Home / Blog / Career and Certs

Intermediate Career and Certs

How to Onboard New Analysts in a Multi Tenant MSP SOC

A 90 day onboarding plan for new analysts in an MSP SOC: least privilege client access, shadowing, graduated queue ownership, SC-200 study and metrics that show readiness.

A new analyst in an MSP SOC has to learn your tools, your processes and a dozen client environments that all behave differently, while the queue keeps moving. This guide gives SOC managers and team leads a practical 90 day onboarding plan that protects client SLAs, keeps access tight, and gives you evidence of when a new starter is ready to work alone.

Who it is forSOC managers, shift leads and senior analysts who mentor
Timeframe90 days, in four phases
OutcomeAn analyst trusted to own the queue across every client
>

Scenario: Two analysts start on Monday. Your SOC monitors fourteen client tenants in Microsoft Sentinel through Azure Lighthouse, each with its own runbook, escalation contacts and SLA. Last time, a new starter closed a true positive as benign on day nine because nobody had told them that one client's finance team runs scheduled PowerShell every night, and another client's does not. You want this intake to go differently.

Prepare access before day one

The fastest way to lose a client's trust is to give a new starter more access than they need. Set up access in the week before they arrive, and grant it in stages that match the plan below.

If you manage Sentinel through Azure Lighthouse, put new starters in a separate group in your managing tenant and map that group to read only roles in each delegation, such as Microsoft Sentinel Reader. Move them into the Responder group, with Microsoft Sentinel Responder, only once they reach the queue ownership phase. Engineering roles such as Microsoft Sentinel Contributor belong behind eligible authorizations, which use Privileged Identity Management (PIM) to grant access just in time, and new starters should not be eligible for them at all in the first 90 days.

For Microsoft 365 and Defender work through Partner Center, the same rule applies to granular delegated admin privileges (GDAP): request the least privileged roles your contract needs, assign them to security groups, and add new starters to the narrowest group.

!

Watch out: Some client contracts restrict who can see their data, for example requiring UK based staff or named, vetted individuals. Check each client's contract schedule before you add a new starter to their delegation, and keep a record of who has access to which client.

  • Accounts created in your managing tenant with phishing resistant MFA registered.
  • Added to the read only group for every client they are cleared to see.
  • Ticketing system, chat channels and on call rota access in place.
  • A named buddy for the first 30 days, on the same shift pattern.
  • A lab or training workspace where they can break things safely.

Follow a four phase plan

Each phase has a clear exit test, so progression is based on evidence rather than time served. Some analysts move faster, and that is fine as long as the test is passed.

  1. D1
    Orientation, days 1 to 10Tools, ticket workflow, escalation paths and a walkthrough of every client profile. Exit test: explain how to escalate a High incident for any client.
    Read only
  2. D11
    Shadowing, days 11 to 30Sit with the buddy, watch triage and write the notes on incidents the buddy works. Exit test: correct triage notes on 20 incidents.
    Read only
  3. D31
    Supervised ownership, days 31 to 60Own Low and Medium incidents with a senior reviewing every closure. Exit test: two weeks of closures with no review corrections.
    Responder
  4. D61
    Independent queue, days 61 to 90Work the full queue, review sampled rather than total, and take part in one client call. Exit test: sign off from the shift lead.
    Responder
+

Tip: Keep a short client profile for each tenant: what the business does, critical systems, known noisy detections, approved admin tools, escalation contacts and SLA. New starters learn faster from fourteen one page profiles than from fourteen full runbooks, and the profiles help your existing team too.

Teach the client context, not just the tools

In a single organisation SOC, an analyst learns one normal. In an MSP, normal changes with every tenant. The day nine mistake in the scenario was not a skills gap, it was a context gap. Build context into the plan deliberately.

Same alert, different meaningPick three detections that fire across many clients and show how the right response differs per tenant.
Who to callEach client has its own contacts and hours. Drill the escalation path until it is automatic.
What the client seesShow real (sanitised) client updates so new starters learn the tone and detail clients expect.

A useful exercise in week two: give the new starter the same anonymised alert in three client contexts and ask them to write the triage note and escalation decision for each. Review the differences with their buddy.

Build certification into the plan

Certifications do not replace time on the queue, but they give new starters structure and give you a common baseline. If your SOC runs on Microsoft Sentinel and Defender XDR, the SC-200 exam (Microsoft Security Operations Analyst) maps closely to daily work: managing the security operations environment, responding to incidents and threat hunting with KQL.

Microsoft updates the SC-200 skills measured from time to time, and recent updates have added newer Sentinel and Security Copilot features. Before you set a study plan, check the current study guide on Microsoft Learn so your team is not revising last year's objectives. The certification renews annually through a free online assessment.

Days 1 to 30

KQL foundations

Filtering, summarising and joining in the training workspace, using the tables your SOC queries most.

2 hours a week, protected
Days 31 to 60

SC-200 study

Work through the Microsoft Learn path alongside real incidents, linking each module to a client case.

Practice assessment by day 60
Days 61 to 90

Exam and apply

Sit the exam, then write one hunting query that solves a real problem for a client.

Query added to the team library

Measure readiness with real queue data

Gut feeling is a poor guide to whether someone is ready. Sentinel records incident ownership and outcomes in the SecurityIncident table, so you can see each analyst's work directly. Because the table keeps a new record each time an incident changes, take the latest record per incident first. Run this from your managing tenant across the client workspaces you can see.

Microsoft SentinelKQL
SecurityIncident
| where TimeGenerated > ago(30d)
| summarize arg_max(TimeGenerated, *) by IncidentNumber, TenantId
| extend Analyst = tostring(Owner.userPrincipalName)
| where Analyst in~ ("new.analyst1@msp.example", "new.analyst2@msp.example")
| where Status == "Closed"
| extend HoursToClose = datetime_diff("minute", ClosedTime, CreatedTime) / 60.0
| summarize Closed = count(), MedianHours = percentile(HoursToClose, 50), TruePositives = countif(Classification == "TruePositive"), FalsePositives = countif(Classification == "FalsePositive"), Undetermined = countif(Classification == "Undetermined") by Analyst, Severity
| order by Analyst asc, Severity asc

The numbers on their own do not prove quality. Pair them with a review sample: pick five closed incidents a week per new starter and check the classification, the notes and whether the client was told what they needed to know.

Readiness check example targets for a new analyst at day 60
40+Incidents closed in 30 days
0Missed SLA breaches
95%Review sample passed first time
100%Client profiles walked through
3Incidents left Undetermined
1Client update drafted and sent
2Escalations made to the wrong contact
YesSC-200 practice assessment passed
Decision: extend supervised ownership by two weeks, focusing on escalation contacts and closing out Undetermined incidents, then reassess.
i

Note: The targets above are illustrations, not benchmarks. Set your own from what your experienced analysts achieve on the same queue, and agree them with the new starter on day one so nothing comes as a surprise.

Hand over safely across shifts

New starters are most likely to drop something at shift change. Make handover a skill you teach and check, not a chat message.

  • Every open incident has an owner, a current status and a next action in the ticket, not only in chat.
  • Anything near an SLA deadline is called out by name and client.
  • Client communications sent during the shift are listed with times.
  • The new starter reads their handover aloud to the incoming lead for the first month.
  • Questions they could not answer go into the handover, so the next shift can pick them up.

Key takeaways

  1. Grant client access in stages, starting read only, and keep engineering roles behind just in time access.
  2. Use a four phase plan with an evidence based exit test for each phase.
  3. Teach client context deliberately, because the same alert means different things in different tenants.
  4. Build SC-200 study into protected time, and check the current skills measured before you plan.
  5. Measure readiness from the SecurityIncident table plus a weekly review sample, not gut feeling.

A good onboarding plan pays for itself in fewer misclassified incidents, fewer awkward client calls and analysts who stay. Write yours down, share it with the new starter on day one, and improve it after every intake.

Written by Stacey, SOC Manager, MSP

Keep going

More in Career and Certs

BeginnerCareer and Certs

How to Build a Home SOC Lab with Free Tools

A beginner friendly plan for a home SOC lab using Windows evaluation builds, Sysmon, a free SIEM and Atomic Red Team, plus how to turn it into interview evidence.

7 Nov 20267 min read