Incident Response Playbooks: A Complete Worked Example
Skip to content

Incident Response Playbooks

An incident response playbook is a documented procedure for handling one type of security incident. It tells the team what to check, what to do, who decides, and when the incident is over.

The clearest way to understand a playbook is to read one. This guide presents a complete playbook for a compromised user account, explains why each part is there, and shows what changes for other incident types.

Why Playbooks Exist

In the first minutes of an incident, the person on duty has to answer several questions at once. Is this real? How serious is it? What do I do first? Who do I tell? What must I avoid touching? Each answer is simple for someone who has handled the situation before, and hard for someone who has not.

A playbook exists to transfer that experience. It captures what a skilled responder would do, in the order they would do it, so that a less experienced person on a night shift produces the same result.

Stress narrows thinking. People under pressure skip steps they would never skip on a calm day. They reset a password and forget to revoke the session. They close the incident without checking whether other accounts were hit. A written sequence prevents this.

Incidents are rare for any one person. A security analyst may handle a compromised account every week. An IT generalist at a mid-sized company may see one a year. Without a playbook, each one is handled from memory and guesswork.

Order matters. Many response actions are only effective in sequence, and some destroy evidence if done first. Shutting down a server before capturing its memory loses what was running. Notifying the user by email when their mailbox is compromised tells the attacker.

Playbooks also make response measurable. When every incident of a type follows the same steps, it becomes possible to say how long each step takes, where delays occur, and whether a change improved things.

They are not a substitute for judgment. A playbook covers the expected path and tells the responder when to stop and escalate. The best ones say plainly where the script ends: If any of the following are found, raise the severity and call the incident lead.

A Complete Playbook: Compromised User Account

PartContent
TriggerA sign-in from an unfamiliar country or network; a new authentication method the user did not add; a new inbox rule or forwarding address; contacts reporting unexpected messages; an identity protection alert rated high.
SeverityMedium by default. High if the account is privileged, belongs to finance or an executive, or more than one account is affected. Critical if administrator actions by the attacker are found.
First 15 minutesStart the incident log, then: 1. Block sign-in if the attacker is active or the account is privileged. 2. Reset the password. 3. Revoke the account’s sessions. 4. Remove what the attacker added (authentication methods, inbox rules, forwarding, application grants, devices), recording each before removal. 5. Confirm in the sign-in logs that nothing from the attacker’s sources succeeds.
Evidence to collectSign-in records for the past 30 days, interactive and non-interactive. Audit records for the account. The record made in step 4 of each rule, forwarding address, delegation, application grant, and device. The phishing message, if one is found.
Questions to answerWhen did access begin? How was the account compromised? What did the attacker read, download, send, or change? Which other accounts signed in from the same source?
Containment optionsDisable the account for an extended period (incident lead). Block the attacker’s networks tenant-wide (incident lead). Force sign-out for all users (executive sponsor).
Decision pointsDoes the mailbox or file store hold personal information the attacker accessed? Do customers or suppliers who received messages need to be warned? Is a payment at risk?
CommunicationThe user and their manager, at once. Finance, if payment requests were sent. Recipients of attacker messages, once identified. The privacy lead, if personal information was accessed.
Exit criteriaNo sign-ins from the attacker’s sources for 48 hours. No unrecognized authentication methods, rules, grants, or devices. Exposure window and affected data recorded. Other affected accounts handled.

This fits on a page. A responder who has never seen this incident type can follow it, and a manager can see at a glance where their decisions come in.

Why Each Part Is There

Trigger

The trigger lists the specific signals that start the playbook, so the person on duty does not have to judge whether this is the right procedure.

Severity

The severity part gives a default and the conditions that raise it. Without them, every incident is argued from scratch.

First 15 Minutes

The first actions are numbered because the sequence does the work. Blocking sign-in comes first where the attacker is active or the account is privileged. The password is reset before sessions are revoked, so the attacker cannot sign in again with the password once the sessions are gone. Whatever the attacker added is removed next, and the sign-in logs then confirm that nothing from the attacker’s sources succeeds.

Evidence to Collect

The evidence part says what to collect and where it comes from. Each setting is recorded before it is removed, because deleting an inbox rule without noting it destroys the evidence of what the attacker wanted.

Questions to Answer

The questions define what the investigation must establish. An incident is not finished when the account is secure. It is finished when these have answers.

Containment Options

Each further action is paired with the role that can approve it. The wider the business impact, the more senior the approver.

Decision Points

The decision points list the choices that belong to people outside the technical team, so they are raised at the right time and not discovered later.

Communication

The communication part names who is told, in order of urgency. Without it, the user’s manager, finance, and the privacy lead hear at different times and from different people.

Exit Criteria

The exit criteria state what must be true before the incident is closed. Most repeat compromises trace back to a closure made without checking.

One Step, Expanded

A playbook stays short by pointing to runbooks for detail. Seeing one step expanded shows where the line falls.

The playbook above says, in the first 15 minutes: Revoke the account’s sessions. That is the right level for a playbook. It states the action and leaves the method to the tool.

The runbook behind it would cover the following, for the organization’s own identity platform.

Where to do it

The exact console, menu, and button, or the command, with a note on which administrator role is needed.

What it does

Invalidates refresh tokens and browser sessions for that user, so they must sign in again on every device.

What it does not do

Access tokens already issued can remain valid for up to about an hour for some services. Revocation also does nothing about a password the attacker still holds, which is why the reset comes first.

How to confirm it worked

The audit log entry to look for, and a check that the user’s next sign-in required authentication.

What the user will notice

They will be signed out of mail and applications on phone and laptop, and should be warned through a channel other than their own email.

How to undo it

Nothing to undo. The user signs in again.

Who to tell

The incident lead, with the time it was done.

A responder in a hurry reads the playbook line. A responder who has never done the task opens the runbook. The incident response plan sits above both.

DocumentPurposeChanges when
Incident response planRoles, authority, severity levels, and communication for any incident.The organization or its obligations change.
PlaybookThe response to one incident type, from trigger to closure.Threats, systems, or lessons from incidents change.
RunbookThe exact steps to perform one task in a specific tool.The tool or its configuration changes.

Keeping the documents apart means a tool change does not require rewriting every playbook.

Branches and Stop Points

Real incidents do not follow a straight line, and a playbook that pretends they do is abandoned at the first surprise. Good playbooks include a small number of explicit branches.

Take the compromised account playbook. After the first actions, the responder reviews what the session did. What they find changes the path.

The account is privileged

Stop treating it as a single-account incident. Assume tenant-wide settings may have been changed, and bring in the incident lead.

A payment request was sent from the mailbox

Start the payment fraud playbook in parallel. Finance must call the bank now.

The mailbox was synchronized or bulk-downloaded

Involve the privacy lead. A breach assessment is needed.

Other accounts signed in from the same source

This is a campaign. Repeat the first actions for each, and look for the phishing message.

The attacker reached internal systems or connected to the VPN

This is a network intrusion. Escalate to the major incident process.

None of these apply

Continue to exit criteria.

Each branch is one line, and each either hands off to another playbook or raises the level. This keeps a single playbook from trying to cover every possibility.

Stop points work the same way. A playbook should name the conditions under which the responder must pause and get a decision, such as before disabling an executive’s account, before contacting a customer, or before wiping a device that may hold evidence.

How Other Playbooks Differ

Every playbook keeps the same parts. The content changes with the incident.

IncidentFirst actionsKey decisionExit criteria
RansomwareIsolate affected systems and leave them powered on. Disconnect backups. Restrict administrator accounts. Move to a separate channel. Call responders, insurer, and counsel.Which systems to take offline, restore priorities, notification, and the response to an extortion demand.Entry point closed, credentials reset, restored systems validated, no attacker activity under monitoring.
Payment fraud by emailAsk the bank to recall the payment. Contain the mailbox. Hold related payments.Whether to warn customers and suppliers, and the breach assessment.Access removed, scope established, bank and affected parties informed.
Data theftPreserve transfer logs before they expire. Restrict the account or system. Identify the data it could reach.Whether a notification threshold is met, and what to tell affected people.Route closed, data identified as far as evidence allows, decisions recorded.
Malware on an endpointIsolate the device. Collect a triage package. Check for the same file elsewhere.Reimage or clean. Whether credentials used on the device need resetting.Device rebuilt, no other hosts affected, entry method understood.
Phishing email reportedRemove the message from every mailbox it reached. Block the sender and the link. Identify who clicked, replied, or opened the attachment.Whether any recipient entered credentials, which starts the compromised account playbook for each.Message removed, indicators blocked, every recipient who interacted handled.
Supplier compromiseSuspend the supplier’s access. Review its recent activity. Contact the supplier through a known route.When to restore access, and contractual notification.Supplier confirms containment, access restored with reviewed permissions.

Full detail for the first two is in the guides to ransomware and business email compromise. The notification decision in the data theft playbook is worked through in the data breach response plan guide.

A Second Example: Lost or Stolen Laptop

A short playbook shows that the same parts scale down. This one is used far more often than a ransomware playbook and is frequently handled inconsistently. Whether the loss is a notifiable breach is assessed with the test in the data breach response plan guide.

Trigger

An employee reports a laptop lost or stolen, or a device has been unaccounted for after travel.

Severity

Low if the disk is encrypted and the device was locked. Medium if encryption cannot be confirmed. High if the device belongs to an administrator or holds regulated data without encryption.

First 15 minutes

Record when and where the device was last seen. Confirm encryption status from the device management system. Mark the device as lost and issue a remote wipe. Reset the user’s password, then revoke the device’s sessions and its access to mail and files.

Evidence to collect

The device management record showing encryption and last check-in. The user’s account of what was on the device and whether it was locked. A police report number if stolen.

Questions to answer

Was the disk encrypted? Was the user signed in? What data was stored locally? Has the device connected to anything since?

Containment options

The security analyst may block the user’s sign-in until the password reset is confirmed. The incident lead may remove the device from the directory so that it cannot rejoin.

Decision points

If encryption cannot be shown, the privacy lead assesses whether the loss is a data breach.

Communication

The user’s manager. The privacy lead where data may be exposed. The insurer if a claim is likely.

Exit criteria

Device access revoked, wipe issued, credentials reset, encryption evidence recorded, and breach assessment completed where needed.

The whole playbook takes a few minutes to run. Its value is in the evidence step. An organization that can show a dated record of encryption has a simple answer to the breach question. One that cannot may have to assume the worst.

Writing Steps People Can Follow

Playbooks are read at 2 a.m. by someone under pressure. The wording decides whether they are followed.

Start each step with a verb

Revoke the account’s sessions, in place of Session revocation should be considered.

One action per step

A step that contains “and then” is two steps.

Name the role

The incident lead approves tells people who to call. “Approval is required” does not.

Say where

Name the system, the log, or the console.

State the condition

If the account is privileged, block sign-in is clearer than leaving it to judgment.

Give the reason once

A short note on why a step comes first helps people adapt when reality differs.

Compare the same instruction written both ways. A weak step reads: Consideration should be given to preserving relevant evidence prior to remediation. A strong one reads: Export the account’s sign-in records for the past 30 days and attach them to the incident log.

Which Playbooks to Write First

Write for the incidents most likely to happen and the ones that would hurt most.

1
Compromised user accountThe most frequent serious incident where cloud email is used.
2
Payment fraud by emailNeeds finance and IT to act together within hours.
3
RansomwareThe highest impact and the most decisions.
4
Data theft or accidental disclosureTriggers privacy assessment and notification.
5
Malware on an endpointCommon, and a test of evidence handling.
6
Lost or stolen deviceFrequent, and quick to write.
7
Supplier compromiseDepends on contracts and access outside the organization’s control.

A few playbooks that have been exercised are worth more than a library that has never been opened.

Automating a Playbook

Parts of a playbook can run automatically. Evidence collection is the safest place to start, since it changes nothing. Routine containment that can be reversed, such as revoking sessions, usually follows.

For each automated step, the playbook should state the condition under which it runs, who is told, and how it is undone.

Automate what the team already does the same way every time. A procedure that has never been run by hand should not be handed to automation.

Measuring Whether Playbooks Work

A playbook is a hypothesis about the best way to respond. Incident records show whether it is right.

Time to first action

From the trigger to the first containment step. A long gap means the trigger is unclear or the responder was unsure what to do.

Time to containment

From trigger to the point where the attacker can no longer act.

Steps skipped or reordered

Where responders departed from the text. Frequent departures mean the playbook does not match reality, and the playbook should change.

Escalations

How often the playbook handed off to a more serious process, and whether those hand-offs came early enough.

Reopened incidents

Cases that were closed and came back. These point to weak exit criteria.

Evidence gaps

Investigations that could not answer one of the playbook’s questions, and which log was missing.

Time spent per incident

A check on whether effort matches severity.

Review these after each significant incident and across all incidents every quarter. A few numbers, tracked over time, show more than a long report produced once.

The most useful signal is the simplest. If responders with different levels of experience reach the same result using the same playbook, it is working.

Keeping Playbooks Current

1
Walk through each playbookDo it against the real environment, with the people who would use it.
2
Check the permissionsConfirm responders hold the access each step needs.
3
Use them in exercisesRun a tabletop exercise against a playbook and note where the team departed from the text.
4
Update after every real useEach incident should produce at least one improvement.
5
Review on changeNew systems, suppliers, and tools all affect the steps.
6
Give each playbook an ownerThe owner approves each change and leads the walkthrough.

Playbooks With ThreatDefence

ThreatDefence prepares incident-specific runbooks as part of our incident response retainer. Under our 24x7 SOC service, containment follows playbooks that each customer has authorized, such as disabling a compromised Microsoft 365 account.

Each action type has its own setting: recommended for a person to decide, held for approval, or executed automatically. Every action has a documented rollback, so a step taken at 3 a.m. can be reversed in the morning.

Response actions in the ThreatDefence portal: revoke sessions, reset password, isolate endpoint and block address, each with its operating mode, status and rollback

Experiencing a breach?

Our incident response team answers 24x7, with or without a retainer.

Talk to Our Experts

Ask about incident-specific runbooks and a readiness review for your organization.

    I'm an MSP, MSSP, or reseller

    By submitting this form, you accept the terms in our privacy policy.

    Frequently Asked Questions

    Where should incident response playbooks be kept?

    Keep playbooks where responders already work, such as the case management or ticketing system, linked from the alert types that trigger them. Keep a second copy outside the corporate environment, printed or on a separate service, because a ransomware playbook stored only on the file server is unavailable when it is needed. Every copy should show its version, date, and owner.

    What is the difference between a playbook and a runbook?

    A playbook describes the response to one incident type from trigger to closure, and says who decides at each point. A runbook gives the exact instructions for a single task in a specific tool, such as revoking sessions in the identity platform. The playbook changes when the organization changes how it responds. The runbook changes when the tool does.

    How many incident response playbooks does an organization need?

    Start with a compromised user account, payment fraud by email, and ransomware. Then add data theft or accidental disclosure, malware on an endpoint, a lost or stolen device, and supplier compromise. Write a new playbook when the same kind of incident has been handled from memory twice. Each one needs an owner and a yearly walkthrough, so keep the set small enough to maintain.

    What should a ransomware playbook include?

    A ransomware playbook opens with the first actions: isolate affected systems and leave them powered on, disconnect backups, restrict administrator accounts, move to a separate channel, and call responders, the insurer, and counsel. It names who decides which systems go offline, the restore priorities, notification, and the response to an extortion demand. Its exit criteria are a closed entry point, reset credentials, validated restored systems, and no attacker activity under monitoring.

    What does a phishing playbook cover?

    A phishing playbook starts when a suspicious email is reported or detected. The first actions are to remove the message from every mailbox it reached, block the sender and any link, and identify who clicked, replied, or opened an attachment. For each person who entered credentials, the compromised account playbook starts. The incident closes when the message is gone and every recipient who interacted has been handled.

    Who should write incident response playbooks?

    Security and IT practitioners should develop the technical procedures with input from the people responsible for business decisions. Legal, privacy, communications, and executive representatives should validate the decisions and notifications relevant to their responsibilities. The people expected to execute the playbook should test it before use.

    Can incident response playbooks be automated?

    Yes, in part. Start with evidence collection, because it changes nothing, then add containment that can be reversed, such as revoking sessions. Each automated step needs a stated condition, a notification, and a way to undo it. Decisions with business or legal consequences, such as disabling an executive’s account or contacting a customer, stay with a named person.

    How often should playbooks be reviewed?

    Review each playbook at least once a year, after every incident in which it was used, and whenever a system, tool, or supplier it names changes. Check the measures each quarter. Steps skipped or reordered, reopened incidents, and evidence gaps show where the text no longer matches reality. A walkthrough with the people who would use it catches steps that can no longer be carried out as written.