Begin with the service and trace the supporting layers: applications, databases, integrations, infrastructure, information flows, access points, owners, monitoring, and backup-and-recovery procedures. The result should show what a failure affects, which dependencies matter, and how the service is restored.