Breach Risk Direct.
Security Controls

Will Your MDR Provider Stop an Attack, or Just Email You About It?

Most MDR proposals promise "response." Some mean isolating a laptop at 3 a.m.; others mean an email you read at 8. Here's how to tell which one you're buying, and how to test it before you sign.

Breach Risk Direct editorial team · November 2026 · 9 min read

What "Response" Means When the Contract Is Signed

Two MDR proposals can look identical on paper. Both offer 24/7 monitoring, both name an EDR platform, and both promise detection and response. Then ransomware starts encrypting a file server at 3 a.m. on a Sunday. One provider isolates the host within minutes. The other opens a ticket, emails your security inbox, tries your on-call phone, and gets no answer.

The difference comes down to what the provider is allowed and able to do without asking you first. MDR response and containment covers the actions an analyst takes to stop an attack from spreading: cutting a machine off the network, disabling a compromised account, blocking a malicious address. Investigation, system rebuilds and data recovery generally sit outside it.

Advertisement

Providers fall into three broad models:

  • Alert-only (notify): The provider detects, investigates and tells you what to do. Your team or IT partner takes the action.
  • Guided or collaborative: The provider recommends an action and runs it once someone on your side approves, usually by phone, chat or a portal button.
  • Active (pre-authorized): The provider acts immediately, within limits you agreed in advance, and tells you afterward.

Several large providers sell all three as configurable modes. Sophos MDR, for example, publicly describes notify, collaborate and authorize options. So the useful question in active response vs alert-only MDR isn't which provider you pick. It's which mode you'll actually run, on which assets, and who signed off on it.

Host Isolation, Account Disablement and Network Blocking, Explained

Almost every containment playbook is built from three actions. Each depends on a different integration, and that's where gaps hide. A provider can be excellent at one and unable to perform another in your environment.

ActionWhat it doesWhat the provider needsCommon gaps
Host isolationCuts an endpoint off from everything except the EDR console, so analysts can still investigate itA supported EDR agent on the device and permission to use the vendor's isolation functionUnmanaged devices, legacy servers, OT systems, and Linux or macOS hosts with limited agent features
Account disablementDisables a user or service account, revokes sessions and forces password or token resetsAdmin-level API access to your identity provider (Entra ID, Okta, Google Workspace, on-premises Active Directory)On-premises AD needs a connector; break-glass and service accounts are often excluded; SaaS sessions may survive
Network blockingBlocks IPs, domains or hashes at the firewall, proxy, email gateway or EDR levelAPI integration with your firewall, SASE or email security platformMany providers only block within their own EDR; firewall blocking is often a premium tier or not offered

Host isolation is the most mature of the three because it lives inside the EDR the provider already operates. Identity response is where buyers get caught out. Many intrusions start with stolen credentials or hijacked session tokens, not malware. If your provider can isolate a laptop but can't kill a compromised Microsoft 365 session, the attacker may simply keep working from their own machine.

Ask which of the three actions are included at your price, which integrations they need, and whether the provider acts on your systems directly or just hands you a script to run.

MDR Rules of Engagement Decide What Happens at 3 A.M.

The rules of engagement turn "active response" into something that happens at night. Some providers call the document a response authorization matrix or a runbook. It spells out which actions the provider can take without calling you, on which assets, under what conditions, and who is told afterward.

Good rules are specific. A workable version for a hypothetical 300-person company might say:

  • Workstations and laptops: isolate immediately on high-confidence ransomware, credential theft or hands-on-keyboard activity. Notify after.
  • Standard user accounts: disable and revoke sessions on confirmed compromise. Notify after.
  • Domain controllers, ERP, payment and production servers: call the escalation list first. If nobody answers within 15 to 30 minutes and encryption or data theft is under way, isolate.
  • Executive and break-glass accounts: never disable without a named approver.
  • Clinical, manufacturing or OT systems: notify only.

The "no answer within X minutes" fallback matters most. Without it, an approval-based model becomes alert-only the moment your on-call person sleeps through a call. List at least three names in the escalation chain, and keep their numbers current.

Local rules can shape the document. In parts of Europe, works councils or employee-monitoring rules may affect how account actions on staff are handled. Regulated sectors such as healthcare, finance and critical infrastructure often have change-control duties that apply to emergency actions. Breach-reporting deadlines vary widely too, from hours in some jurisdictions (India's CERT-In rules are among the shortest) to 72 hours under the GDPR. Your rules should name who starts that clock. Have legal counsel review the authorization language, especially if the provider acts across several countries.

Where MDR Response Ends and Incident Response Begins

Containment stops the spread. It doesn't clean up. Even strong MDR response and containment usually stops short of:

  • Full forensic investigation and root-cause reports that hold up for insurers, regulators or courts
  • Eradicating persistence across the environment and rebuilding systems
  • Restoring from backup and checking data integrity
  • Ransom negotiation, breach counsel coordination and regulatory notifications

Some providers bundle a set number of incident response hours. CrowdStrike, Arctic Wolf and eSentire, among others, offer IR in higher tiers or as a separate retainer. Others hand you to a third party. Get the handoff in writing: how long it takes, who pays, and whether the IR team can work from the MDR provider's telemetry or must deploy its own tools.

Bring in your cyber insurer early. Many policies require an approved panel of IR firms and breach counsel, or at least the insurer's consent before you hire one. If your MDR provider's IR team isn't on the panel, you may end up paying for that work yourself. Confirm the details with your broker, since wording varies by insurer and market.

What Active Response Costs and What Drives the Price

Active response isn't always a line item, but it changes what you pay. Providers that act on your behalf take on more liability and need more integrations, and the price reflects that. The ranges below are typical 2026 market ranges. They vary by contract, region and volume, so confirm current pricing with vendors.

Service levelTypical indicative range (USD)What's usually included
Alert-only MDR$6 to $12 per endpoint per month24/7 monitoring, triage, notification with remediation guidance
Active endpoint response$10 to $25 per endpoint per monthPre-authorized host isolation, process kill, file quarantine
Identity response add-on$2 to $6 per user per monthAccount disablement, session revocation, identity threat detection
Network and cloud responseOften quoted per integration or as a premium tierFirewall, SASE, email and cloud workload actions
IR retainer$15,000 to $75,000+ per yearPre-agreed hours, guaranteed response SLA, forensic support

The main cost drivers are the number of endpoints and identities in scope, how many third-party integrations are needed, whether you bring your own EDR licenses or buy them through the provider, and how many after-hours escalations you expect. Prices also differ by region. Providers with local security operations centers in the EU, the Middle East or Asia-Pacific may charge more for data residency or in-region analysts.

Say a hypothetical 150-endpoint company pays roughly $900 to $1,800 a month for alert-only coverage, or about $1,500 to $3,750 for active response before any identity add-on. The extra annual spend is real but modest next to a single weekend of uncontained ransomware.

Red Flags in MDR Proposals and Contracts

When you compare managed detection and response capabilities, the warning signs usually sit in the wording, not the feature list. Watch for these:

  • "Response" without verbs. If the proposal never says isolate, disable or block, assume you're buying notification.
  • SLAs that measure only notification. A 15-minute "time to respond" that ends when an email goes out says nothing about containment. Ask for separate time-to-notify and time-to-contain targets.
  • Containment limited to the provider's own agent. Fine if you run their EDR everywhere. If you don't, servers and cloud workloads may fall outside scope.
  • No written rules of engagement before go-live. If onboarding ends without a signed authorization matrix, the default is almost always notify-only.
  • Liability caps that rule out active response. Some contracts disclaim all responsibility for business disruption caused by containment. That's reasonable. Disclaiming failure to act on agreed rules isn't.
  • Vague answers about who does the work. Find out whether the provider's own analysts run actions or a subcontracted operations center does, and where those analysts sit.

How to Test a Provider's Response Before You Sign

Demos show dashboards. They don't show what happens when an analyst must act on incomplete information at an awkward hour. Push for a proof of value that tests response directly:

  1. Run it on your own systems. Ask for a 14- to 30-day trial on a representative mix of laptops, at least one server and your identity provider, not a vendor sandbox.
  2. Agree on draft rules of engagement first. Test the mode you'd actually run in production.
  3. Simulate attacks after hours. Use a scheduled adversary simulation, such as the open-source Atomic Red Team or a commercial breach-and-attack simulation tool, to trigger credential theft and lateral movement on a Saturday night. Tell the provider a test window is coming, but not the exact time.
  4. Measure four timestamps: detection, analyst pickup, containment and your notification. Get them from the provider's own logs.
  5. Test an unanswered escalation. Have your on-call contact deliberately not pick up, and see whether the fallback rule fires.
  6. Run a tabletop on a critical server. Walk through a domain controller compromise with the provider's analysts and check that their reasoning matches what you wrote down.

The decision rule: if a provider won't join an after-hours response test during evaluation, or the containment timestamp is missing from their results, treat them as alert-only whatever the proposal says. Price them that way, and budget for your own on-call coverage.

Frequently asked questions

Is alert-only MDR ever the right choice?

Yes, if you already have a 24/7 internal security team or an IT partner who can act within minutes at any hour. It can also suit environments where any automated action is too risky, such as some OT or clinical systems. For most small and mid-size companies without round-the-clock staff, alert-only leaves the riskiest hours uncovered.

Can MDR host isolation break business operations?

It can. Isolating a server that runs payments, production or patient systems stops that service until it's released. That's why good rules of engagement treat critical servers differently from laptops, and require a phone escalation or a time-based fallback before isolating them.

Do cyber insurers require active response from MDR providers?

Requirements vary by insurer and country. Many underwriters ask about EDR coverage and 24/7 monitoring, and some ask whether a provider can contain threats without customer approval. Check your application and policy wording with your broker, and confirm whether your provider's incident response team is on your insurer's approved panel.

Who is liable if the MDR provider isolates the wrong machine?

Most MDR contracts limit the provider's liability for disruption caused by containment actions taken in good faith under agreed rules of engagement. In return, expect the provider to accept responsibility for failing to act when the rules said it should. Have legal counsel review these clauses, because enforceability and standard terms differ across jurisdictions.

How often should rules of engagement be reviewed?

At minimum once a year, and whenever you add critical systems, change identity providers, restructure escalation contacts or acquire a business. A review after every real incident is also sensible, to adjust what the provider can do automatically based on what happened.

Written by the Breach Risk Direct editorial team. Prices, features and rules change often - confirm details with the vendor or a licensed professional before you decide.