Guide, read 10 Nov 2026
The first 30 days should end with every in-scope source reporting, a signed authority matrix and one tested escalation
Week one: people and permissions
Name two contacts on your side for each shift, with phone numbers that ring. Give the provider a list of critical systems, a map of the network and a note of anything fragile. Agree the channel for urgent calls and the channel for routine tickets, and test both before the end of day three.
Set the authority matrix now, not after the first alert. The guide on authority to act on this site covers it. If the service runs in Microsoft's Defender Experts, check which access level you granted, because the page read on 10 Nov 2026 says experts act only with Security Operator access.
Week two: data sources
The UK NCSC describes onboarding as adding systems to the scope of a SOC so that their logs are collected and monitored, and says onboarding is never complete because no IT estate stays the same. Treat it that way: keep a list of sources with three columns: connected, verified and owner.
Verified means a test event was seen by the provider, not that a connector shows green. Check endpoint agents, identity logs, email, cloud audit logs and firewall logs in that order. Decide log retention at the same time. NCSC says it can be months before incidents are detected, and recommends storing your most important logs for at least 6 months. Ask where the provider holds your data and for how long.
- Endpoint agents: percent of assets reporting, by group.
- Identity: sign-in and admin-change logs.
- Email and cloud audit logs.
- Network and firewall logs, if in scope.
- Anything excluded, written down with the reason.
Week three: tuning
Expect noise at the start. NIST's glossary defines a false positive as incorrectly classifying benign activity as malicious. Each week, review every alert the provider closed as benign and every one it escalated. Agree exclusions in writing, with an owner and a review date, so a quiet console does not hide a blind spot.
Ask the provider whether it runs new detections against stored history on deployment. IDC says Kroll runs new detections against 30 to 90 days of historical client SIEM and EDR data on deployment. Whether another provider does the same is a fair question to put to each, not an assumption.
Week four: a live test and a review
Run a controlled test. Many providers publish targets you can measure. Sophos lists service level targets of 2 minutes to create a case and 30 minutes from case creation to first outreach, evidence collection or response action, and says these are targets. Time your own test against whatever your contract says, with the clock start written down.
Hold a review at day 30 with the provider's lead analyst. Bring the source list, the closed-alert review, the test result and any action taken outside the matrix. Leave with a dated list of changes.
What does done look like after 30 days of MDR?
By day 30, every in-scope source reports and is verified. The authority matrix is signed. A test alert reached the right person within the contracted time, and the undo for an automatic action was tried. The exclusions list is short and owned. If any item is missing, ask for an extension in writing before the first invoice for the steady-state service.
Related pages
- All guides
- MDR vs MSSP, the full comparison
- MDR glossary
- The MDR Service Matrix
- Kinds of managed detection service
- Authority to act in an MDR service
- What to check in an MDR contract
- How MDR is priced
Sources
- NCSC, Onboarding systems and log sources, read 10 Nov 2026
- NCSC, 10 Steps: Logging and monitoring, read 10 Nov 2026
- NIST CSRC glossary, false positive, read 10 Nov 2026
- Microsoft, Defender Experts managed response, read 10 Nov 2026
- IDC MarketScape MDR 2024 (copy hosted by Arctic Wolf), read 10 Nov 2026
- Sophos, MDR service description, read 10 Nov 2026
Last reviewed 10 Nov 2026.