Insights

What cyber underwriters can check from the outside before they quote

Which outside-in signals cyber insurers and MGAs use before quoting SMEs, what they reveal about the proposal form, and where outside-in data stops.

A small-business cyber proposal form is short by design. Brokers want quotes in minutes and applicants won't fill in twenty pages. The cost is that most of what the underwriter knows about the risk is self-declared, once, by someone who may not know the answer.

That is why more carriers, MGAs and brokers now pair the form with an outside-in look at the applicant's domain. This article covers what that look can and can't tell an underwriter, and how to use it without overreaching.

Why it matters now

The market is already moving this way. A few public examples:

  • Howden Greece says it uses Cyberwrite software to identify the biggest online threats facing a business that buys its Cyber+ cover (Howden, Cyber+).
  • AIG Cyprus says it "continuously monitors clients' publicly-facing internet infrastructures for key vulnerabilities, infections and misconfigurations" (AIG Cyprus appetite guide).
  • Hiscox Germany offers a free scan that checks a domain for security risks visible from outside, passing it to its partner Bitsight (Hiscox DE).
  • DUAL UK added KYND in April 2026 for external scanning, continuous monitoring and vulnerability analysis, at both individual-risk and portfolio level (Insurance Business, 9 Apr 2026).
  • Coalition offers policyholders Coalition Control, which monitors "risks across your entire external digital footprint" (Coalition Control).

The claims mix explains part of the interest. In Coalition's 2025 Cyber Claims Report, business email compromise and funds transfer fraud together accounted for 60% of its 2024 claims, against 21% for ransomware (Coalition, May 2025). Email configuration is one of the things you can read from outside.

What the proposal form asks, and what it misses

Take a public example: Beazley's cyber application for businesses under £/€250m revenue (Beazley application). It asks about MFA for remote access and webmail, incoming email screening, phishing training, endpoint protection, backup frequency, critical patching of internet-facing systems, EDR vendors, and out-of-band verification of supplier bank-detail changes.

It does not ask directly about exposed remote desktop, SPF, DKIM or DMARC, or end-of-life software. Those are exactly the items an outside-in scan can see. The two sources are complementary: the form covers what only the applicant knows, the scan covers what the applicant may not know they expose.

What an outside-in check can actually see

Email authentication

SPF, DKIM and DMARC records are public DNS. A scan shows whether they exist and how strict the DMARC policy is. Microsoft's documentation describes the three policies: p=none gives "no suggested action for messages that fail DMARC", p=quarantine means they "should be accepted but marked", and p=reject means they "should be rejected" (Microsoft Learn). A domain stuck at p=none can still be spoofed.

Exposed services

Remote desktop, databases, file shares or admin panels answering on the public internet. This is the outside-in signal closest to a form question ("is remote access protected?") and the one most likely to contradict it.

Patch hygiene and end-of-life software

Servers often announce product and version. A scan can flag versions past vendor support. It can't see internal patching.

TLS, certificates and web headers

Expired certificates, old TLS versions and missing security headers. Treat them as maintenance signals: they show whether anyone looks after the internet-facing estate.

Credentials in breaches

Whether addresses on the company's domain appear in known breach data. Useful as a phishing and credential-stuffing signal.

How to use the signals in underwriting

Use Question it answers Outside-in signal
Pre-quote triage Is anything obviously wrong before we spend time? Exposed remote access, end-of-life software
Form verification Do the answers match what we can see? Remote access exposure vs MFA answer; patching answer vs software versions
Subjectivities What must be fixed before binding? DMARC at enforcement, close exposed RDP
Portfolio monitoring Has an insured's exposure changed mid-term? New exposed service, removed DMARC record
Loss prevention What can the insured fix to cut frequency? Prioritised fix list

Two practical rules help:

  1. Report coverage, not just a score. If a scan could not assess a category, it should say so. "Not assessed" scored as "pass" flatters the risk.
  2. Treat a finding as a question, not a verdict. An open port may be a honeypot, a vendor's system or a stale DNS record. Ask the broker before declining.

Where outside-in data stops

Be honest about the limits.

  • It can't see MFA, backups, EDR or training. Research from Marsh McLennan's Cyber Risk Analytics Center, pairing its claims data with self-assessment questionnaires, linked lower incident likelihood to controls such as hardening baselines, broadly deployed MFA and patching high-severity vulnerabilities within seven days (Marsh McLennan, April 2023). Most of those are invisible from outside.
  • Public sources show capability, not proven loss correlation. The vendor and carrier pages above describe what the tools do. None of them publishes a link between specific external findings and claims outcomes. Ask any provider, including us, for evidence before you build pricing on a score.
  • Small businesses have small footprints. One domain and a hosted mailbox may produce few signals. Fewer findings isn't the same as low risk.

Common mistakes

  • Using the score as the price. Use it to prioritise questions and subjectivities, not as a rating factor without validation.
  • Scanning once. Exposure changes. A clean pre-bind scan says little about month eight.
  • Sending the insured a PDF with no fixes. The insured can't act on a grade; they can act on "close port 3389 and move remote access behind a VPN with MFA".
  • Ignoring consent and data protection. Make sure the applicant knows their domain will be checked, for example through the proposal form, and that your contract with the data provider covers it. This is general information, not legal advice.

Try it on applicants you have already assessed

Velontech is in private beta. VelonRadar checks a domain daily from the outside, using public information only: email authentication, HTTPS and security headers, end-of-life software, exposed services on verified domains and leaked accounts. VelonGrade turns that into an A–F grade with coverage shown and the fixes that raise it. It doesn't see inside controls unless the insured connects their tools. If you underwrite or place SME cyber in Greece, the simplest test is a free pilot on 10 to 20 applicants you have already assessed, with their consent: see the insurer pilot, or run a free check on your own domain at velontech.eu.

Want us to check this for you?

A free check using public information only. We reply within one working day.

Free check