Skip to content
Tutorials Center
My learning

Topics

Lesson 20 of 37 ยท Incident Response

Triage and Alert Investigation

Learn how SOC analysts triage alerts: gather context, ask the right questions, decide true or false positive, and escalate with a clear written summary.

A busy SOC receives far more alerts than it can study in depth. Triage is the quick, structured first look that decides what happens to each alert: close it as a false positive, investigate further, or escalate. Good triage is a skill built on a short list of questions and a habit of checking context before reaching a conclusion.

This lesson gives you a repeatable method and two small programs that practice it on made-up data. Everything here is a stand-in for the screens of a real SIEM; no real systems are touched.

The triage loop

flowchart TD
    A["Alert arrives"] --> B["Read the rule: what was it looking for?"]
    B --> C["Check the raw events"]
    C --> D["Add context: user, host, asset value, history"]
    D --> E{"Benign explanation?"}
    E -- "Yes, proven" --> F["Close as false positive and note why"]
    E -- "No or unsure" --> G["Investigate or escalate"]
    G --> H["Write a clear summary"]

Closing needs a reason you can write down. "Looks fine" is not a reason.

Questions to ask every time

  1. What does the rule detect, and why did it fire? Read the logic and the matching events. A rule titled "Impossible travel" fires because two logins were far apart in a short time, not because an account is hacked.
  2. Is the data correct? Check timestamps, time zones and field values. A parsing error can create a convincing fake alert.
  3. Who and what is involved? Which user, which host, how valuable is the asset, is the user an administrator?
  4. Is this normal here? Compare with the user's history, with their peers, and with scheduled work such as a vulnerability scan or a release.
  5. What happened before and after? One event rarely tells the story. Look at the surrounding minutes: a failed login followed by a success is very different from failures alone.
  6. Can I prove a benign explanation? If you can confirm it (for example a ticket for the maintenance window), close with that reference. If not, do not guess.

True and false positives

Real threat No threat
Alert fired True positive False positive
No alert False negative (the worst case) True negative

False positives waste time and cause alert fatigue, where analysts start ignoring everything. False negatives are the attacks you never saw. Triage notes feed back into alert tuning, which reduces false positives without creating false negatives.

Never close an alert just because the source address belongs to a well-known company or cloud provider. Attackers rent infrastructure from the same places everyone else does.

Enriching an alert

Enrichment adds facts to the alert so a decision is possible: asset owner and criticality, the user's department, whether the address is internal, previous alerts for the same entity, and threat intelligence (IOCs and enrichment). In a SIEM this happens automatically or by pivoting through linked searches.

Try it: a simple priority score

Many teams compute a numeric priority from the rule severity, asset value and whether the user is privileged. The weights below are invented for teaching; tune yours with real incident data.

Python Alert priority score

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

Try it: look before and after the alert

An analyst almost always pivots to the surrounding events. Here the "alert" is a successful login, and the surrounding window shows whether it followed a brute-force run.

Python Context window around an alert

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

Writing the triage summary

Whether you close or escalate, leave a note the next person can use without asking you. A good summary has four parts:

  • What fired: rule name and what it matched.
  • What I checked: sources and queries, with results.
  • Conclusion: false positive, benign true positive, or incident, and the reason.
  • Next step: closed, escalated to whom, or waiting for what.

Example: "Rule 'Failed logins burst' fired for svc-backup from 203.0.113.50. Four failures in six seconds, then a success, then a group change on the same account. Source is external and has no history. Likely credential guessing followed by privilege change. Escalated to Tier 2 as High; account disabled per runbook step 3."

Time-boxing and escalation

Set a time budget for triage (for example 15 minutes for Tier 1). If you cannot reach a confident decision in that time, escalate with what you have. Escalate immediately, without waiting, for anything that involves privileged accounts, domain controllers, data leaving the network, or ransomware indicators.

Common mistakes

  • Closing based on the alert title alone, without reading the events.
  • Ignoring time zones, which turns normal daytime work into "3 a.m. activity".
  • Investigating one host forever when the real story is on a second one.
  • Forgetting to record negative findings ("checked proxy logs, nothing found"), which saves the next analyst repeating the work.

Check your understanding

Quick check ยท 4 questions

of 4 answered You scored / 4 ยท perfect!

  1. 1. What is a false negative?

    A false negative is a missed detection. It is the most dangerous failure because nobody knows to look.
  2. 2. Which is a valid reason to close an alert as a false positive?

    Closure needs a verifiable benign explanation, such as an approved scan window.
  3. 3. Why look at events before and after an alert?

    Context reveals sequences, which is how many attacks are recognized.
  4. 4. Which situation should be escalated immediately?

    Privileged changes on critical systems carry high impact and need fast senior attention.

Summary

  • Triage is a fast, structured decision: close with a reason, investigate, or escalate.
  • Read the rule, check the data, add context, compare with normal, look around the event.
  • Priority combines severity, asset value and privilege; write down why you decided.
  • Leave clear notes, including what you checked and found nothing in.

Next: Containment, Eradication and Recovery