Choose events that answer real questions

Start with administrator sign-ins, account changes, permission changes, failed authentication, software deployment and important data exports. Include events from email, cloud services, websites, servers and network controls that support critical work.

For each log source, ask what an investigator would need to know: the actor, time, system, action and result. Check whether the records use a consistent time setting and identify the relevant account or service.

Protect logs and personal information

Restrict who can read, change or delete logs. Send important records to a separate system where a compromised host cannot easily alter them. Set retention based on business, legal and investigation needs.

Do not store passwords, access tokens or unnecessary personal data in logs. Review what the application records and redact secrets before they reach a central platform.

Give alerts an owner

Create alerts for events that need a response, such as an unexpected administrator login or a new privileged key. Name the person who receives each alert and define what action they should take.

Test the alert path and review whether alerts are useful. Logging without review can create a false sense of coverage. An assessment can identify important events that your systems do not currently record.