, who to ask.