CertPrepNow
CrowdStrikeCCFA-200b6 domains

CCFA-200b Exam Notes

Last-minute traps, must-know facts, and scenario tips for the CrowdStrike Certified Falcon Administrator exam.

General Exam Tips

  • 1.Read ALL answer options before choosing — many questions are designed so the first plausible option is wrong, and the correct one requires understanding nuance
  • 2.No penalty for wrong answers — never leave a question blank; guess if you must
  • 3.Drag-and-drop and multi-select questions count as single questions — read the instructions carefully before committing your answer
  • 4.When stuck between two answers, ask: does this preserve telemetry? Does this follow least privilege? The option that maintains visibility and uses minimum permissions is usually correct
  • 5.Policy questions almost always hinge on 'first match wins, not most restrictive wins' — watch for answer options that incorrectly say the more restrictive or more specific policy applies
  • 6.Any question about remote access to an endpoint requires BOTH conditions to be true: the right policy setting AND the right user role — if the question omits one, the answer is that the missing condition is the problem
  • 7.Dashboard and UI-specific questions trip up many candidates — study the exact console section names (Activity, Investigate, Configuration, Support and Resources) and what lives under each
  • 8.When a question describes a false positive scenario, the correct answer always involves creating an exclusion, not just closing the alert
  • 9.Sensor questions that mention 'not working' or 'limited' on Linux almost always point to kernel compatibility and RFM, not a misconfigured CID
Domain 125% of exam

Policy Configuration and Management

Must-Know Facts

  • Cloud ML and sensor ML are completely independent slider settings — each has its own levels (Disabled, Cautious, Moderate, Aggressive, Extra Aggressive) and each must be configured separately
  • When a host is offline, ONLY sensor ML can analyze files — cloud ML requires an active CrowdStrike cloud connection
  • Policy precedence = the first policy in the ordered list whose host group matches the host — NOT the most restrictive, NOT the most specific, the first match wins
  • Every policy type has a default policy that catches hosts not covered by any other policy assignment — always configure defaults with at least a baseline security posture
  • Containment blocks ALL outbound traffic except to CrowdStrike cloud — IP exclusions must be set BEFORE containment if critical services (DNS, SIEM) need to remain reachable
  • Device control targets USB mass storage class by default — keyboards, mice, and audio USB devices are NOT blocked unless explicitly configured
  • Firewall rules are evaluated top-down with first match wins — a broad ALLOW above a narrow DENY permits the traffic
  • Response policy enables RTR on the endpoint side — but RTR role must also be assigned to the user; both are required (see user-access-management domain for RTR role details)
  • Sensor update policy stages: N (latest, test environments), N-1 (standard production), N-2 (critical/conservative servers)

Common Traps

TrapThe more restrictive policy applies when a host matches multiple host groups
RealityCrowdStrike uses strict precedence ordering. The policy with the highest precedence (lowest number / first in the list) applies. The most restrictive setting does NOT win — order wins.
TrapSetting all production endpoints to N (latest) sensor version keeps you most secure
RealityN (latest) with no staging means an untested sensor version may be pushed to all production endpoints simultaneously. Best practice is N for test groups, N-1 for general production, N-2 for critical servers.
TrapContainment also blocks the host from reaching CrowdStrike cloud
RealityContained hosts ALWAYS maintain their connection to the CrowdStrike cloud for ongoing management and telemetry. Only other network traffic is blocked.
TrapDevice control policies block all USB devices from connecting
RealityDevice control only applies to USB mass storage class by default. Keyboards, mice, headsets, and other USB HID devices continue to function normally unless additional USB classes are explicitly configured.
TrapCloud ML and sensor ML are the same engine at different sensitivities
RealityThey are completely different systems. Cloud ML runs analysis in CrowdStrike infrastructure (requires internet). Sensor ML runs locally on the endpoint (works offline). Air-gapped or intermittently connected hosts depend entirely on sensor ML.
TrapOnce you enable a new prevention setting, it immediately protects at full strength
RealityAlways deploy new prevention features in Detect mode first. Without a tuning period, enabling Prevent on a misconfigured setting can block legitimate business applications immediately.

Confusing Pairs

Detect ModePrevent Mode

Detect = alerts generated but execution is NOT blocked. Prevent = execution is actively stopped. Newly enabled settings should always start in Detect, then switch to Prevent after confirming no false positives.

Policy Precedence (first-match)Most Restrictive Wins

CrowdStrike uses first-match-wins for policy resolution. Most restrictive wins is a concept from other security tools (like firewall rules in some systems). In Falcon, the admin controls precedence by reordering the policy list — and the first match is the only one applied.

N-1 Sensor UpdateN-2 Sensor Update

N-1 = one version behind latest, standard for general production environments, allows validation time. N-2 = two versions behind, for the most conservative/critical servers. Neither is for test environments — use N (latest) for test groups.

Sensor Update PolicyPrevention Policy

Sensor update policy controls WHICH VERSION of the Falcon sensor runs on endpoints. Prevention policy controls WHAT the sensor detects and blocks. They are separate policy types, both assigned to host groups independently.

Scenario Tips

If the question asks about:

A host belongs to the 'Servers' group (Policy A, ranked #1) and the 'Critical' group (Policy B, ranked #3). The question asks which policy applies.

Answer:

Policy A applies because it has higher precedence (rank #1). Rank position is the only deciding factor — not the name, not restrictiveness, not group specificity.

Distractor to avoid:

Policy B because Critical servers should have stricter policy. Wrong — precedence is determined by list position, not semantics.

If the question asks about:

A laptop goes offline and encounters a never-before-seen malware sample. Cloud ML is set to Aggressive, sensor ML to Moderate. What happens?

Answer:

Sensor ML at Moderate analyzes the file. Cloud ML is unavailable — it requires CrowdStrike cloud connectivity. The sensor provides the only protection.

Distractor to avoid:

Cloud ML at Aggressive level because it is the highest configured. Wrong — cloud ML requires connectivity; offline hosts can only use sensor ML.

If the question asks about:

A contained host needs to reach the internal ticketing system at 192.168.1.100 during incident response. How is this achieved?

Answer:

Add 192.168.1.100 as a containment IP exclusion BEFORE containing the host. IP exclusions allow specific IPs to remain reachable during network isolation.

Distractor to avoid:

Lift containment, add the rule, then re-contain. This creates a gap in isolation. IP exclusions should be pre-configured before initiating containment.

If the question asks about:

A new script-based attack is being enabled in Prevent mode before any tuning. What is the correct practice?

Answer:

Enable in Detect mode first. Monitor for false positives over a tuning period. Only switch to Prevent after confirming legitimate tools are not affected.

Distractor to avoid:

Enable in Prevent mode immediately since Detect doesn't stop attacks. Wrong — false positives in Prevent mode can block critical business applications.

Last-Minute Facts

1Policy precedence: 5 policy types (prevention, sensor update, device control, firewall, response) each have their own ordered list
2Default policy is a common exam trap: a host not matching any host-group assignment still gets the default — leaving defaults at low security is an instant coverage gap
3Sensor update notation: N = latest, N-1 = previous, N-2 = two back
4Containment always preserves CrowdStrike cloud connection — this is by design
5USB device control default scope: mass storage class only (NOT all USB devices)
Domain 220% of exam

Sensor Deployment and Management

Must-Know Facts

  • Windows sensor requires no reboot after installation — protection begins immediately
  • macOS requires a system extension approval (SEXT) via MDM profile deployed BEFORE the sensor — without it, the sensor runs in Reduced Functionality Mode
  • Linux requires kernel compatibility verification before deployment — the sensor version must support the running kernel; incompatible kernels trigger RFM
  • Uninstalling the sensor requires an uninstall token retrieved from the Falcon console — this anti-tamper protection prevents adversaries from disabling endpoint coverage
  • CID (Customer ID) includes a checksum suffix — the full CID+checksum is required during installation for the sensor to register to the correct tenant
  • RFM (Reduced Functionality Mode) means the sensor is installed and running but with limited capability — it is NOT the same as the sensor being offline or unregistered
  • Large-scale deployment tools: GPO, SCCM/MECM, Intune (Windows), Jamf/Intune (macOS), Ansible/Puppet/Chef (Linux)

Common Traps

TrapWindows endpoints require a reboot after sensor installation to begin protecting
RealityThe Windows Falcon sensor begins protecting immediately after installation with no reboot required. This is a key differentiator from traditional AV products.
TrapRFM means the sensor is not installed or not communicating
RealityRFM (Reduced Functionality Mode) means the sensor IS installed and IS communicating, but cannot provide full protection. On Linux, the most common cause is an unsupported kernel version. On macOS, an unapproved system extension is common.
TrapThe CID alone is sufficient for sensor registration
RealityThe CID includes a checksum that must be appended. The full CID+checksum string is what registers the sensor to your specific tenant.
TrapYou can remove the Falcon sensor using standard OS uninstall methods
RealityThe sensor has anti-tamper protection. Removal requires an uninstall token from the Falcon console. Without the token, the uninstall command fails.
TrapmacOS sensor works fully as soon as the package is installed
RealitymacOS enforces system extension approval. If the MDM profile for SEXT approval was not deployed beforehand, the sensor operates in RFM. The MDM profile must be deployed BEFORE the sensor installer runs.

Confusing Pairs

Reduced Functionality Mode (RFM)Sensor Offline

RFM = sensor is installed, running, and checking in with the cloud, but cannot enforce full protection (commonly from kernel incompatibility on Linux or unapproved SEXT on macOS). Offline = sensor cannot reach the CrowdStrike cloud at all, last seen timestamp goes stale.

CIDCCID

CID = Customer ID without the checksum (tenant identifier). CCID = CID + checksum suffix (the full string used during sensor installation). The exam may use both terms — know that the full installation parameter is the CCID.

macOS System Extension ApprovalWindows Silent Install

macOS requires MDM profile deployment before sensor installation to pre-approve the system extension — user interaction or manual approval is not needed with MDM, but the profile must exist first. Windows installation has no OS-level gating mechanism; it runs silently with a CID parameter.

Scenario Tips

If the question asks about:

Multiple Linux hosts appear in the console with RFM status after a fresh sensor deployment. What is the most likely cause?

Answer:

The running Linux kernel version is not supported by the installed sensor version. RFM on Linux is almost always a kernel compatibility issue. Check the sensor support matrix and either update the kernel or downgrade the sensor version.

Distractor to avoid:

The CID was entered incorrectly. Wrong — an incorrect CID would prevent registration entirely; the host would not appear in the console at all.

If the question asks about:

An IT team deploying to macOS via Jamf reports sensors are installing but operating in RFM. What was missed?

Answer:

The PPPC and System Extension MDM profiles were not deployed before the sensor installation. The MDM profile must be pushed via Jamf BEFORE the sensor is installed so macOS auto-approves the extension.

Distractor to avoid:

A reboot is required on macOS after installation. Wrong — no reboot is required. The issue is the missing MDM approval profile.

If the question asks about:

An administrator tries to uninstall the sensor from a Windows laptop but the command fails with an access denied error. What is needed?

Answer:

An uninstall token generated from the Falcon console (Support and Resources or Host Management section). The token must be provided as a parameter to the uninstall command to bypass the anti-tamper protection.

Distractor to avoid:

Run the uninstall as a local administrator. Wrong — even with local admin rights, the anti-tamper mechanism requires the console-generated token.

Last-Minute Facts

1Windows: no reboot required, admin rights required, no .NET dependency
2macOS: MDM profile (SEXT approval) BEFORE sensor install; no reboot required
3Linux: verify kernel compatibility BEFORE deployment; no reboot required
4All platforms: no reboot required after installation
5RFM root causes: Linux = unsupported kernel; macOS = unapproved system extension
6Uninstall token is an anti-tamper control — generated from the Falcon console, required for sensor removal
Domain 315% of exam

User and Access Management

Must-Know Facts

  • RTR Active Responder and RTR Administrator are different roles — Administrator is a superset with additional write capabilities including custom scripts, file uploads, and put commands
  • Roles are additive — a user assigned both Analyst and Administrator roles has the combined permissions of both
  • RTR requires TWO conditions: the response policy must have RTR enabled AND the user must have an RTR role — one condition alone is insufficient
  • API clients use OAuth2 and have granular scopes (read vs. write per resource type) — a client with Hosts:Read cannot modify host groups; it needs Host Groups:Write
  • SSO/SAML: always maintain at least one local administrator account for break-glass access — if the IdP goes down, all SSO users lose access
  • API client scopes follow least privilege — assign only the scopes the client actually needs, not a broad 'admin' scope

Common Traps

TrapRTR is available to any user as long as the response policy has RTR enabled
RealityRTR requires BOTH: (1) RTR enabled in the response policy, AND (2) the user must have either RTR Active Responder or RTR Administrator role. A user with only Falcon Analyst role cannot initiate RTR sessions even with RTR policy enabled.
TrapRTR Active Responder can run custom scripts on endpoints
RealityCustom script execution requires RTR Administrator role. RTR Active Responder can only run built-in read-only and basic remediation commands. Custom scripts and file uploads (put command) are exclusively RTR Administrator capabilities.
TrapConfiguring SSO eliminates the need for local user accounts
RealityAt least one local administrator account must be maintained as a break-glass account. If the identity provider goes down, SSO authentication fails for all users, and only local accounts can log in.
TrapAn API client with Administrator-level scope has access to all API endpoints
RealityThere is no catch-all 'Administrator' API scope. API client access is defined by specific resource-level scopes (e.g., Hosts:Read, Host Groups:Write, Prevention Policies:Read/Write). Each scope is independently assigned.

Confusing Pairs

RTR Active ResponderRTR Administrator

Active Responder = read-only commands and basic remediation (ls, cat, ps, netstat, kill process, delete file). Administrator = everything Active Responder can do PLUS custom scripts, file uploads (put), and advanced write operations. Know which capabilities require which role.

Falcon AnalystFalcon Investigator

Analyst = triage detections, view host details, basic investigation. Investigator = deeper investigation including Event Search, raw telemetry, and process timelines. Neither can modify policies or manage users. Investigator has broader visibility than Analyst.

API Client Read ScopeAPI Client Write Scope

Read scope = view/query resources only. Write scope = create, modify, delete resources. Each resource type (Hosts, Host Groups, Prevention Policies, etc.) has separate read and write scopes. Automation that only reads data needs read scope; automation that manages resources needs write scope.

Scenario Tips

If the question asks about:

A user reports they can view detections and investigate alerts but cannot start an RTR session on a host. RTR is enabled in the response policy. What is the cause?

Answer:

The user does not have an RTR role (Active Responder or Administrator) assigned to their account. Both the policy setting and the user role are required — the policy alone does not grant access.

Distractor to avoid:

The host must be contained first before RTR can be used. Wrong — containment is not a prerequisite for RTR.

If the question asks about:

A script needs to automate creation of new host groups based on CMDB data. What API scope is needed?

Answer:

Host Groups: Write scope. Creating or modifying host groups requires write access to the Host Groups resource. Read scope only allows querying existing groups.

Distractor to avoid:

Hosts: Write scope. Wrong — the Hosts scope controls host management operations, not host group management.

If the question asks about:

After enabling SSO for all users, the identity provider suffers an outage. No one can log in to the Falcon console. What should have been done?

Answer:

Maintain at least one local administrator account as a break-glass account. This account bypasses SSO and allows console access during IdP outages.

Distractor to avoid:

Configure a backup SAML certificate. Wrong — a backup certificate does not help if the IdP infrastructure is down.

Last-Minute Facts

1RTR requires: policy with RTR enabled + user with RTR role (both required)
2RTR Active Responder: read commands + basic remediation only
3RTR Administrator: everything Active Responder can do + custom scripts + file uploads
4Roles are additive — multiple roles combine their permissions
5API scopes are per-resource and per-action (read vs write) — no global admin scope
6SSO: always keep at least one local admin account for break-glass
Domain 415% of exam

Platform Navigation and Core Functionality

Must-Know Facts

  • Four main console sections: Activity (detections dashboard), Investigate (deep investigation and event search), Configuration (all policies, host groups, exclusions, allowlists), Support and Resources (docs, audit logs, API docs)
  • Activity shows DETECTIONS and security events — it does NOT show all raw sensor telemetry; raw telemetry is in Investigate > Event Search
  • Host Management shows all registered endpoints: sensor version, OS, last seen timestamp, applied policies, host group membership — use it to find stale/offline sensors
  • The process tree in a detection highlights in red the specific node where malicious behavior was detected — this is the detection point, NOT necessarily the root cause or the first infected process
  • Detection severity (Informational, Low, Medium, High, Critical) is set by CrowdStrike Intelligence — administrators cannot change the severity of a detection
  • Searching by hash in Investigate shows ALL endpoint activity related to that file across the organization — more comprehensive than Activity > Detections which only shows detection events

Common Traps

TrapThe Activity dashboard shows all sensor telemetry events
RealityActivity shows detections and security-relevant alerts, not all raw telemetry. For raw event data (process executions, network connections, file writes), use Investigate > Event Search.
TrapA stale 'last seen' timestamp on a host means the host is compromised or the sensor was removed
RealityA stale last seen timestamp simply means the host has not checked in with the CrowdStrike cloud recently. Common causes: the host is powered off, offline, or on a network without cloud access. It does not indicate compromise.
TrapThe red-highlighted node in the process tree is where the attacker first gained access
RealityThe red node is where the specific malicious BEHAVIOR was detected by the Falcon sensor. It may be deep in a chain where earlier processes (like explorer.exe or cmd.exe) are legitimate parents. The red node marks detection, not necessarily patient zero.
TrapAdministrators can adjust detection severity if they disagree with the CrowdStrike assessment
RealityDetection severity is fixed and set by CrowdStrike Intelligence based on threat analysis. Admins can tune which detections appear (via exclusions) or how sensitive the policies are, but cannot change the severity label on an existing detection.

Confusing Pairs

Activity > DetectionsInvestigate > Event Search

Activity Detections = organized, severity-labeled security alerts ready for triage. Event Search = raw sensor telemetry (all process starts, network connections, file writes). Use Activity for triage workflows; use Event Search for forensic investigation of specific activity.

Investigate (hash search)Configuration > Allowlists (hash entry)

Investigate hash search = find all historical activity related to a file hash across the environment (read-only investigation). Allowlists hash entry = configure the platform to permanently allow that hash everywhere (write action with security implications). One is investigation, one is a configuration action.

Host Management (last seen / sensor health)Activity > Detections (alert triage)

Host Management = operational view of endpoint inventory — sensor version, OS, last seen timestamp, RFM status, host group membership. Use it to find stale, offline, or out-of-date sensors. Activity > Detections = security view — alerts generated by the sensor's threat engines. A host can appear healthy in Host Management but have open Critical detections in Activity. Exam scenarios often present one section as the answer to a question that clearly belongs to the other.

Scenario Tips

If the question asks about:

A CISO wants to see all High and Critical detections from the past 48 hours for a board report. Where in the console should they look?

Answer:

Activity > Detections, filtered by severity (High, Critical) and time range (48 hours). This section is designed for detection triage and management with filtering capabilities.

Distractor to avoid:

Investigate > Event Search. Wrong — Event Search shows raw telemetry, not organized detection alerts. It is for forensic investigation, not executive reporting.

If the question asks about:

A threat analyst wants to find every endpoint that has executed a specific suspicious file (known SHA256) in the past 30 days across the entire organization.

Answer:

Investigate > search by hash. Investigate shows all endpoint activity related to the hash across the org, including hosts where the file appeared, execution history, and any related detections.

Distractor to avoid:

Activity > Detections filtered by hash. Partial — this only shows detections, not all occurrences where the file may have run without triggering a detection.

Last-Minute Facts

1Console sections: Activity (alerts) | Investigate (search + raw telemetry) | Configuration (policies, groups, exclusions) | Support and Resources (docs, audit logs)
2Host last seen = last cloud check-in — stale does NOT mean compromised
3Process tree red node = where behavior was DETECTED, not where attacker entered
4Detection severity is read-only for admins — set by CrowdStrike Intelligence
5Audit logs are under Support and Resources — track who changed what and when
Domain 515% of exam

Detection and Prevention

Must-Know Facts

  • IOA (Indicator of Attack) = behavioral-based — detects malicious patterns and TTPs regardless of specific malware. Custom IOA rules let admins define organization-specific behavioral detections
  • IOC (Indicator of Compromise) = artifact-based — matches specific known SHA256 hashes, domains, or IP addresses uploaded to Falcon
  • Custom IOC action options: 'Detect' generates an alert but allows execution; 'Prevent' blocks execution. Always test with Detect before switching to Prevent
  • Marking a detection as 'false positive' ONLY closes that alert — it does NOT prevent future detections of the same activity. A separate exclusion must be created
  • Releasing a quarantined file returns it to its original location — add it to the allowlist BEFORE releasing to prevent immediate re-quarantine
  • Three exclusion types in order of impact: IOA exclusions (suppress behavioral alerts), ML exclusions (suppress machine learning verdicts), Sensor Visibility exclusions (eliminates ALL telemetry — use as last resort)

Common Traps

TrapMarking a detection as false positive prevents it from recurring
RealityFalse positive status only closes that specific detection instance. The underlying activity continues to trigger new detections until an appropriate exclusion (IOA, ML) or allowlist entry is created.
TrapIOA exclusions and ML exclusions do the same thing
RealityIOA exclusions suppress behavioral/pattern-based detections (Indicators of Attack). ML exclusions suppress machine learning verdicts (cloud ML and sensor ML). If a file triggers both IOA and ML detections, you need BOTH types of exclusions to fully suppress alerts on it.
TrapA custom IOC set to 'Detect' quarantines matching files immediately
RealityDetect mode only generates an alert — it does NOT block or quarantine. The file executes normally. Only 'Prevent' action stops execution. Custom IOCs in Detect mode are for monitoring and tuning before committing to blocking.
TrapSensor visibility exclusions are the preferred way to handle false positives
RealitySensor visibility exclusions eliminate ALL telemetry from a path/process — the sensor becomes blind. This is the last resort option for severe performance issues only. For false positives, use IOA or ML exclusions which suppress specific alert types while preserving telemetry.
TrapCustom IOA rules require selecting a specific malware signature
RealityCustom IOA rules are behavioral — they match process behavior patterns (e.g., PowerShell downloading and executing, process spawning unexpected child processes). They do not require a known malware hash or signature.

Confusing Pairs

IOA (Indicator of Attack)IOC (Indicator of Compromise)

IOA = behavioral detection based on TTPs (how attackers behave). Detects unknown threats and fileless attacks. IOC = artifact matching (specific SHA256, domain, IP). Detects known bad items. The exam will distinguish between 'we see suspicious behavior' (needs IOA) vs 'we have a known-bad hash' (needs IOC).

IOA ExclusionML Exclusion

IOA exclusion = tells Falcon to ignore a specific behavior/pattern for a file or path. ML exclusion = tells cloud ML and sensor ML engines to stop flagging a specific file. A file causing both behavioral and ML alerts needs both exclusion types. They target different detection engines.

Custom IOA RuleCustom IOC Upload

Custom IOA rule = define a behavioral pattern to detect (process spawning, file write, registry change). Catches unknown threats. Custom IOC = upload a known-bad hash/domain/IP with an action. Matches known artifacts. IOA is proactive behavioral; IOC is reactive artifact-based.

Scenario Tips

If the question asks about:

An admin marks a detection as false positive but the same application triggers the same detection the next day. What is missing?

Answer:

An exclusion was not created. Marking false positive only dismisses that specific alert instance. To prevent recurring alerts, create an IOA exclusion (if behavioral) or ML exclusion (if machine learning flagged it) for the application.

Distractor to avoid:

The sensor needs to be updated for the false positive to take effect. Wrong — false positive status is a workflow marker, not a configuration change.

If the question asks about:

You need a rule that alerts when any process matching 'cmd.exe' spawns 'powershell.exe' with an encoded command. Which feature handles this?

Answer:

Custom IOA rule. This is a behavioral pattern (parent-child process relationship with suspicious arguments). Custom IOA rules detect TTPs without requiring a specific file hash.

Distractor to avoid:

Custom IOC with a SHA256 hash of powershell.exe. Wrong — hashing powershell.exe would block all PowerShell usage, and IOCs match specific files, not behavioral patterns.

If the question asks about:

A legitimate IT tool is being repeatedly quarantined by sensor ML. You want the tool to stop generating ML alerts but still have the sensor monitor its file activity. Which exclusion type is correct?

Answer:

ML exclusion. This suppresses the machine learning verdict for the specific file while the sensor continues to collect telemetry. Sensor visibility exclusion would eliminate all monitoring of the file — that is too aggressive for this case.

Distractor to avoid:

Sensor visibility exclusion for better performance. Wrong — sensor visibility exclusions eliminate all telemetry from the sensor, which eliminates investigation capability if the tool is later compromised.

Last-Minute Facts

1IOA = behavioral (TTPs). IOC = artifact (hash, domain, IP)
2Custom IOC actions: Detect (alert only, execution allowed) vs Prevent (blocks execution)
3False positive marking does NOT create an exclusion — they are separate actions
4Release quarantined file AFTER adding it to the allowlist — not before
5Exclusion priority for false positives: IOA or ML exclusion first; sensor visibility exclusion only for performance emergencies
6Custom IOA rules: always deploy in Detect mode first, switch to Prevent only after confirming no false positives
Domain 610% of exam

Reporting and Administration

Must-Know Facts

  • Audit logs track all administrative actions in the Falcon console: who made changes, which object was changed, and when — used for compliance accountability
  • Scheduled reports can be configured with specific content, frequency, and email recipients — they continue running even if a recipient leaves the organization until manually deleted
  • Dashboard data is subject to the platform's retention period — data older than the retention window is not visible; Falcon Data Replicator (FDR) is required for long-term raw event retention
  • Sensor health monitoring: use Host Management to find offline sensors, RFM hosts, outdated sensors, and coverage gaps
  • Notification thresholds should be tuned — alerting on every low-severity detection creates noise and alert fatigue

Common Traps

TrapAudit logs show both who made a change and what the setting was before the change
RealityAudit logs capture who performed an action and when, but do not always capture the previous/before value. Document your baseline configuration separately if rollback capability is important.
TrapHistorical detection data is always accessible in the Falcon dashboard
RealityDashboard and detection data is limited by the platform's data retention period. Data older than the retention window requires Falcon Data Replicator (FDR) to have been configured to export to external long-term storage.
TrapFalcon Data Replicator (FDR) is used to generate PDF reports for executives
RealityFDR streams raw sensor event data to external storage (AWS S3, Azure Blob) for SIEM integration and long-term retention. Executive PDF/email reports use the Scheduled Reports feature, not FDR.

Confusing Pairs

Scheduled ReportsFalcon Data Replicator (FDR)

Scheduled Reports = formatted summaries delivered via email on a recurring schedule (daily, weekly, monthly). For humans. FDR = raw event data stream exported to cloud storage for SIEM ingestion and long-term data retention. For machines/analysts doing deep queries. Different purposes, different audiences.

Audit LogsDetection Logs (Activity)

Audit Logs = record of admin actions in the console (who changed a policy, who created a user, who modified an exclusion). Activity Detections = security events generated by endpoint threats. Compliance teams want audit logs; SOC analysts want detection logs.

Scenario Tips

If the question asks about:

A compliance officer asks who changed a prevention policy configuration on a specific date. Where should the Falcon administrator look?

Answer:

Audit logs under Support and Resources. Audit logs record all administrative actions including policy changes, with the user, action type, and timestamp.

Distractor to avoid:

Activity > Detections. Wrong — this shows security events, not administrative change history.

If the question asks about:

A CISO wants automatic weekly email reports on detection trends without logging into the console. What is the correct configuration?

Answer:

Configure a Scheduled Report with detection summary content, weekly frequency, and the CISO's email as the recipient. This delivers automated reports without requiring console access.

Distractor to avoid:

Set up FDR to stream data to the CISO's email. Wrong — FDR streams raw event data to cloud storage, not formatted email reports.

Last-Minute Facts

1Audit logs: who + what + when for ALL admin actions; find in Support and Resources
2Scheduled reports: continue emailing even after recipients leave the org until deleted
3FDR = raw events to S3/Azure Blob for SIEM — not for email reports
4Host Management last seen filter = fastest way to find stale/offline sensors
5Dashboard retention is limited — FDR required for data beyond the retention window

Feeling confident?

Put your knowledge to the test with a timed CCFA-200b mock exam.