Remote Support Start download

When the cyber insurer refuses to pay: the typical reasons and how to neutralise them in advance

Cyber InsuranceComplianceRisk
When the cyber insurer refuses to pay: the typical reasons and how to neutralise them in advance

In the mid-market, cyber insurance is often treated as a safety net: pay the premium, risk covered. Reality in 2026 looks different. Anyone talking to claims adjusters hears the same story on repeat: contract current, premium paid — and yet payout denied or significantly cut. Not because of a formal mistake, but because the protective measures promised in the questionnaire were not demonstrably in place when it mattered.

This is no longer an edge case. Insurers learned from the ransomware wave of the early 2020s. Questionnaires are more detailed, forensic audits during claims are tougher, and policy terms contain obligations whose breach triggers loss of cover. For the mid-market, this means the policy alone is not enough. The infrastructure has to match the questionnaire — and continuously, not just on the day of signing.

This article summarises why cyber insurers reject claims and which four construction sites appear most often in our advisory practice. For context on the general 2026 requirements, start with Cyber insurance 2026: what insurers require from SMBs.

Why a cyber insurance policy denies a claim: the mechanics

First, the basic question: how can a running policy deny a claim at all? Nearly every cyber insurance contract contains so-called obligations — the policyholder’s duty to maintain certain minimum standards. These standards are set out partly in the pre-contract questionnaire, partly in the general terms.

If the policyholder breaches an obligation through gross negligence or intent, the insurer can refuse the claim in whole or in part. “Gross negligence” in German case law does not mean “with malicious intent”; it means “violating ordinary diligence to a particularly severe degree”. A perimeter server left unpatched for months falls into that bracket quickly.

In addition, there are risk exclusions: specific causes of loss excluded by the contract. If, for example, the required multi-factor authentication is missing on a critical access path, and the attack happens exactly through that path, the “insufficient access protection” exclusion often applies. The non-payment is then not a discretionary decision but a clean contractual conclusion.

The typical denial reasons boil down to four areas: patch management, access protection, backups and logs. Everything else — training, segmentation, incident response plan — is important, but the disputed claims are almost always decided on one of these four points.

Reason 1: patches are missing — the classic

By far the most common denial reason is the unpatched server. Specifically: a known vulnerability on a publicly reachable system, for which a vendor patch has been available for weeks or months, becomes the entry point. In a claim, the insurer checks when the vulnerability became public, when the patch was available, and when it was installed. The larger the gap between patch availability and attack, the more likely the coverage reduction.

Questionnaires now contain concrete SLA figures: critical patches within 14 days, high-severity within 30 days. Whoever ticks that box will be measured against it in a claim. Not against “we patch regularly”, but against the audit trail: when was which patch rolled out to which system?

What helps here is not heroism but process. Centralised patch management with reporting shows the history during an audit. For Linux servers that can be Ansible or an RMM; for Windows, WSUS, Intune or a commercial RMM tool. What matters is the verifiable record, not the specific product.

Three system classes typically under-patch in mid-market environments:

  • Perimeter firewalls and VPN gateways — the first thing the insurer looks at
  • Hypervisors (Proxmox, VMware ESXi, Hyper-V) — a compromised hypervisor is the worst case
  • Mail gateways and reverse proxies — publicly reachable, often forgotten during patch windows

Teams running Proxmox clusters will find a workable approach in our Proxmox SSH key management and cluster rotation article.

What the insurer actually wants to know in a claim

The forensic investigation after an incident reconstructs the attack path. It also surfaces when a system has not been updated in months. The insurer then asks for proof of the patch process. If this is missing or shows gaps exactly on the compromised systems, “we are insured” quickly turns into “contributory negligence — payout cut by 50 percent”.

Reason 2: no MFA — or MFA with holes

Multi-factor authentication is the first mandatory field in every current questionnaire. The wording is typically “MFA for all administrative and external access”. That sounds clear, but in practice the holes open up at the exceptions:

  • Service accounts without MFA that have been running for years
  • Legacy protocols such as IMAP, POP3 or SMTP-Auth that bypass MFA
  • Break-glass admins with password only, “just in case MFA fails”
  • Vendor accounts set up years ago and never reviewed

Each of these exceptions is the point where an attacker walks past the questionnaire. And each is an argument for a coverage cut in a claim. The insurer argues, in effect: “You promised MFA. The attack came in via an account without MFA. The promise therefore was not kept.”

The solution is not an exotic product but an honest inventory. Which access paths exist? Which are externally reachable? Where does MFA run, where not, why not? Every exception has to be justified and documented, ideally with compensating controls — network isolation, additional logging, time or location restrictions.

For moving towards passwordless login and FIDO2 see Passkeys for mid-market: passwordless. For SSO-based access consolidation Authentik: self-hosted single sign-on is a practical starting point.

SMS-based MFA no longer counts as MFA in 2026

A side point that increasingly appears in questionnaires: SMS-based MFA is no longer accepted as full MFA by many insurers. The reason is SIM swapping. Anyone declaring SMS MFA in newer policies gets either an underwriter note or an outright contractual exclusion. Better: app-based TOTP or hardware tokens.

Reason 3: the backup obligation is not met — but everyone thinks it is

The third large construction site is backups. Every questionnaire asks about the backup strategy: are backups made, how often, where do they live, are they offline or immutable, is restore tested regularly? The answers in German SMBs are rarely fully correct.

The typical gaps:

  • “3-2-1 is implemented” but all three copies sit on connected systems that ransomware reaches
  • “Backups are immutable” but the immutability flag is a software policy, not a true write-once medium
  • “We run regular restore tests” but the last documented test is two years old
  • “Air-gap backup in place” but the air-gap server is online 24/7, just on a different AD

In a claim the insurer does not test the theory; the insurer looks at what actually happened. If the backups were encrypted along with everything else, the 3-2-1 rule was clearly not fulfilled. If the last successful restore was a year ago, the promise of “regular restore tests” does not hold up.

In practice we recommend a clean backup vault: immutable backups with S3 Object Lock, or Proxmox Backup Server with a genuinely separate credential set, physically or at least logically strongly isolated. Done correctly, this satisfies the policy terms and provides the best ransomware resilience. See TrueNAS immutable backup with Veeam and Object Lock and the 3-2-1-1-0 backup formula.

On the hardware side: a dedicated TrueNAS backup target is often the most economical choice for the mid-market. The right configuration can be scoped in the TrueNAS configurator; we handle ongoing operations through TrueNAS Support.

The restore evidence is the decisive document

Any post-incident audit asks for restore evidence: on what date was which system restored for test purposes, how long did it take, was the result complete? See Backup 30-day restore drill for a process that has proven itself. Without this evidence, any backup promise is attackable.

Reason 4: logs are missing — the SIEM gap as denial reason

The fourth large reason is the fastest-growing one: missing or insufficient logs. In 2026 insurers increasingly ask explicitly for a SIEM, or at least central log aggregation with forensic retention. The background is that without logs the attack path cannot be reconstructed cleanly after an incident. For the insurer this means an unclear evidence picture, often to the policyholder’s disadvantage.

Typical gaps in mid-market environments:

  • Logs sit locally on the compromised system and are destroyed with it
  • Firewall logs are overwritten after a few days — but the attack happened weeks earlier
  • No central timeline — each system has its own logs, no correlation possible
  • No alerts — logs are written but nobody looks at them

What is enough is usually not a large SIEM suite. A central log server (Loki, Graylog, Wazuh, Elastic) with the critical sources configured, at least 90 days of retention for firewall, VPN and AD logs, and a few sharp alerts (failed logins, new admin rights, unusual VPN locations) cover the bulk.

Crucially, logs must not live on the same system as what they monitor. A log server in the same VLAN as production helps the attacker cover their tracks. Better: separate management network, separate backup server, logs with WORM character.

For a mid-market SIEM implementation see Wazuh for SMBs: open-source SIEM in practice and OPNsense logging and monitoring.

How to neutralise the four traps in advance: the honest questionnaire

The most effective preparation for cyber insurance — and for any eventual claim — is an honestly filled-out questionnaire before it is submitted. Every “no” in this internal pass is a to-do item, not a reason to tick anyway. The four construction sites above set the priorities:

  1. Patch audit: which externally reachable systems run which version? Where do the known CVEs stand? Is there a documented patch process?
  2. MFA inventory: which access paths are external? Which have MFA, which do not, which exceptions exist?
  3. Backup stress test: last documented restore? Are the backups genuinely isolated or only “offline” in the same network?
  4. Log check: are firewall, VPN and AD logs centrally retained for at least 90 days? Are there alerts someone actually reads?

Teams with these four points in order not only meet most insurance requirements — they measurably reduce real attack surface. That is the core point: the measures insurers demand are not compliance theatre. They are the foundations of modern IT security, derived from the last decade of actual claims.

DATAZONE accompanies mid-market companies through this preparation calmly and without panic. Often a manageable effort is enough to reach a state where the questionnaire can be answered honestly with “yes” and will not become a problem in a claim. Concrete projects start through our IT service or within managed services.

FAQ

Can cyber insurance really deny payment entirely?

Yes. In the case of grossly negligent breach of a contractually promised obligation, full denial is possible. More common is a proportional reduction depending on severity. In practice we see cuts between 25 and 100 percent, depending on how clearly the gap is documented and how directly it caused the loss.

Is it enough to fill in the questionnaire correctly at contract signing?

No. The obligations apply across the full contract term. Anyone ticking “MFA everywhere” at signing and two years later allowing an MFA-free legacy path breaches the promise from that moment on. A recurring internal check is therefore sensible, ideally annually and after larger infrastructure changes.

What is the most common denial reason for mid-market companies?

In our experience it is the combination of unpatched perimeter system plus missing logs. The attack comes in via a vulnerability known for months, the course cannot be reconstructed forensically, the insurer documents contributory negligence and cuts the payout. Both are gaps that are closed with process, not with a product purchase.

Does a higher sum insured help against denial?

No. The sum insured is the upper limit of payment, not the quality of the contract. A policy with ten million euro cover denies just as readily as one with one million, when the obligations are breached. More important than the headline figure are the sublimits, the exclusions, the obligations — and your own implementation of the minimum standards.

What is the difference between breach of obligation and risk exclusion?

A breach of obligation is a policyholder’s violation of a duty, triggering loss of cover or reduction. A risk exclusion is a contractually excluded loss cause where the insurer does not pay from the outset. Example: “war and cyber war” is typically an exclusion, “missing MFA” is a breach of obligation. Both lead to non-payment, with different legal reasoning.

Is cyber insurance worth it at all for a 20-person company?

In principle yes, but only when the foundations are in place. Without MFA, patches, immutable backups and logs the policy is, in a dispute, worth little and the money is better invested in those measures first. Once the foundations are in place, the policy is a reasonable add-on for residual risks such as business interruption, data-protection follow-up costs or crisis communication. The order matters: technology first, policy second.

Need IT consulting?

Contact us for a no-obligation consultation on Proxmox, OPNsense, TrueNAS and more.

Get in touch