Start with evidence
Each troubleshooting page is written around observable state: logs, portal status, Graph output, event IDs, policy assignments, and device-side symptoms.
Avoid destructive first fixes
AdminSignal favours diagnosis before reset actions. Clear caches, delete registry keys, or re-enrol devices only when evidence points there and rollback impact is understood.
Close the loop
Validation matters as much as the fix. Use the follow-up checks to confirm policy state, reporting, user impact, and whether the same fault is likely to recur.