Every incident opens as a numbered path to where to look

Up to seven steps, each with one next click

The page reads top to bottom as numbered steps. Needs attention now says how many customers are affected, how many failed once and how many repeatedly, how much the error rate changed, and how the affected endpoint's response time compares with before. What changed lists the deploys, feature flags, config, and dependency upgrades from the hours before it started, labeled potentially relevant. The likely cause follows, with its match confidence, its evidence, the pull request when a deploy is the match, and a recommended action.

Then the affected endpoint and its slowest captured request, the slowest database query in that request, and the outside service it was waiting on, each with one button to open it. A step with no data behind it is left out and the rest renumber, so there's never an empty stop. An issue opens the same way: what needs attention now, then Where to look, with the likely cause and a short numbered list of next steps.

The top of the path: who it's hitting, what changed just before it, and the likely cause.