Auditvault Pro.
Underwriting

EDR Requirements for Cyber Insurance: Six Deployment Gaps Underwriters Keep Finding

Ticking "yes, we have EDR" is easy. Underwriters now want to know where it runs, whether it blocks, who watches it overnight and which systems you left out. Here's how to close each gap before renewal.

Auditvault Pro editorial team · November 2026 · 8 min read

The Coverage Percentage You Report Is Probably Wrong

Your renewal application asks whether endpoint detection and response is deployed on all endpoints. Most finance and risk teams forward the question to IT, get a yes and tick the box. That box now carries real weight. EDR requirements for cyber insurance have moved from "do you own a product" to "show us how it's deployed." Carriers and MGAs in London, continental Europe, North America, Singapore and Australia increasingly follow up the questionnaire with supplemental ransomware forms and underwriting calls. After a loss, the forensic report shows exactly which machines had a working agent.

If that report contradicts your application, you have a problem. How serious it is depends on where you are. In the UK, the Insurance Act 2015 generally applies proportionate remedies when a commercial policyholder breaches the duty of fair presentation. US outcomes vary by state, and civil-law countries have their own misrepresentation rules. Everywhere, a disputed claim is the worst time to find out your answer was optimistic.

Advertisement

The first gap is the denominator. IT usually calculates coverage as agents in the console divided by agents in the console. That's 100% by definition. The number that matters is healthy agents divided by every device that should have one.

What to do instead: reconcile the EDR console against independent inventories: your directory (Active Directory or Entra ID), MDM, cloud asset lists, vulnerability scanner and network discovery. Any device that appears in one of those but not in EDR is either a gap or a documented exception. Take the result to renewal: total in-scope devices, devices with a healthy agent, the resulting percentage and a named exception list.

Servers and Cloud Workloads Left Outside the Agent Rollout

Laptops get EDR because MDM pushes it automatically. Servers often don't. A different team owns them, and that team worries about performance, reboots and vendor support. Cloud workloads slip through for another reason: auto-scaling groups launch from images built before anyone added the agent. Ransomware operators target exactly these systems, because domain controllers, file servers, backup servers and hypervisors are where encryption does the most damage. That's why underwriters now ask about server coverage specifically.

Red flags to check for:

  • Server and workstation coverage are reported as one figure, so a 70% server number hides inside a 98% total.
  • Linux hosts were never in scope because the original rollout only covered Windows.
  • Cloud virtual machines spin up and down faster than anyone can install agents by hand.
  • Hypervisors such as VMware ESXi can't run a conventional agent, and nobody has documented what protects them.

What to do instead: bake the agent into golden images and infrastructure-as-code templates so every new instance registers at boot. Turn on your vendor's cloud options, such as Microsoft Defender for Servers through Defender for Cloud or the cloud workload modules from CrowdStrike, SentinelOne or Trend Micro. For hypervisors, document the compensating controls: an isolated management network, MFA on the management console, lockdown mode and EDR on every guest. Report server coverage as its own line, and cloud coverage by account or subscription.

Detect-Only Policies That Watch Ransomware Run

Many organizations deploy EDR in audit or detect-only mode so it won't break business applications during rollout. Then the project closes and nobody flips the switch. In that state, the tool records encryption while it happens. Underwriters ask about prevention mode because a detect-only sensor depends on a person reacting in time, and at 3 a.m. on a holiday weekend that rarely happens.

A quieter version of the same gap is the exclusion list that only grows. Every time an application broke, someone added a folder or process exclusion. Years later, broad entries such as a whole drive letter or a scripting engine leave blind spots attackers find quickly.

What to do instead: move to blocking in stages. Run a pilot group for two to four weeks, tune, then roll out by policy group, with servers last because they need the most testing. Review exclusions quarterly and give each an owner and a reason. At renewal, bring a policy export or screenshots showing prevention enabled per group, plus the date of the last exclusion review. If some groups must stay in detect mode, say so and give the switch-over date. A dated plan reads far better to an underwriter than an unexplained gap.

Alerts Nobody Reads After Business Hours

Alerts only help if someone acts on them. Most intrusions that end in ransomware leave a window, sometimes hours and sometimes days, between initial access and encryption. If your console is watched 9 to 5 in one time zone, most of that window goes unwatched. That's why 24/7 EDR monitoring now appears in questionnaires for mid-size buyers and, increasingly, smaller ones.

There are three realistic ways to cover it. The figures below are typical 2026 market ranges; they vary by region, contract length and endpoint count, so confirm with vendors.

OptionTypical costBest fitWatch for
Vendor MDR (e.g., CrowdStrike Falcon Complete, SentinelOne's MDR service, Sophos MDR)Roughly $8 to $25 per endpoint per month, EDR license includedCompanies with 50 to 2,000 endpoints and no SOCWhether the service can isolate hosts or only notifies you
Independent MDR or MSSP (e.g., Arctic Wolf, Expel, regional providers)Roughly $5 to $20 per endpoint per month on top of the EDR licenseMulti-vendor estates and in-region data residency needsAnalyst locations, language coverage, response SLAs
In-house 24/7 SOCOften $500,000 to $1.5M+ per year in high-salary markets, since round-the-clock shifts need around five or six analysts; lower where salaries are lowerLarge enterprises with existing security teamsCoverage holes from sickness, turnover and holidays

Hypothetical example: a 300-employee logistics firm with offices in Dubai and Johannesburg runs about 380 endpoints. At $12 to $18 per endpoint per month, vendor MDR costs roughly $55,000 to $82,000 a year, a fraction of what a single in-house shift rotation would cost. Below about 2,000 endpoints, MDR is the better choice for most companies on cost alone, and it gives you a contract and SLA to show the underwriter.

Evidence to bring: the MDR contract or order form, the response SLA and written confirmation that the provider may contain hosts without waiting for your approval.

Stale Agents and the Legacy Systems Nobody Wants to Touch

Agents that are installed but not working

An agent that hasn't checked in for 45 days still counts toward your coverage number. It protects nothing. The same goes for sensors on versions the vendor no longer supports, and agents with tamper protection off, which a local admin or an attacker with admin rights can simply uninstall. Post-incident investigations often find this pattern: EDR was "deployed," but not running on the machine where the attack began.

What to do instead: run a monthly health report covering last check-in, sensor version and tamper-protection status. Reinstall or retire anything silent for more than 14 to 30 days. Use staged auto-update rings to keep versions supported, and require an uninstall token or console approval before any agent can be removed.

Legacy and operational systems carved out of scope

Old Windows Server builds, factory control systems, medical and lab equipment and point-of-sale terminals often can't run a modern agent, or their manufacturer won't support one. Excluding them can be reasonable. Excluding them silently is not.

What to do instead: keep an exception register listing each system, its business reason, its owner, its compensating controls (segmentation, no internet access, application allowlisting, extended security updates, monitored jump hosts) and a retirement or upgrade date. Underwriters generally accept a short, controlled exception list. They don't accept discovering those systems after a claim.

Building the Renewal Evidence Pack Gap by Gap

Start about 90 days before renewal. That leaves time to fix gaps rather than just describe them, and to brief your broker before the underwriting call.

GapWhat underwriters probeEvidence to present
Coverage miscountedTrue EDR coverage percentageReconciliation against directory, MDM and cloud inventories
Servers and cloud missingServer and workload coverage reported separatelyCoverage by asset class; golden-image policy
Detect-onlyPrevention or blocking enabledPolicy export; exclusion review date
No 24/7 monitoringWho responds after hours, and how fastMDR contract, SLA, containment authority
Stale agentsAgent health and tamper protectionMonthly health report
Legacy exclusionsUnprotected systems and their riskException register with compensating controls and end dates

Before you sign, have your broker or a licensed adviser in your jurisdiction check the wording for any warranty or condition tied to EDR. Some forms link coverage, a sublimit or coinsurance to controls being in place at the time of loss. The decision rule is simple: fix any weak row before you answer the questionnaire, and attach a dated remediation plan for whatever is still open.

Questions buyers ask

What EDR coverage percentage do insurers expect?

Many applications ask whether EDR covers all endpoints and servers, and underwriters generally look for figures in the high 90s with any shortfall explained. Thresholds differ by carrier, market and company size. Being able to show how you calculated the number, and what's excluded, matters more than the headline figure.

Does Microsoft Defender count as EDR for insurance purposes?

Usually, if it's an EDR-capable tier such as Defender for Endpoint Plan 2, Defender for Business or Defender for Servers, not just the built-in antivirus. Underwriters care about capabilities like behavioral detection, host isolation and central management rather than the brand. Confirm which tier you're licensed for before describing it on the application.

Can our internal IT team provide the 24/7 monitoring instead of an MDR service?

Yes, if there are genuinely staffed shifts or an on-call rotation with defined response times and authority to isolate machines. One administrator getting alert emails on a phone rarely qualifies. For most companies without a security operations team, MDR is cheaper and easier to defend at renewal.

What if a critical legacy system can't run an EDR agent?

Disclose it and document compensating controls such as network segmentation, no internet access, application allowlisting and a monitored jump host, along with an upgrade or retirement date. A short exception register with a named owner per item is generally viewed far more favorably than an undisclosed gap.

Could a claim be reduced or denied if EDR wasn't deployed as we stated?

Possibly. The outcome depends on the policy wording, any control warranties and the misrepresentation rules where you are, which differ considerably between countries and between US states. Have a broker or licensed coverage counsel review the wording, and make sure your answers match what a forensic investigator would find.

Will better EDR deployment lower our cyber insurance premium?

It can improve your terms, though the bigger gains are often access to higher limits, better ransomware sublimits or retentions, or avoiding a declinature rather than a large price cut. Results vary by carrier and market cycle, so ask your broker to take your evidence pack to several markets.

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