Co-managed security needs a tested decision boundary
Co-managed security means the customer and provider share operational responsibilities. It can combine local business knowledge with specialist monitoring or investigation, but the arrangement succeeds only when the handoffs work. An alert may be technically correct and still lead nowhere if each team assumes that the other owns the next decision. The most useful acceptance exercise tests that boundary directly.
This guide focuses on authority and escalation drills for a co-managed SOC. It explains how to define decisions, build safe scenarios, handle unavailable approvers and preserve a useful incident record. The examples are hypothetical. They do not involve unauthorised testing or demonstrate a real customer incident. Any operational action must remain within the agreed exercise scope.
For the overall service context, see our managed security services homepage. A co-managed model should make responsibilities more precise, not leave the customer with an ambiguous share of every task. The desired outcome is that both teams can explain who observes, investigates, approves, acts and verifies for a particular incident scenario.
Start with decisions instead of job titles
A responsibility chart that lists a CISO, IT manager and SOC analyst is not enough. Identify the decisions those roles must make. Examples include confirming whether activity is expected, authorising collection, approving account restriction, accepting business disruption and declaring recovery complete. Different decisions may belong to different people even within the same department.
For each decision, record the minimum information needed and the time sensitivity. A business owner confirming a maintenance window needs different evidence from an executive approving a major service interruption. The provider should frame its request accordingly. Sending every recipient a large technical alert and expecting them to infer their responsibility creates delay and confusion.
Separate advice from authority. The SOC may recommend an action without being authorised to perform it. The internal team may be authorised but lack the technical access to execute it. Identify the operator as well as the approver. This prevents the false assumption that an approved recommendation has already changed the environment.
Give every decision a fallback. The fallback may be an alternate approver, a narrower pre-authorised action or an explicit instruction to preserve evidence while waiting. It should be an organisational choice, not something an analyst invents under pressure. Record when that fallback applies and which evidence must accompany it.
Define the response boundary by action and asset
Use an action register with the target population, operation, prerequisite, approving role and verification owner. For example, a provider may be allowed to isolate a defined test endpoint during an exercise but not a production server supporting a critical application. The distinction should be visible in the register and reflected in technical permissions where practical.
Account for business dependencies. An identity used by an integration may affect several systems; a server may support a process that is not obvious from its name. The customer should supply the relevant context, and the provider should state when that context is missing. An incomplete asset record is a decision constraint, not an invitation to guess.
Distinguish routine, urgent and exceptional authority. Routine work can follow the normal change process. Urgent response may have a pre-agreed expedited route. Exceptional actions outside those boundaries require an explicit decision. Combining all three under “emergency response” makes it difficult to know which controls still apply during an incident.
Review the technical role against the action register. Excessive permission does not create contractual authority, and insufficient permission can prevent an authorised action. Both mismatches deserve attention. The customer and provider should verify the operating path in a controlled setting rather than discovering the mismatch during a real event.

Choose one scenario with a meaningful handoff
Select a scenario that requires information from both teams. A synthetic privileged change, suspicious test sign-in or agreed endpoint event can work if it leads to a business-context question. The scenario should be narrow enough to follow from observation to closure. A sprawling fictional crisis often produces discussion without demonstrating an actual operating process.
Write the expected decision path before execution. Identify the event source, analyst task, customer question, approval route, possible action and closure condition. Keep the expected path separate from the observed result. This prevents the exercise report from describing the plan as though it happened merely because everyone agreed with it in a meeting.
Use safe test assets and harmless markers. Avoid real customer data, live payments or disruptive changes. If an action is necessary to test the handoff, use a reversible operation against a disposable resource and establish a recovery route. The exercise should improve coordination without creating an incident of its own.
Decide whether the participants know the scenario in advance. A tabletop checks understanding; an operational drill checks the actual tools and contact routes. Both can be useful, but they provide different evidence. Label the exercise accurately and do not present a verbal walkthrough as proof that live collection or response access works.
Set the exercise controls
Name a coordinator with authority to pause the exercise. Confirm the window, permitted operations, participating identities and business systems excluded from the test. Establish a stop condition for unexpected impact or an unrelated real incident. These controls allow the teams to test realistically while maintaining a clear boundary around the activity.
Agree how real and synthetic activity will be distinguished by the coordinator. A unique harmless marker can help reconcile records, but the detection should not be tuned solely to that marker. The exercise is intended to test a capability relevant to the service, not a special rule created only to recognise the demonstration.
Protect the exercise record. Contact details, role assignments and system identifiers may be sensitive even when the scenario is synthetic. Store the detailed record in an appropriate restricted location and use a concise summary for broader management review. There is no need to publish an operational contact tree to prove that a drill occurred.
Tell participants how observations will be used. The purpose is to identify process defects and improve the service, not to embarrass a person who follows an unclear procedure. A constructive exercise encourages staff to reveal uncertainty. Concealing that uncertainty produces a polished report while leaving the real decision problem unresolved.
Make the escalation request decision-ready
The escalation should identify what was observed, which asset or identity is affected, why it matters and what the recipient is asked to decide. Include the relevant evidence reference and current uncertainty. The recipient should not need to open several inaccessible tools before understanding whether the message requires context, approval or operational action.
Use a concise distinction between facts and assessment. “A privileged setting changed at this time” is an observation; “the change may be unauthorised” is an assessment pending context. The customer can then confirm an approved activity or request further investigation. Presenting every hypothesis as a fact encourages unnecessary disruption and erodes trust.
Specify the consequence of waiting. This is not a tactic for manufacturing urgency. It helps the approver understand whether evidence may expire, unwanted access may continue or a business operation remains paused. Where the consequence is uncertain, say so. The decision should be based on an honest account of risk and operational impact.
Check that the recipient can access the supporting material through the agreed channel. A link into a provider-only console may be useful internally but useless to the customer. Provide an authorised view or a proportionate summary with references. The handoff is complete only when the receiving role has enough information to act.

Exercise an unavailable approver
During the drill, make the primary approver unavailable through simulation rather than actually disabling their account or device. Observe whether the SOC follows the documented alternate path. Record whom it contacts, what information it sends and whether the alternate recognises their authority. A contact list is not proven until its participants can use it.
The provider should remain within its authorised boundary while waiting. A missed call does not expand permission. If a pre-authorised temporary action exists, the record should identify its prerequisites and limits. If no such action exists, the exercise may reveal a decision the organisation must make before the service is accepted.
Test the customer’s internal escalation as well as the provider’s. The external team may reach the IT duty contact quickly, yet the internal route to a business owner may fail. Measure and describe those stages separately. Otherwise the review can blame the wrong team and leave the actual bottleneck unchanged.
After the exercise, ask whether the fallback is realistic outside office hours or during staff absence. Do not promise continuous availability based on one successful daytime call. Operating hours and response commitments belong in the agreement, and the drill should reflect the conditions the customer is actually buying.

Verify that the operator can carry out the approved step
Once a decision is approved, identify the operator and the intended access path. Ask that person to perform only the agreed harmless test action. Record the identity used, target, operation and result. A separate administrator performing the action on their behalf does not demonstrate that the ordinary response arrangement works.
If the operation fails, preserve the failure and diagnose it. The problem might be a role restriction, network dependency, expired credential or mistaken target. Correct the specific issue through the approved process. Broadly increasing permissions to make the drill pass can create a larger problem than the one the exercise uncovered.
Check whether the action affects investigation access. Some containment operations can change connectivity or identity state in ways that make further collection harder. The playbook should account for that dependency. The exercise is a safe place to discover the sequence needed to preserve evidence while reducing unwanted activity.
Record the recovery path and execute it before closing the test. Confirm that the test resource returns to its expected state and that temporary permissions are removed. If recovery belongs to another team, test that handoff too. Co-management should connect the whole agreed workflow, including the transition from response to normal operations.
Separate an accepted request from an effective result
A platform reporting that an action succeeded does not necessarily establish the intended business effect. Verify the relevant state independently. If the exercise restricts a test identity, check the specific access path being assessed. If it isolates a test resource, confirm both the unwanted test communication and any deliberately preserved management route.
The verification owner should know the expected result before the action. Otherwise the team may accept whatever outcome occurs. Record the observation and any difference from expectation. A partial result can still be useful if its limitation is clear; it should not be converted into a full pass for convenience.
Include the customer’s business check where relevant. The technical state may be correct while a dependent process is unexpectedly affected. A co-managed service benefits from the customer’s local knowledge, so use it. The provider should not claim that operations are unaffected solely because its own management connection remains available.
Close the loop with the approver. Report what was done, what was observed and what remains unresolved. The person who authorised an action should not have to infer completion from silence. This final handoff supports accountable decisions and prevents temporary controls from remaining in place without an owner.
Measure time in stages
Capture event creation, collection, analyst acknowledgement, escalation, customer receipt, approval, execution and verification separately. A single elapsed-time figure hides the distinction between technical processing and human decisions. The purpose is to find where the service needs improvement, not to produce the most flattering number.
Use a consistent time basis and note known source delays. If an event appears late in the monitoring platform, the analyst cannot investigate it before arrival. Conversely, quick ingestion does not excuse a case waiting in an unmonitored customer mailbox. Each stage should have an owner who can explain its observed behaviour.
Compare the results with the actual agreement. A response target may refer to acknowledgement rather than containment. State the definition clearly. Do not advertise an exercise’s fastest result as a universal commitment, especially if the participants knew a test was underway or the scenario required fewer approvals than a real incident.
Look for repeat patterns across exercises. An occasional delay may have a specific explanation; a recurring delay at the same handoff suggests a structural issue. Use the service review to change the responsibility, contact route or prerequisite that causes it. Repeating the same drill without resolving the bottleneck adds little value.

Keep a record that survives a change of analyst
The incident record should identify observations, decisions, actions, evidence references and remaining questions. Another authorised analyst should be able to continue from it without relying on the original person’s memory. This is particularly important when internal and external shifts change at different times or work spans several organisations.
Use a short handover check during the drill. Give the record to a second participant and ask them to explain the current state and next action. Their questions reveal missing context. Improve the record while the exercise is fresh rather than waiting for a real incident to expose the gap.
Capture decisions made on calls or alternate channels in the official record. The summary should state who decided what and on which basis, without transcribing every conversation. A provider’s ticket and the customer’s internal record can be linked, but the relationship and authoritative owner should be clear.
Protect sensitive details and preserve the agreed retention. An incident record may include employee information, internal architecture and access references. Limit distribution to those who need it. A useful management summary can explain the process finding without reproducing the technical evidence pack or personal contact details.
Worked example: everyone sees the alert, nobody owns containment
In a hypothetical drill, the external SOC identifies a synthetic suspicious sign-in and notifies the customer’s IT team. The IT team assumes that the provider will restrict the test account. The provider assumes that the customer must approve and execute the restriction. Both teams acknowledge the ticket, but no action follows.
The coordinator pauses the exercise at the planned decision point and asks each team to explain its understanding. The contract describes monitoring, while an informal onboarding discussion had mentioned response assistance. The discrepancy is recorded as an authority gap rather than an analyst-performance failure. The customer now has a concrete decision to make.
The teams agree a revised action register: the provider gathers the evidence and recommends a step; the customer’s named role approves; the designated internal operator performs it; the provider verifies the agreed test access path. They update the ticket template and contact route to reflect that sequence.
The repeated drill completes the sequence and records each handoff. The result demonstrates the revised workflow for the selected scenario. It does not imply that the provider has gained blanket authority over every identity or system. The value comes from making a specific shared responsibility work and documenting the boundary accurately.

Turn findings into changes that can be retested
Write findings in operational terms. “Escalation process unclear” is a useful starting observation, but the corrective action should name the decision, recipient and required update. “The duty contact could not identify the approver for account restriction” gives the customer a precise ownership problem to resolve.
Assign a verification method to each change. A new phone number should be checked, a role correction should be exercised and a revised ticket should be tested with a receiving participant. Documenting the fix is not the same as demonstrating that it works. Retest the affected step without unnecessarily repeating the entire scenario.
Prioritise defects by their practical consequence. A missing cosmetic field in a ticket differs from an unavailable approval route for a critical system. Use the likely effect on investigation, containment and business decisions to order the work. Avoid inflating every exercise observation into a severe security finding.
Keep accepted limitations visible. The organisation may decide that some actions remain advisory or available only during defined hours. Those are service boundaries to communicate, not necessarily defects to conceal. The acceptance record should make it easy for a future incident coordinator to understand which capabilities are dependable under which conditions.
Maintain the agreement as the teams change
Repeat focused drills after changes to the provider, internal duty rota, identity system, critical applications or response tooling. A workflow depends on people and access as well as documentation. A new service manager or changed approval structure can break a previously successful handoff without any change to the detection platform.
Review one complete case during the regular service meeting. Trace the observation, escalation, decision and closure, then ask whether the customer received useful information at each stage. This creates a practical quality discussion instead of limiting the meeting to alert counts and uptime statistics.
The NIST SP 800-61 Revision 3 guidance connects incident response with broader risk management and improvement. A co-managed drill is one way to apply that principle: it exposes the decisions and dependencies that must work across organisational boundaries, then gives them owners and verification steps.
The CISA logging guidance is a useful starting point for the observation side of the workflow. Collection remains necessary, but this article’s emphasis is what happens after information becomes available. The service must translate evidence into a decision the customer can actually make.

Write a co-managed service brief around the gaps
Describe your internal team, critical systems, operating hours and current tools. Identify which responsibilities you want the provider to take and which decisions must remain with the customer. Ask for an action register and onboarding drill as deliverables. This produces a more comparable proposal than requesting a generic co-managed SOC package.
Ask the provider to explain an unavailable-approver scenario and show a synthetic escalation record. Confirm how temporary access, evidence transfer and recovery handoff are handled. These questions reveal operational maturity without requiring disclosure of another customer’s confidential incident. A clear bounded answer is more useful than an unqualified promise to handle everything.
Use our managed security RFP builder or contact Atlant Security to discuss a co-managed arrangement. Start with responsibilities and business outcomes rather than credentials or sensitive incident exports. Access and evidence exchange can be agreed once the scope is understood.
Reviewed 25 October 2026. This guide does not promise particular response times or continuous monitoring. Those commitments depend on the agreed service and customer prerequisites. The method helps an organisation verify the arrangement it has chosen, identify its limits and improve the decisions that connect detection to effective action.
Published by Atlant Security. Sources, editorial policy and corrections.

