T Techclick ← All lessons
Microsoft Defender for Endpoint · Evidence desk · Interactive lesson

Prove MDE is working — first tool + proof field

01:40. Slack: “Is Defender even working?” Then: “Why didn’t it block?” The CIO is already in the channel. A tray-icon photo is not proof. This desk is five official surfaces on security.microsoft.comDevice page sensor health / last seen, Timeline, Alert / Incident, Action center, ASR / next-generation protection — each mapped to one ticket, one first click, and one field you paste before you isolate, exclude, or flip a rule.

~20 min read · L2 primary · Quiz at end · Blog 1 · Factory

⚡ Quick Answer

How you prove Microsoft Defender for Endpoint is working: Device page sensor health / last seen, Timeline, Alert / Incident, Action center, ASR / next-gen protection. Five tickets with first tool and one proof field.

After this page you can

Quick answer (say this out loud)

Device page answers “did this sensor check in?” Timeline answers “what parent launched what child, under which MITRE technique?” Alert / Incident answers “what severity, in what status, with what classification?” Action center answers “did Isolate device / collect / live response land?” ASR / next-generation protection answers “was this host even allowed to block?” High risk level is not Isolated. Audit is not a miss. A green Windows Security icon is not Last seen.

1. Why “is MDE working?” is five questions

Operators collapse five failures into one sentence. The sensor is Inactive. The device page Last seen is days old. The Office-child ASR rule is in Audit. The alert is still New while the laptop sits on the LAN. Isolate device is Pending because the host is dark. Those are five first clicks.

This page is the night-shift desk for proof. The factory taught ASR Audit versus Block, risk versus isolate, and why a living-off-the-land alert is not a patch ticket. Here you learn the five Microsoft Defender portal surfaces you actually open, in order, when someone asks you to prove MDE is working — or to explain why it did not block.

Hero · five tiles, one ticket
Night-shift operations desk with five glowing MDE proof tiles on a wall monitor
Notice: five tiles, not one “Defender dashboard.” You pick the tile that matches the question, then you quote one field.
Interview line

If they say “prove Defender is working,” do not say “I opened security.microsoft.com.” Say: “I prove the sensor with Device page Sensor health state and Last seen, the chain with Timeline process tree + MITRE, the case with Alert Severity / Status, the click with Action center, and the block decision with ASR mode or Defender Antivirus Active versus Passive.”

2. Mental model — five proof tools

Memorise five named objects before you click. Each tool is allowed to prove one thing. Over-claiming a field is how you isolate a leftover renamed device or flip ASR tenant-wide at 02:00.

1 · Device page

Assets → Devices (security.microsoft.com/machines), then the device. Proves the sensor: Sensor health state, Last seen, Onboarding status, Risk level. Does not prove a MITRE technique or an isolate click.

2 · Timeline

Device page → Timeline. Proves parent → child, command line, MITRE ATT&CK technique, event time. A filename in Slack is not the tree. Empty Timeline on an Inactive sensor is expected.

3 · Alert / Incident

Incidents & alerts → Alerts (security.microsoft.com/alerts), or the device Incidents and alerts tab. Proves Severity, Status (New / In progress / Resolved), Classification. High is not Isolated.

4 · Action center

Action center (security.microsoft.com/action-center), or the Action center button on the device. Proves Isolate device / Collect investigation package / Live response / Run antivirus scan — Success, Pending, Failed, or Skipped. The button is not the proof.

5 · ASR / next-gen protection

Device Effective settings, or Endpoint security policies at security.microsoft.com/policy-inventory, or Reports → Attack surface reduction rules. Proves Audit versus Block, and Defender Antivirus Active versus Passive. A Detect/Audit mode is configuration, not a miss.

Hard words, once

Sensor health state = Active / Inactive / Misconfigured. Last seen = last connection on the device page (may lag Timeline). Isolate device = response action (needs Active remediation actions). Contain device = different action for unmanaged hosts. ASR = attack surface reduction. NGP = next-generation protection (Defender Antivirus).

Flow 1 · five tools, one question each
Write hostname + UTC first · then pick the tool Is MDE working? five questions, not one Device page Sensor talking? health · last seen Assets → Devices onboard · risk not a MITRE ID Timeline Who launched it? parent → child MITRE technique Device → Timeline not live isolate Alert / Incident This case? Severity · Status Classification Incidents & alerts High ≠ Isolated Action center Did the click land? Isolate · collect Success / Pending action-center needs a live sensor ASR / NGP Allowed to block? Audit vs Block AV Active / Passive Effective settings mode is not a miss Empty Timeline / empty Alerts is data. It usually means the sensor never landed. Do not invent a miss from an empty queue. Start at Assets → Devices · Sensor health state + Last seen.

Read left → right. Each box is allowed one claim. If you cannot name the field, you are not proving — you are guessing.

Say this out loud

I prove the sensor, then the timeline, then the alert, then the Action center row, then the assigned ASR / antivirus mode. I do not isolate, exclude, or flip Audit → Block until I can quote the field that made me do it.

3. Decision flow — ticket → first tool

Flowchart first. Do not open Endpoint security policies or click Isolate device until a diamond says so.

Path · pick the branch before the menu
Abstract diamond splitting into five MDE proof paths
Notice: the diamond is the ticket. The path is the tool. The field comes last. Do not reverse that order.
Flow 2 · first-tool diamond
Symptom first · tool second · field third What must we prove? Sensor live? or already inside? “Is MDE up?” Device page health · last seen “Why no block?” ASR / NGP Audit · Passive Live process tree Timeline parent · MITRE Need isolate proof Action center Success / Pending High / still New Alert / Incident Severity · Status Inactive / Last seen > 7 days / Misconfigured → stop. There is no live isolate and no “MDE miss.” Fix the sensor (proxy, Sense, renamed leftover). Then re-open Timeline. Diamond = decision. Do not isolate from the bottom box. Do not flip ASR from an empty Timeline. Official Last seen caveat: the device-page clock can lag newer Timeline events. Quote both if they disagree.

Read the diamond first. “Why didn’t it block?” never starts in Action center. A dark Last seen never starts in Endpoint security policies. A blocked demo can be policy success.

4. How to choose — first tool + proof field

Print this next to the Microsoft Defender portal. If you cannot recite the proof field, you are not ready to isolate or change a rule.

If the ticket says…First tool (official path)Proof fieldDo not open first
Laptop / hotel / “is Defender even working?” Assets → Devices → device Overview (security.microsoft.com/machines) Sensor health state (Active / Inactive / Misconfigured) + Last seen + Onboarding status A new ASR Block, or Isolate device
“Why didn’t it block?” after Word spawned PowerShell Device Effective settings, or Reports → Attack surface reduction rules, or security.microsoft.com/policy-inventory ASR rule mode Audit (not Block), or Defender Antivirus Passive on the Device health status card Tenant-wide ASR flip
Need the process chain / who launched whom Device page → Timeline (from inventory, an alert, or an incident) Event time + parent → child + MITRE technique (and Additional information) The filename argument in Slack
High alert, still New, “just isolate it” Incidents & alerts → Alerts, then device Overview isolation state Severity + Status + Classification. High risk is not Isolated. Waiting for risk to self-contain
Did Isolate / collect / live response land? Action center (security.microsoft.com/action-center) · Pending / History Action type + status (Success / Pending / Failed / Skipped) + submitting user + time Click Isolate twice. Isolate a host Last seen days ago and call it done.
Last seen caveat (official)

Microsoft documents a product constraint: the device profile does not consider all cyber evidence when it computes Last seen. The Device page clock can show an older time even when newer alerts or Timeline rows exist. If Last seen looks stale but Timeline is fresh, quote both. Do not declare the sensor dead from Last seen alone when Timeline is still writing events.

5. Runbook Side A → B → C

Side A proves the sensor is on the wire. Side B proves what MDE saw and whether policy was allowed to block. Side C proves the response click landed. On a messy Sev-2, do them in this order until a field lights up.

Side A — Device page (sensor health / last seen)

  1. Open Assets → Devices, not the alert editor

    Path: Assets → Devices, or go directly to security.microsoft.com/machines. Search the hostname. If two rows appear after a reinstall or rename, Microsoft documents that a new device entity is created and the previous one stays Inactive — work the live name. Source: Explore devices in the device inventory; Fix unhealthy sensors.

  2. Read the three sensor columns that close “is MDE working?”

    Sensor health stateActive (reporting), Inactive (no report for more than seven days in the past month), or Misconfigured (impaired communications / no sensor data). Last seen — last connection on the device page. Onboarding status — onboarded versus discovered / can be onboarded. Source: Check the device health; Device inventory columns.

  3. If Inactive or Last seen is days, stop. This is a sensor ticket

    Do not start live response. Do not isolate and call it contained (a dark host retries isolate for up to three days, then gives up). Do not call it a miss. Check Sense / WinHTTP / proxy to Defender URLs, a rename leftover, or offboarding. Then come back.

  4. If Misconfigured, quote the subtype before you hunt a “miss”

    Impaired communications = limited talk to the service. No sensor data = the device talks but only reports partial data. Official next checks: Internet + WinHTTP, service URLs, diagnostic data service, Defender Antivirus ELAM not disabled by policy. Empty Timeline is expected until that is fixed.

security.microsoft.com/machines · Assets → Devices
Training mock · not live

Assets / Devices / All devices

Device inventory

DEVICE-LAB-17
Any
DeviceOnboardingSensor healthLast seenRiskIsolation
DEVICE-LAB-17OnboardedActive22s agoHighNot isolated
DEVICE-LAB-17-OLDOnboardedInactive12 days agoNo knownNot isolated

Source: Microsoft Learn — Explore devices in the device inventory (security.microsoft.com/machines); Check sensor health state (Active / Inactive / Misconfigured); Fix unhealthy sensors (rename leftover stays Inactive). Two names, one desk — work the 22s row. Lab identities only. Training mock · not live.

Side B — Alert / Incident, Timeline, ASR / next-gen protection

  1. Open the alert, not Slack’s filename

    Path: Incidents & alerts → Alerts (security.microsoft.com/alerts), or the device Incidents and alerts tab. Official columns: short description, Severity (high / medium / low / informational), Status (New / In progress / Resolved), Classification (not set / false alert / true alert), investigation state, category, assignee, last activity. Source: Investigate devices; Investigate alerts in Microsoft Defender XDR.

  2. Open Timeline before you type live response

    Device page → Timeline. Select the event → side panel with process tree. Quote parent, child, command line, and the MITRE technique (bold + blue icon). Additional information can say Remediation successful / unsuccessful, Suspicious script detected, or the alert category. Timeline is historical telemetry. Live response is live. Do not skip the tree to collect a file you already have in the graph.

  3. If the ticket is “why didn’t it block?”, open Effective settings — then the ASR report

    Device page → Configuration management → Effective settings. Official: setting name, policy type, effective value, source (Defender / Intune / Group Policy / registry), last report time. Complex settings such as ASR rules break down each rule, its source, and exclusions. Then confirm detections: Reports → Endpoints → Attack surface reduction rules and the Blocked/Audited? column.

  4. Also read Defender Antivirus mode. ASR needs Active

    Device health status card (highest-priority message includes Defender Antivirus not active). ASR rules require Microsoft Defender Antivirus in Active mode with real-time protection on — not Passive, not Limited periodic scanning, not Off. If AV is Passive, Audit/Block on an ASR policy will not do what the ticket expects. That is configuration, not a product miss. Source: ASR rules overview — Requirements.

security.microsoft.com · Assets → Devices → DEVICE-LAB-17 → Timeline
Training mock · not live

Assets / Devices / DEVICE-LAB-17 / Timeline

Device timeline

Last 24 hours
ALR-1042 · Suspicious PowerShell
Time (UTC)Action typeProcess / MITREAdditional information
10:41:02FileOpenedwinword.exe · invoice.docm
10:41:08ProcessCreatedpowershell.exe · T1059.001Suspicious script detected
10:41:12ConnectionSuccessrundll32.exe → 203.0.113.88Alert category · Command and Control

Source: Microsoft Learn — Investigate devices (Timeline tab; process tree; MITRE techniques; Additional information). Lab command lines and 203.0.113.88 only. Training mock · not live.

Alert + Timeline + policy — fields you write in the ticket
Path:            Incidents & alerts → Alerts   (or device Incidents and alerts)
Quote:           Severity + Status + Classification
Then:            Device → Timeline  (parent → child → MITRE + Additional information)
If “no block”:   Device → Effective settings   and/or
                 Reports → Endpoints → Attack surface reduction rules
Quote:           ASR rule mode Audit vs Block   ·  Blocked/Audited?
Also quote:      Device health status · Defender Antivirus Active / Passive
If empty queue:  Assets → Devices · Sensor health state / Last seen first
security.microsoft.com · DEVICE-LAB-17 → Effective settings · ASR
Training mock · not live

DEVICE-LAB-17 / Configuration management / Effective settings

Effective settings

Active
Intune · NGP-Finance-Lab
Audit
Block
ASR rule d4f940ab-401b-4efc-aadc-ad5f3c50688a: Audit (code 2)
ASR rule 9e6c4e1f-7d60-472f-ba1a-a39ef669e4b2: Block (code 1)
Reports → ASR rules → Blocked/Audited?: Audited for the Office-child event
Product did what this host is configured to do. Promote under change control.

Source: Microsoft Learn — Investigate devices (Effective settings); ASR rules overview (modes Off 0 / Block 1 / Audit 2 / Not configured 5 / Warn 6; Office-child GUID); ASR rules report (Blocked/Audited?); Endpoint security policies at security.microsoft.com/policy-inventory. Lab policy names only. Training mock · not live.

Side C — Action center (isolate / collect / live response)

  1. Isolate a live egress host from the device page — prove it in Action center

    Response actions sit along the top of the device page: Isolate device, Restrict app execution, Run antivirus scan, Collect investigation package, Initiate live response session, Action center. Official isolate: type a comment → Confirm. Isolation disconnects the device from the network while it stays connected to the Defender service. You need the Active remediation actions role and device-group access. Isolation is automatically lifted after seven days. Microsoft’s word on the device is Isolate device (Contain device is a different action for unmanaged hosts). Source: Take response actions on a device.

  2. Do not trust the button. Open Action center

    Path: device → Action center, or security.microsoft.com/action-center. Tabs: Pending (approve / reject) and History. Quote action type, submission time, submitting user, and whether it succeeded or failed. If the host was offline, Microsoft retries isolate for up to three days — Pending is not Isolated.

  3. Live response and collect only after Last seen is seconds

    Initiate live response session needs a talking sensor. Collect investigation package can fail on low battery or a metered link — the zip is downloaded from Action center when the package is available. Both are change-control in most shops. A dark Inactive device does not collect tonight.

security.microsoft.com/action-center · History
Training mock · not live

Action center / History / Isolate device

Action center

Isolate device
Success
DEVICE-LAB-17
soc.oncall@lab.example
10:52Z Isolate device · DEVICE-LAB-17 · Success
comment: ALR-1042 egress 203.0.113.88 — IR-LAB-17
Release from isolation is the undo. Isolation auto-lifts after 7 days.

Source: Microsoft Learn — Take response actions on a device; Overview of the Action center; View and manage actions (security.microsoft.com/action-center). Isolation success is evidence. Do not isolate twice. Lab identities only. Training mock · not live.

Green success on each side

6. Five tickets as full stories

These five land every quarter. Memorise first tool + proof field. Times and identities below are lab-only.

Journey · one amber hop is the ticket
Laptop to cloud path with four healthy cyan nodes and one cracked amber hop
Notice: an Active sensor can still leave Word→PowerShell running if ASR is Audit. That is a policy ticket, not a “Defender is down” ticket.
TicketSymptomFirst toolProof field
MDE-ED-01WFH laptop: “Defender is down, icon looks installed”Assets → DevicesSensor health state + Last seen — or the 12-day Inactive leftover
MDE-ED-02Word spawned PowerShell; process still running; “why no block?”Effective settings / ASR reportOffice-child rule = Audit · Blocked/Audited? = Audited
MDE-ED-03Need the chain / who launched the childDevice → TimelineParent → child + MITRE T1059.001 + event time
MDE-ED-04High / New; user still on the laptopIncidents & alerts → AlertsSeverity + Status = New · risk High is not Isolated
MDE-ED-05“I clicked Isolate — is it isolated?”Action centerIsolate device · Success / Pending / Failed + submitting user

MDE-ED-01 — Prove the sensor (Device page)

01:42 · P2. Priya on a hotel network. Phone photo of a Windows Security shield. L1 already drafted “Defender missed the malware.” Timeline for her hostname is empty.

First tool: Assets → Devices at security.microsoft.com/machines. Search DEVICE-LAB-17.

If Sensor health state is Inactive / Misconfigured, or Last seen is days: quote that pair. Empty Timeline is expected. Next check is sensor — WinHTTP / proxy to service URLs, Sense, rename leftover, offboarding — not a new ASR Block.

If Active and Last seen is seconds: the sensor is talking. Now you are allowed to open Timeline and Alerts for that device and UTC window. A tray icon is not Last seen.

Trap

Two rows after a reinstall. Microsoft keeps the old entity Inactive. Isolating DEVICE-LAB-17-OLD does not touch the laptop on the desk. Sort Last seen. Work the Active row.

MDE-ED-02 — Prove why it did not block (ASR / next-gen protection)

02:05 · P2. winword.exe started powershell.exe. The process is still running. Someone typed “Defender failed” and wants every ASR rule set to Block for Finance so this never happens again tonight.

First tool: device Effective settings, then Reports → Endpoints → Attack surface reduction rules. Confirm Defender Antivirus mode on the Device health status card.

Proof field: rule Block all Office applications from creating child processes (GUID d4f940ab-401b-4efc-aadc-ad5f3c50688a) = Audit (code 2). ASR report Blocked/Audited? = Audited. Antivirus = Active — so the rule could have blocked if it were in Block. That is the ticket. MDE did what this host is configured to do. Isolate if Timeline shows live egress. Flip Audit → Block under change control on a ring. Do not call it a miss and do not save a tenant-wide policy at 02:10.

Close

I would not call this a miss. I would quote Audit on that one rule, isolate the live host if egress is on the Timeline, and open a change to move that ring to Block. If the health card said Defender Antivirus not active, I would quote Passive first — ASR requires Active.

MDE-ED-03 — Prove the chain (Timeline)

02:20 · P1. ALR-1042. User still in Outlook. L1 wants to argue whether the file is named invoice.docm or Invoice (1).docm.

First tool: device page → Timeline (from the alert, the incident, or Assets → Devices). Select the 10:41Z ProcessCreated row → process tree.

Proof field: 10:41:02 winword.exe opened the document → 10:41:08 powershell.exe (MITRE T1059.001, Additional information: Suspicious script detected) → 10:41:12 rundll32.exe to 203.0.113.88. Last seen 22s, not isolated. Isolate first. Filename later. Hunt siblings with advanced hunting after the tree names the account.

Close

Quote event time + parent → child + MITRE. Isolate the live device. Do not spend the bridge on the attachment name. Source: Investigate devices — Timeline, process tree, MITRE techniques.

MDE-ED-04 — Prove the case (Alert / Incident)

02:40 · P1. Queue shows Suspicious PowerShell, Severity High, Status New. Chat says “risk is High so it must already be isolated.”

First tool: Incidents & alerts → Alerts, then the device Overview cards.

Proof field: Severity = High, Status = New, Classification = not set. Device Risk level = High. Isolation = Not isolated. Microsoft’s risk field is for prioritisation (and optional Intune compliance). Isolate device is a separate response action. Assign the alert. Do not leave New while the laptop sits on the LAN.

Trap

Exposure level is posture (pending recommendations). Risk level is alert-driven. Neither is Isolate device. Waiting for auto-isolate because risk is High is how egress continues.

MDE-ED-05 — Prove the click (Action center)

03:00 · P2. Someone already clicked Isolate device. The channel asks “are we contained?” A second person is about to click it again.

First tool: Action center → History, filtered to that device. Also re-read the device page isolation state.

Proof field: Isolate device · Success · submitting user · 10:52Z. If the row is Pending and Last seen is hours, you have a sensor ticket — Microsoft will retry isolate for up to three days, then stop. If Failed, quote the failure and fix reachability; do not click Isolate a third time. Release from isolation is the undo (comment + Confirm).

Close

I would paste the Action center row, not a second isolate. If Success, hunt siblings. If Pending on a dark host, treat it as a sensor ticket. Isolation auto-lifts after seven days — put a calendar note on the IR ticket.

7. Traps + close-the-ticket proof

Proof · named field, then Closed
Operations desk with abstract green health checks and one highlighted process-tree node
Notice: the close is a named column on a timestamp, not a screenshot of the user’s Windows Security icon.
You seeWeak closeStrong close
Empty Timeline / empty Alerts“Defender missed it”Assets → Devices · Sensor health state + Last seen first
Tray icon on a phone photo“MDE is working”You only proved a bitmap. Quote Last seen on that device
Office-child event, process still running“Defender failed”Effective settings: that ASR rule = Audit · Blocked/Audited? = Audited
Risk level HighWait for auto-isolateHigh is not Isolated. Timeline, then Isolate device, then Action center
Two device rows after a renameIsolate both / live-response the old oneInactive leftover. Work the Active Last seen
Last seen 8 days / InactiveStart live response / isolate and closeSensor ticket. Isolate retries then stops. Collect will not land
Device page Last seen older than Timeline“Sensor is dead”Official Last seen caveat. Quote both clocks
Defender Antivirus not activeFlip every ASR rule to BlockASR requires AV Active. Quote the health-card mode first
Isolate button clickedClick it againAction center History · Success / Pending / Failed
CVE list pasted into the IR chatConvert the ticket to patchingStay on Timeline + ASR. Vuln is a follow-up ticket (factory)
Proof checklist before you leave the bridge
Interview close

I name the question, then the first tool, then one official field. Device page proves the sensor. Timeline proves the chain. Alert proves the case. Action center proves the click. ASR / next-generation protection proves whether MDE was allowed to block. I isolate a live host. I do not start with “Defender missed it.” Factory model: ASR audit vs block.

Knowledge check

Six night-shift judgments. Each maps to a first tool or a proof field. Check answers, then Reset if you picked the wrong surface.

Q1

WFH user: “Is Defender even working?” Timeline for her hostname is empty. You have not opened a policy yet. First proof?

Correct: b. Empty Timeline is data. Official first surface is the device inventory / device page. Re-read Side A and MDE-ED-01.
Q2

Word spawned PowerShell an hour ago. The process is still running. Which proof field closes “why didn’t it block?”

Correct: a. Audit is configuration, not a miss. Action center and fleet counts answer different tickets. Re-read Side B steps 3–4 and MDE-ED-02.
Q3

Sensor health state is Inactive. Last seen is eight days. IR wants live response get of a script. What do you do first?

Correct: c. Official live response and collect need a talking sensor. Isolate on a dark host retries then stops. Re-read Flow 2 bottom box and MDE-ED-05.
Q4

ALR-1042 is High / New. Timeline shows rundll32 to 203.0.113.88. Last seen is 22s. Isolation is Not isolated. Next?

Correct: d. Official action is Isolate device. Risk is not isolation. Filename is not the first field. Re-read Side C and MDE-ED-03 / MDE-ED-04.
Q5

You already have the alert. You need to know who launched the child process. First surface?

Correct: b. Timeline is historical EDR telemetry. Live response is live. Do not lift isolation to “make the tree work.” Re-read Side B step 2 and MDE-ED-03.
Q6

Device risk is High. Isolation still says Not isolated. What is that pair allowed to mean?

Correct: a. Official device page: risk level is not Isolate device. Re-read MDE-ED-04, the How to choose table, and the factory. Vuln is a follow-up ticket.

Sources

Related: Blog 1 · Session factory — ASR audit vs block · Defender practice dashboard · Dummy lab · MDE interview Q&A