top of page
40051abd5a76713af8f015988fc6780e-blue-phone-icon-with-a-wave-on-it.webp
Logo-Zalo-Arc.webp
Facebook_Messenger_logo_2020.svg.webp
whatsapp

EDR and XDR in SOC: From endpoint alert to response

Nov 10
10 min read

Enterprises have deployed EDR or XDR, but security teams still spend a lot of time verifying alerts, contacting the IT department, and deciding whether to isolate devices? The issue may not lie in the tool's detection capabilities, but in how technology is connected to cybersecurity operational processes.

EDR/XDR in SOC plays the role of providing data and detection signals, helping cybersecurity teams correlate events, verify threats, and execute responses based on risk levels. However, to convert an endpoint alert into a handled incident, enterprises need to synchronously establish six components: telemetry, correlation, triage, containment, escalation, and KPIs.

What role do EDR and XDR play in the SOC operational process?

In the enterprise security model, Endpoint Detection and Response (EDR) monitors endpoint activities to detect, investigate, and facilitate response to suspicious behaviors. Extended Detection and Response (XDR) expands the capability to analyze and correlate signals from multiple security layers, depending on the architecture and data sources supported by the platform.

Meanwhile, the Security Operations Center (SOC) is the operational department responsible for monitoring, analyzing, coordinating, and tracking the cybersecurity incident response process.

Therefore, integrating EDR and XDR into the SOC is not simply connecting two systems via API. The goal is to build a process so that detection signals are transformed into decisions and actions with accountable individuals.

edr-xdr-in-soc
EDR/XDR in SOC operation flow

A typical operational flow includes:

  1. Endpoints generate data regarding processes, files, network connections, or user activities.

  2. EDR/XDR collects and analyzes the data, generating alerts or grouping related signals.

  3. The SOC receives the information and adds context from other systems.

  4. Analysts verify the risk level, scope of impact, and priority.

  5. The SOC coordinates to execute containment measures, escalate, and monitor remediation.

  6. The incident's outcome is verified, the case is closed, and it is included in operational efficiency reports.

In reality, some XDR platforms can perform correlation and response directly before forwarding information to the SOC. In other systems, SIEM or SOAR is used to support these steps. Therefore, enterprises do not necessarily have to push all raw data through all platforms.

If there is a need to understand the overall structure and functions of a cybersecurity operations center, enterprises can refer to the article What is a SOC and how does it work. With a multi-source detection architecture, the IPSIP Vietnam’s XDR solution page provides additional information on the scope of technology application.

Endpoint telemetry: What data does the SOC need to receive?

Endpoint telemetry is the recorded information of activities on workstations, servers, and monitored endpoints. This is a critical data source for the SOC to determine what happened on the device before, during, and after an alert appears.

Groups of data of interest include:

  • Process telemetry: Process name, execution path, command line, parent process, and child process.

  • File activity: Files created, modified, or executed; file hashes and malware detection results.

  • Network activity: Inbound and outbound connections, IP addresses, ports, domains, and the process initiating the connection.

  • User activity: Login accounts, sessions, and activities related to access privileges.

  • Detection metadata: Alert code, detection time, severity level, involved devices, and protective actions taken.

  • Agent health: Operational status, agent version, last data submission time, and collection errors.

Not all platforms provide all of the above data fields across every operating system or license tier. Thus, the first step when integrating EDR/XDR into the SOC is to identify the data that can realistically be collected and forwarded.

Checking data connection capabilities

Enterprises need to verify supported integration methods, such as APIs, native connectors, event streaming, or mechanisms for forwarding alerts to a SIEM/SOAR.

Four issues that should be checked before moving into official operation are:

  • Completeness: Does the SOC receive the necessary data fields to verify alerts, or only the threat name and severity?

  • Consistency: Are device IDs, user IDs, and timestamps standardized to correlate events across systems?

  • Latency: How long does it take for alerts to reach the SOC? Is there a mechanism to retrieve data if the connection is interrupted?

  • Coverage scope: How many devices within the protection scope have active agents sending valid data?

For example, a suspicious PowerShell alert will be difficult to investigate if the SOC only receives the process name but lacks the command line, execution account, or relationship to the parent process.

Conversely, collecting all types of telemetry without a usage plan can increase storage, processing, and data management costs. Enterprises should determine the necessary level of data for each detection use case instead of defaulting to maximum collection.

Correlation: From endpoint alert to attack context

A single isolated event is not always sufficient to conclude that the system is under attack. For instance, PowerShell is a legitimate administrative tool, but it can also be abused in a malicious behavioral chain.

Correlation is the process of linking multiple related events based on devices, users, time, processes, or technical indicators.

When deploying EDR XDR in the SOC, the correlation capability helps analysts see a chain of behaviors rather than having to inspect each alert in isolation.

Illustrative scenario

Suppose the system records the following events on a workstation:

  1. An account logs in from an unusual location or device.

  2. After logging in, a PowerShell process is launched with suspicious parameters.

  3. This process connects to a domain that has never appeared in the device's activity history.

  4. The endpoint continues to generate behaviors accessing sensitive data.

If looking at each event individually, the SOC may find it difficult to determine the severity. When correlating them, the analyst has a stronger basis to hypothesize that the account has been compromised or the device is executing an anomalous chain of activities.

The SOC can use EDR data alongside identity logs, DNS, firewalls, or cloud audit logs to test the hypothesis.

However, correlation does not automatically prove an attack has occurred. Legitimate administrative activities can still generate similar event chains. Therefore, correlation results need to move to the triage step for verification.

To build technically grounded use cases, enterprises can refer to the MITRE ATT&CK Data Components, which describes data components like process creation and network connection creation for behavioral detection.

Triage and enrichment: How does the SOC verify alerts?

Triage is the process of initial screening and assessment to determine which alerts require deep investigation or immediate response. Enrichment adds information so the analyst better understands the subject, behavior, and scope of impact.

This is the decisive stage in determining whether an alert from EDR/XDR becomes an incident that needs handling.

Step 1: Verify the detection signal

The SOC inspects the detection rule, the triggering behavior, the time of occurrence, and the raw data used to generate the alert.

Specialists must distinguish truly suspicious behavior from authorized administrative activities, software updates, or automated tasks.

Step 2: Add context

An alert on a test machine may require a different prioritization approach than the same behavior appearing on a server containing production data.

Enrichment information should include:

  • The role and criticality of the device.

  • The user or administrator of the device.

  • History of related alerts.

  • Indicators such as file hashes, domains, or IPs.

  • Threat intelligence information and behavioral analysis results.

  • Approved system changes.

Step 3: Evaluate severity and priority

Severity reflects the technical seriousness of the alert or incident. Priority dictates the urgency of handling based on actual risk, potential for lateral movement, and business impact.

These two metrics are not always identical. A medium severity technical alert on a critical system may require higher priority than a critical alert that has already been blocked on a test device.

Step 4: Decide on the handling direction

After triage, the SOC can classify alerts as legitimate activities, false positives, events lacking sufficient evidence, or confirmed incidents requiring a response.

Cases lacking sufficient evidence must be monitored or escalated for further investigation, and should not be automatically closed just to reduce the alert volume.

NIST SP 800-61 Revision 3, published in April 2025, recommends integrating incident response activities into cybersecurity risk management, including detection, response, and recovery. This reinforces the need to link triage with the overall handling process rather than treating it as an isolated action on a dashboard.

Enterprises wishing to delve deeper into the investigation process can refer to the SOC incident investigation handbook for L1/L2/L3.

The practical value of reducing false positives

In the outsourced XDR deployment case study, IPSIP Vietnam announced a project protecting 550 devices across 25 branches in 3 countries. After deployment and connection to the 24/7 SOC, the project recorded a 48% reduction in false positives compared to pre-deployment.

This is a result within a specific project scope, not a default improvement level for all systems. However, this case demonstrates that the quality of classification and operational organization can be just as important as the volume of alerts detected by the technology.

Containment: Turning verified alerts into containment actions

Containment is the activity of limiting the ability of an incident to cause further harm, such as preventing a device from communicating with other systems or restricting a malicious process from operating.

After verifying the risk, the SOC must decide whether to respond immediately, execute controlled actions, or escalate for approval.

Depending on the platform, permissions, and deployment configuration, actions may include:

  • Isolating the endpoint from most network connections.

  • Stopping or restricting the execution of malicious processes.

  • Quarantining detected files.

  • Blocking file hashes, IPs, or domains according to supported mechanisms.

  • Collecting additional evidence before restoring the device.

The Microsoft Defender for Endpoint documentation describes response capabilities such as device isolation, collecting investigation packages, and restricting application execution. However, availability depends on the license tier, operating system, configuration conditions, and granted permissions.

When is the SOC allowed to automatically isolate endpoints?

The answer must be based on the playbook agreed upon with the enterprise.

For standard devices and high-confidence dangerous behaviors, enterprises may allow automatic responses based on predefined conditions.

Conversely, isolating production servers, industrial control devices, or high-availability systems can cause severe disruptions. These cases typically require an approval process or alternative containment plans.

Before enabling automation, enterprises should determine:

  • What types of devices are allowed to be isolated automatically?

  • What are the required technical conditions and confidence levels?

  • Who has the authority to approve or cancel actions?

  • How to maintain necessary administrative channels?

  • What are the procedures for restoring connectivity and verifying safety?

Successful containment does not mean the incident is over. After containment, the response team still needs to investigate the root cause, determine the scope of impact, eradicate malicious components, and verify conditions for restoration.

Escalation and ownership: Who is responsible for handling the incident?

One of the major gaps when integrating EDR/XDR into the SOC is the lack of responsibility delegation among the monitoring unit, internal IT team, and the authorized decision-makers.

The SOC can detect incidents but may not have the authority to change production system configurations. Conversely, IT has the authority to operate but may not have enough data to determine which action is appropriate from a security perspective.

Therefore, the escalation process needs to answer three questions: who receives it, who executes it, and who makes the decision.

Delegating responsibilities by activity

In a reference model:

  • SOC Analyst is responsible for verifying signals, collecting evidence, assessing risks, and proposing actions.

  • Incident Owner oversees the entire handling process, coordinates parties, and ensures the incident is not abandoned upon escalation.

  • IT/Endpoint Team supports operations related to configuration, connectivity, device operation, and system restoration.

  • Business/System Owner assesses the impact on business operations and approves actions that may significantly affect service continuity, according to prescribed authority.

In enterprises with a dedicated incident response team, this group can take on deep investigation, threat eradication, and recovery coordination.

Establishing escalation conditions

An escalation process can be based on severity, asset type, signs of lateral movement, unresolved time, or the need for approval.

For example, upon detecting suspicious behavior on a financial server, the SOC needs a mechanism to directly contact the designated focal point, clearly stating the reasons for proposing containment and potential risks that may arise.

A minimum handover record should include the incident ID, affected devices, critical evidence, impact assessment, actions taken, assignee, and next steps.

For enterprises using outsourced services, the scope of responsibility should be detailed in the contract and SLA, rather than assuming the provider is permitted to handle all types of incidents by default.

Enterprises can learn more about the differences between SOC and MDR, or consider an outsourced SOC model when internal teams lack the resources to maintain continuous monitoring and coordination activities.

Which KPIs show that EDR/XDR has integrated effectively with the SOC?

Successfully connecting EDR/XDR technically does not prove that the SOC operational process is effective. Enterprises need to measure data quality, processing capability, and response outcomes.

Six priority KPI groups include:

1. Endpoint telemetry coverage: The percentage of endpoints within the monitoring scope that have active agents and send valid data. This should be cross-referenced with the actual asset inventory, not just the number of devices appearing on the console.

2. Alert-to-incident conversion rate: The percentage of alerts verified and converted into incidents requiring handling. This metric helps track classification quality, but neither a higher nor a lower rate should always be considered good.

3. Mean Time to Detect (MTTD): The average time from when a suspicious activity begins to when it is detected, within the scope of incidents with reliable time data.

4. Mean Time to Contain (MTTC): The average time from a predefined milestone until the containment measure takes effect. This is a particularly useful metric when evaluating the ability to move from detection to action.

5. Escalation SLA compliance: The percentage of incidents notified, acknowledged, or escalated within the committed timeframe for each priority level.

6. Incident closure quality: The percentage of records with sufficient evidence, confirmed actions, accountable persons, remediation outcomes, and incident closure conditions.

KPIs must have consistent definitions for start time, end time, and measurement scope. For example, MTTR can be understood as Mean Time to Respond, Remediate, or Recover depending on the organization; comparisons should not be made between two systems without unified definitions.

When assessing the effectiveness of an EDR XDR integrated SOC, enterprises should also inspect after-hours operational capabilities, processes for handling unverified alerts, and coordination mechanisms when a response action fails.

If the alert volume decreases but telemetry coverage declines or many devices lose connection, the result cannot be considered a genuine improvement.

When do enterprises need to re-evaluate the operational model?

Enterprises should review the integration process when there are frequent occurrences of unassigned alerts, many investigative operations having to be performed manually, a lack of response permissions, or an inability to prove processing times through reports.

An effective system is not necessarily the one with the most tools, but one with sufficient data, analytical capabilities, and accountability to respond appropriately to risks.

Has your enterprise deployed EDR/XDR but has yet to optimize the monitoring and response process?

Explore IPSIP Vietnam's 24/7 SOC service to evaluate a monitoring, analysis, and response coordination model tailored to your current system. Enterprises can also book a consultation with IPSIP to discuss integration scope, response permissions, and operational metrics to track.

ipsip-vietnam

EDR and XDR in the SOC unlock their value when endpoint data is connected to a controlled detection, verification, and incident handling process. The six components of telemetry, correlation, triage, containment, escalation, and KPIs form the foundation for enterprises to assess their operational readiness.

The crucial factor is not only how many alerts the SOC sees, but which alerts are verified, what actions are taken, who is accountable, and how the results are validated.

References:

follow ipsip vietnam.png
bottom of page