SIEM query practice and a search reference
A search needs a question and a boundary. “Find authentication failures in this time window” is narrower than “find suspicious activity.” Inspect the fields before writing the query: source, action and outcome describe different properties.
A bounded search
This example uses the course’s generic teaching language and fictional event schema:
EVENTS
| WHERE source = "auth" AND outcome = "failure"
| WHERE time >= "2026-10-17T09:00:00Z" AND time < "2026-10-17T10:00:00Z"
| SORT time ASC
The interval includes its start and excludes its end. Adjacent windows can then share a boundary without double-counting it. Keep the time zone visible. In these examples, ISO timestamps use UTC.
Counts and scope
| Stage | Result |
|---|---|
WHERE user IS NOT NULL |
Keeps records with an identified user |
STATS COUNT AS attempts BY user |
Counts events for each user |
STATS DISTINCT(user) AS users |
Counts distinct non-null users |
FIELDS time, user, outcome |
Selects fields to inspect |
LIMIT 10 |
Keeps only the first ten current rows |
Filter the event set before aggregating it. Five failures by one account are five events and one distinct user. A row limit changes what you can conclude from a count. Read the raw events again if a summary hides the evidence you need.
Platform boundaries
The exercise checks this small generic language. It does not execute Splunk SPL or Microsoft KQL. Use Microsoft’s KQL overview or Splunk’s search manual when you move to those platforms; field names, operators and result behavior need checking there.
A failed authentication alone does not prove a compromise. Connect it to the other events and state the missing evidence before choosing an incident action. The SOC analyst course practices that handoff.