Monitored Recovery
Reconnecting systems to the network is not the end of the incident. Attackers routinely attempt to reaccess environments after eviction. The period between restoration and return-to-normal monitoring is when reinfections happen — and when they're hardest to detect because the IR team has stood down.
An IR team spends three weeks responding to a sophisticated intrusion. Eradication is declared complete. Systems are restored. The CISO declares the incident closed and the IR team reverts to normal operations. Monitoring intensity drops back to pre-incident levels.
Three weeks later, a laptop that returned from an employee's vacation — a laptop that had been offline during the entire incident window and therefore never re-imaged — comes back online. The employee connects to VPN. The laptop still has the beacon installed from six weeks before the incident was detected. The beacon calls home, gets a response, and the attacker is back in the environment. This time, they move faster because they know what they're looking for and where the valuable data is.
Monitored recovery is the operational phase that catches exactly this scenario. It is not optional.
What "Heightened Monitoring" Actually Means
"We'll monitor closely during recovery" is a common IR commitment that means very different things to different people. Operationally, heightened monitoring is a specific set of changes to alert thresholds, detection rules, and analyst assignments — not a vague intention to pay more attention.
Heightened Monitoring: Specific Changes Required
═══════════════════════════════════════════════════════════════════
ALERT THRESHOLD CHANGES
─────────────────────────────────────────────────────────────────
→ Lower severity threshold for alerts from previously compromised hosts
(treat Medium alerts as High for 30 days post-recovery)
→ Enable all suppressed detection rules that were suppressed for
noise reasons — the noise is acceptable during this window
→ Set specific alerts for any authentication from accounts that were
known-compromised during the incident (even after password reset)
→ Alert on ANY connection to previously identified C2 infrastructure
(not just the confirmed-active C2 — any known-bad IP or domain
from this actor)
DETECTION RULE ADDITIONS
─────────────────────────────────────────────────────────────────
→ New rules written from incident timeline (every event with no
corresponding alert is a new rule — see Chapter 15)
→ IOC-based rules for all indicators from this incident
(file hashes, process names, C2 domains, mutex names, pipe names)
→ Re-compromise indicator rules:
- Any process execution from paths used by attacker
- Any authentication from IPs associated with attacker
- Same attack chain (initial access → persistence mechanism pattern)
ANALYST COVERAGE
─────────────────────────────────────────────────────────────────
→ Named analyst owns monitored recovery for the incident
(not "the SOC watches it" — a specific person checks daily)
→ Daily review of authentication logs for formerly-compromised accounts
→ Daily review of DNS/proxy logs for connections to incident-related domains
→ Weekly EOD summary to IR lead during monitored recovery window
SCOPE
─────────────────────────────────────────────────────────────────
→ Every host in the incident scope is in the monitored recovery scope
→ Hosts that come online during the recovery window (from vacation,
travel, repair) are immediately evaluated: were they in scope?
If yes: re-image before reconnecting to the clean environment.
If no: standard onboarding but with incident IOC scan.
Attacker Re-Check Behavior
One of the most consistent patterns observed across IR engagements is attacker return behavior. After being evicted from an environment, a significant percentage of threat actors attempt to re-access the same environment within days to weeks. Understanding why this happens tells you when and how to watch for it.
| Actor type | Typical re-check timeline | Re-entry method | Implication for monitoring |
|---|---|---|---|
| Financially motivated (ransomware operators) | 2–6 weeks after apparent eviction | Same initial access vector if unpatched; credential reuse; cached credentials on un-remediated hosts | Extend monitored recovery to 6 weeks minimum; verify all initial access vectors are closed |
| Espionage / nation-state | Months to years — highly patient; may wait for monitoring to relax | Different initial access vector; previously established but undetected persistence; supply chain reentry | Permanent elevated monitoring for nation-state incidents; assume they have persistence you didn't find |
| Access broker (selling access) | As soon as new buyer is found — may be days | Credentials sold to buyer; same access path; or multiple buyers accessing simultaneously | Monitor for credential reuse by multiple different source IPs/geographies |
| Opportunistic / commodity malware | No specific re-check intention; reinfection via same vulnerable service | Same CVE scan and exploit; same phishing campaign | Verify the initial access vulnerability is patched; block the delivery mechanism |
An attacker who spent weeks in your environment has already paid the intelligence cost of mapping your network. They know where the data is, which accounts have the most access, which systems are poorly monitored. From their perspective, the environment is now a known quantity with high expected yield. Reentry is worth the effort even after being evicted because the mapping intelligence persists. This is why re-compromises after eviction are often faster and more damaging than the original compromise — the second time, the attacker knows exactly what to do and where to go.
Specific Re-Compromise Indicators
Monitored recovery works because the attacker, re-entering an environment they've been evicted from, tends to use familiar patterns. They know the environment; they don't necessarily know what changed in the IR. The indicators that catch re-compromise are the same TTP-level patterns that caught the original intrusion — plus new IOC-based detections from the incident's specific artifacts.
Re-Compromise Detection Framework
═══════════════════════════════════════════════════════════════════
CATEGORY 1: Known-Bad Infrastructure
─────────────────────────────────────────────────────────────────
Alert on ANY connection (DNS query, proxy request, direct connection)
to infrastructure associated with this incident:
→ All C2 domains (even if they appear to have rotated)
→ All C2 IP addresses
→ All certificate thumbprints from incident TLS C2
→ JA3 hashes from incident C2 traffic (if captured)
Rationale: attackers sometimes reuse infrastructure, especially
for initial callback after reentry. Even if they rotated to new
C2, old C2 traffic is a confirmation of compromise.
CATEGORY 2: Known-Compromised Account Activity
─────────────────────────────────────────────────────────────────
Alert on successful authentication from any account that was
confirmed compromised during the incident, even after:
→ Password was reset
→ MFA was re-enrolled
→ Account was disabled and re-enabled
Rationale: the attacker may have established persistence tied
to the account (OAuth tokens, session cookies, saved refresh
tokens, Kerberos TGT cached on remote host) that survives
credential rotation.
CATEGORY 3: Same TTP Pattern
─────────────────────────────────────────────────────────────────
The attacker's techniques are harder to change than their tools.
Alert on the same attack chain pattern:
→ If initial access was phishing → VBA macro → PowerShell download:
alert on any Word document launching PowerShell within 30 seconds
→ If lateral movement was PsExec to DCs:
alert on SMB+ADMIN$ write followed by service creation on DCs
→ If persistence was scheduled task with base64-encoded command:
alert on any new scheduled task with encoded command line
CATEGORY 4: New Hosts Coming Online
─────────────────────────────────────────────────────────────────
Any host that was not present during eradication validation
that joins the network during the recovery window:
→ Run automated IOC scan against incident indicators
→ Check against scope definition: was this host in scope?
→ If yes: re-image before allowing network connectivity
→ If no: enhanced monitoring for 30 days
When Monitored Recovery Ends
Monitored recovery should not end on a calendar date — it should end when specific criteria are met. A 30-day timer without criteria leads to either premature stand-down (if nothing has happened) or indefinite continuation (if something minor keeps resetting the clock). Criteria-based exit is cleaner.
| Criterion | Met when | Who verifies |
|---|---|---|
| All in-scope hosts restored and validated | Every host in the incident scope has been either re-imaged/validated or explicitly cleared from scope with documented justification | IR lead + system owners |
| All offline hosts assessed on return | No hosts remain that were offline during the incident window and have not been assessed on return to the network | IT asset management + EDR coverage report |
| Authentication baseline normalized | Failed logon rates, explicit credential use rates, and remote authentication patterns have returned to pre-incident baseline for 14+ consecutive days | SIEM analyst |
| No C2 traffic to incident-related infrastructure | Zero connections to incident-identified C2 infrastructure for 30+ days | Firewall/proxy analyst |
| Detection uplift implemented | New detection rules from Chapter 15 lessons learned are deployed and tested | Detection engineer |
| Initial access vector confirmed closed | Vulnerability scan confirms the initial access CVE is patched on all systems; external pen test or scan confirms no reachable attack surface matches the initial access pattern | Vulnerability management |
The monitored recovery phase is not just defensive — it is the bridge between reacting to an incident and improving the security program. The detection rules you write during monitored recovery should become permanent detections. The authentication monitoring you implement for formerly-compromised accounts should become a template for monitoring privileged accounts generally. The log reviews you do during monitored recovery expose logging gaps that should be closed permanently. Every improvement made during monitored recovery because of the incident is a security investment that pays dividends on all future incidents.
Detection Uplift During Monitored Recovery
The incident produced a timeline of attacker activity. Many items in that timeline had no corresponding alert — they were discovered through forensic analysis, not detection. Each one of those items is a detection gap. Closing detection gaps during monitored recovery turns a reactive incident into a proactive security improvement.
Timeline-to-Detection Conversion
═══════════════════════════════════════════════════════════════════
For each timeline entry that had no alert:
Timeline entry: "Attacker used comsvcs.dll MiniDump to dump LSASS
at T+8 hours. Discovered forensically post-incident."
Detection gap: No alert fired for rundll32.exe comsvcs.dll MiniDump
targeting lsass.exe
New detection rule:
─────────────────────────────────────────────────────────────────
title: LSASS Memory Dump via comsvcs.dll MiniDump
logsource:
product: windows
category: process_creation
detection:
selection:
Image|endswith: '\rundll32.exe'
CommandLine|contains|all:
- 'comsvcs.dll'
- 'MiniDump'
condition: selection
level: high
Apply this process to every item in the timeline that was
discovered forensically rather than through a live detection.
The result is a set of detection rules that are:
→ Validated against a real attack (not theoretical)
→ Tuned for your environment (you know it fired in your env)
→ Immediately deployable (no FP-testing phase needed — you
know the context in which this detection applies)
Q & A
Q: How long should monitored recovery last for a ransomware incident?
For a financially motivated ransomware group (RaaS operator), the observed re-check window is typically 2–6 weeks. A monitored recovery period of 30 days is reasonable, extended to 60 days if the attacker's dwell time before detection was long (suggesting they are patient and methodical). The 30-day minimum is not arbitrary — it covers one full cycle of the attacker's typical re-check behavior. It also allows time for offline devices to return from travel or repair and be assessed. The criteria-based exit (authentication baseline normalized, no C2 traffic, all returned devices assessed) should be met regardless of timer — if the 30-day timer expires but a device from a business trip just returned and hasn't been assessed yet, the monitored recovery continues until it is.
Q: A user's personal laptop was used for some work during the incident. It's not corporate-owned and isn't subject to our endpoint management. What do we do?
This is a scope decision with business and legal dimensions. The personal device is potentially compromised if it was used to access corporate resources during the incident window (VPN, webmail, corporate SaaS). The options: (1) Ask the employee to submit the device for forensic assessment (voluntary — you cannot compel access to a personal device without legal process). (2) Revoke all access tokens and session cookies for all corporate services and require full re-authentication from a managed device. (3) Treat the personal device as compromised and block it from accessing corporate resources until it has been assessed. In practice, option 2 is almost always the right first step — it limits the damage regardless of whether the device is actually compromised. Option 3 is the right policy going forward: corporate resources should not be accessible from unmanaged devices, and this incident is the argument for enforcing that policy.
Q: The threat intel team says they see the same threat actor in another organization's environment right now. Does that affect our monitored recovery?
Yes, positively. Active threat intelligence about the actor's current TTPs can improve your detection rules during monitored recovery. If they are currently using a new C2 infrastructure or a new toolset in their current operation, and you have IOCs for it, add those to your monitoring. Also consider the implication: if the actor is actively engaged against another organization right now, they may be less focused on re-checking your environment in the immediate term. However, this does not justify reducing monitoring — threat actors can run multiple simultaneous operations, and active engagement elsewhere does not mean they won't check on your environment.
Q: We found during monitored recovery that an alert would have fired three weeks earlier if we'd had Script Block Logging enabled. Do we tell the business?
Yes, in the lessons learned report — framed as "a detection gap we identified and have now closed" rather than "we could have caught this three weeks earlier." The business impact statement should focus on what improved: "We have implemented Script Block Logging and [N] other detection improvements that would have reduced our detection time in this incident. We estimate these changes reduce detection time for similar techniques by approximately [X] days." The framing matters: the goal is to communicate security improvement, not to assign blame or create legal liability. The specific counterfactual ("the attacker was in our environment for three additional weeks that we could have prevented") is accurate but the way you communicate it to leadership shapes whether it drives security investment or just creates blame.
Q: An executive is asking why we're still monitoring intensely three weeks after "the incident was closed." How do you explain the monitored recovery phase?
Brief explanation that works for non-technical executives: "When an attacker is evicted from an environment, they typically try to get back in within the next few weeks using credentials they cached or access paths they didn't use during the original attack. We've seen this in [X]% of incidents where attackers are evicted. We run heightened monitoring for 30 days specifically to catch this — it's the period when re-compromise is most likely and when we'd catch it fastest. The cost is a few analyst hours per week. The cost of missing a re-compromise that leads to a second incident is another full incident response. We'll stand down monitoring when [specific criteria] are met, which we expect to be [date range]." Having a concrete exit criteria helps — executives are more comfortable with "we'll stop when X, Y, and Z are confirmed" than with an open-ended enhanced monitoring commitment.
Q: We think we found evidence of re-compromise during monitored recovery. What do we do?
Treat it as a new incident with all the same rigor as the original — not as a continuation of the closed incident, and not as a minor event that can be handled informally. Immediately: (1) Preserve the evidence that suggests re-compromise — memory image if possible, EDR telemetry, network captures. (2) Declare a new incident P1 and re-engage the IR team. (3) Conduct rapid triage: is this the same actor using the same access, a different actor using a different path, or a false positive? (4) Brief the CISO within the hour. The key mistake is treating re-compromise as an extension of the old incident that can be resolved quickly because "we know this actor." You don't know what they did between eviction and reentry. Treat it as a full investigation.