EDR versus NDR is primarily a difference in observation: endpoint detection and response examines activity on covered hosts, while network detection and response examines traffic visible to its sensors. For a cleared cyber analyst, the useful skill is connecting those observations while explaining what each source cannot establish.
A host process, a remote login, and a network connection can describe related parts of an investigation. They are not interchangeable records. Your interview example should explain how you connected them and where the evidence stopped.
Key takeaways
- Endpoint context: Process and host activity can explain what executed on an instrumented system.
- Network context: Observed connections can show relationships across assets without proving the initiating process or user.
- Correlation: Time, asset identity, session context, and source coverage determine whether a pivot is defensible.
- Career evidence: Build an investigation record that includes uncertainty and collection gaps, alongside the result.
EDR versus NDR: what evidence does each provide?
Endpoint telemetry can provide process creation, command execution, and host changes where collection is supported and enabled. Network telemetry can provide connections, protocol observations, and traffic patterns where sensors can see the traffic. Actual fields and response options depend on the deployed tools, permissions, and configuration.
NIST SP 800-94, published in February 2007, distinguishes host-based and network-based intrusion monitoring, alongside other approaches. It is background for understanding observation points, rather than a specification for today’s EDR or NDR products. NIST notes that its later draft revision never became final and was retired.
For current investigation planning, MITRE ATT&CK describes process creation data separately from network connection observations. That distinction gives you a practical question for every record: did this source observe a process, a connection, or a result inferred by an analytic? Preserve the difference in your case notes.
| Observation | Potential evidence | Limit to explain |
|---|---|---|
| Endpoint activity | Process ancestry, command execution, host changes | Collection depends on host coverage and available events |
| Network activity | Observed peers, connection timing, protocols, traffic volume | Sensor position and encryption affect what is visible |
| Authentication activity | Account, session, source, and authentication outcome | A successful account login does not establish the human operator |
| Central analysis | Queries across collected sources and correlated alerts | A central platform cannot analyze an event it never received |
Use product-specific study, such as cleared endpoint security skills, to learn an interface after you understand the underlying observation. On a resume, identify the task you performed in that interface. Product access alone is a weak description of your investigative contribution.
What can each sensor reveal during possible lateral movement?
Endpoint evidence may connect a login with subsequent process or file activity. Network evidence may show a source reaching another asset or an unusual pattern of internal connections. A lateral-movement assessment needs the surrounding activity and authorization context; remote administration also has legitimate uses.
MITRE’s remote-service login detection strategy, version 1.0 and updated in May 2026, correlates remote logons with later commands, file activity, or network connections. Its examples show why a connection alert is a starting point for investigation. It does not make every remote login malicious.
Build a fictional scenario around an approved administrator connecting from a lab workstation to a lab server. Keep the expected activity record. Then ask what you could establish if you had only the connection, only the destination login, or only the host’s process events. Compare the answers before combining the sources.
A network observation might justify checking the destination host. Host evidence might then justify reviewing a related account session. Neither step should erase uncertainty: a shared address, incomplete collection, or ambiguous session mapping may prevent a confident association. State that limitation in the result instead of filling the gap with a plausible story.
How do you pivot between hosts, flows, and identity?
Establish a common time basis, identify the assets and sessions involved, and preserve the records that support each connection. Treat every pivot as a testable association. Similar timestamps or matching addresses can support an investigation, but do not independently prove common causation or a particular human operator.
The Zeek 9.0.0 connection-log documentation explains network records containing participants, timing, duration, protocols, and traffic measures. These fields make useful lab material for a network-to-host pivot. A connection log should not be described as a full process execution record.
Start a worksheet with the network event and the question it raises. Find the relevant endpoint record, compare event times, and note how the host was identified at that time. Then inspect available authentication evidence. If you cannot associate the connection with a particular session, record the unresolved association explicitly.
Time is part of the evidence. Preserve original timestamps and record how you normalized them for comparison. If an analyst on the next shift cannot distinguish an event’s original time from its ingestion time or your converted display time, the timeline may appear more precise than the underlying records justify.
Practice packet-analysis skills for cleared network roles alongside host investigation. When a reviewer asks what a packet trace proves, point to the visible fields. Separate the data captured from the conclusion you reached after consulting another source.
Where do unmanaged assets and encrypted traffic leave gaps?
An endpoint without the relevant collection cannot supply the same host evidence as a covered device. A network sensor sees only traffic that reaches its observation point. Encryption can hide payload content while leaving useful connection observations. Coverage must be checked against the specific question being investigated.
NIST’s older intrusion-monitoring guidance explicitly discusses encrypted traffic and sensor placement. Retain the underlying distinction: observing that a connection occurred is different from reading its protected content. Avoid describing encryption as complete invisibility or assuming a network tool can reconstruct every action inside an endpoint.
For an unmanaged lab asset, mark which network paths you can observe and which host events you do not collect. Write a question the connection data can answer, then a question it cannot. The exercise helps you explain the value of network visibility without claiming that it replaces an endpoint agent.
Coverage also needs operational evidence. Before interpreting an empty search as a clean result, ask whether the source was active, the event type enabled, the relevant route observed, and the records available for the interval. Check ingestion health. A search that returns nothing can reflect a collection failure rather than absence of suspicious activity.
The joint event-logging guidance from August 2024 emphasizes useful events, correlation, and protected logging. It also warns that apparently suspicious observations can be legitimate. Those points support a disciplined investigation; they do not prescribe universal retention periods or guarantee detection coverage.
Which lab exercises demonstrate cross-tool fluency?
Choose an authorized activity you can observe through endpoint, network, and authentication records. Build a timeline, test a plausible alternative explanation, and document a collection gap. The deliverable should let another analyst repeat your reasoning without needing access to a commercial product console.
Follow an approved remote administration session
Use isolated lab systems and fictional accounts. Record the intended connection and approved action, then collect the available host and network evidence. Your write-up should identify the initiating asset, destination, observed session, and subsequent host activity, while stating any association you could not verify.
Make the alternative explanation explicit. If the same visible behavior could arise from a scheduled administrative task, identify the additional evidence needed to distinguish that task from an unauthorized session rather than claiming that the connection pattern alone identifies an intruder or that a successful login demonstrates who was using the account.
Remove an observation and reassess the conclusion
Keep the original evidence intact, but repeat the analysis with a source withheld from your working dataset. Label the exercise. Explain which conclusions remain supported and which become uncertain, then list the next collection request you would make. This tests reasoning about visibility rather than familiarity with a dashboard.
Write an analyst handoff
Include the question, source coverage, queries or filters, observed facts, unresolved issues, and recommended next step. Keep response authorization separate from technical ability. In a cleared role, a public interview artifact should contain synthetic data and generic system descriptions, with no customer records or sensitive environment details.
MITRE’s older data-source catalog was deprecated in October 2025. Use the current detection strategies and data components when planning exercises, and record the page version you consulted. An older checklist may still explain terminology, but should not be presented as a current catalog that continues to receive new sources.
How should you describe these skills in applications?
Describe the investigation, sources, reasoning, and result. Distinguish professional work from lab practice and distinguish reviewing an alert from configuring collection or executing response. Use the posting’s terminology when it accurately describes your experience, then support it with a concrete example.
For example, a lab-focused resume statement could describe correlating endpoint process events, network connection logs, and authentication records to reconstruct an approved remote administration session. Add that you documented visibility gaps and validated the timeline. Keep it clearly labeled as a lab exercise; do not imply production incident ownership.
When comparing EDR with a SIEM, remember that central analysis works on collected records. The May 2025 SIEM practitioner guidance describes collection, centralization, and analysis. Your SIEM investigation examples should identify the source telemetry behind a result, rather than attributing every observation to the platform displaying it.
Other labels, including managed or extended detection offerings, require questions about actual scope. Ask which sources the team collects, who investigates, and who approves response. The answers help you compare duties without assuming that an acronym specifies either a complete sensor inventory or your personal responsibilities.
Frequently Asked Questions
Can NDR replace endpoint investigation?
Network evidence can support investigation where host telemetry is unavailable, but it does not automatically supply process ancestry or local changes. Compare the question you need to answer with the available data. Explain what network observations establish and which host-level questions remain open.
Can EDR show every connection across a network?
Evaluate the actual endpoint coverage and collected events. Evidence from covered hosts does not establish complete observation of every asset or route. In an interview, describe how you would check coverage before interpreting an empty search as evidence that no relevant activity occurred.
Does encrypted traffic prevent useful network analysis?
Encryption affects payload access, but connection timing, participants, and other visible metadata may still support analysis. State what your sensor actually recorded. If the question requires protected content that was not available, explain the limit and identify another authorized source that could address it.
What should I prepare for a cross-sensor interview?
Prepare a sanitized lab timeline that joins endpoint, network, and authentication observations. Include a plausible legitimate explanation and a documented coverage gap. Revise the handoff until another analyst can follow each association, reproduce your filters, and identify the next evidence request without guessing.
