Quantification

Putting a Dollar Figure on Cyber Risk, and Showing the Work

XR
Xcigence Research
Cyber Risk Intelligence Team
5 min read

Boards ask a fair question: how much could a cyber incident cost us? A single confident number is easy to produce and hard to trust. We built our answer around the opposite idea: every figure shows what supports it, what does not, and how it was checked.

Why the number matters, and why it is easy to get wrong

Published research gives a sense of scale. IBM's Cost of a Data Breach Report 2026 puts the global average cost of a breach at $4.99 million, up 12% on the year. Cyentia's Information Risk Insights Study 2025 reports that a typical firm has about a 9.3% chance of a significant security incident in a year, and a median reported loss of roughly $603,000, with the worst 5% of incidents reaching $32 million.[1][2]

Those figures describe populations, not your organisation. IBM's average comes from a non-statistical sample of breaches between about 2,600 and 115,000 records, so it is a mean, not a worst case. A number copied from a report into a board paper is a guess with a citation. We wanted something that starts from your own evidence.

1. Start with what the platform can actually observe

Our cyber-risk model takes its inputs from real platform data: your sector, how exposed your assets are, whether scan findings include vulnerabilities known to be exploited in the wild, how mature your controls are, and whether any compromise indicators have been raised. Before you run anything, an evidence readiness view lists each input as observed or not observed, explains why, and says what to do next.

When the platform has no real signal for an input, it says so. The model falls back on a general prior and the evidence grade drops. We do not assume the best because nobody has looked.

2. Label every default as a default

Any loss model needs parameters: how often attacks are attempted, how often they succeed, what a loss costs. Ours start as clearly marked defaults. A parameter only counts as sourced when it has validated values, a citation naming the publisher, report, edition and figure used, and a named reviewer with a stated credential who is independent of the person entering the values.

Sign-offs expire after a year, because evidence ages. A sign-off can be withdrawn with a reason, and the history is kept. Until every scenario in a run is sourced and signed off, the result is described as indicative and the grade is capped. We would rather show you a lower grade than a flattering one.

3. Check the result against published studies, honestly

Each run is compared with published reference points: IBM's industry averages and Cyentia's incident probability. A result outside the expected range is a prompt to investigate, not proof of an error, and a result inside it is not proof of accuracy. Nothing in that comparison changes the answer.

One rule we added on purpose
If a signed-off parameter cites the same publication a check compares against, the check is labelled “not independent”. Agreement with the source you copied from shows nothing, and we will not present it as validation.

4. Make it reproducible

Every result records the model version, the inputs fingerprint and the random seed, and the same inputs reproduce the same figures exactly. An auditor can rerun a number and get the number.

5. Be plain about what it is

The output is a range from a scenario model, tested against limits that are set per sector. It is not a forecast and not a professional opinion on any individual loss. Sector limits are proposed starting points until your risk advisers review them for your organisation. The aim is a better conversation with your board, your insurer and your regulator, one in which people can see what the number rests on.

What this changes in practice

  • Security teams can show which missing evidence is holding the grade down and fix it.
  • Finance and risk teams get ranges and a sector limit test, with the assumptions visible.
  • Auditors and assessors can trace a figure back to its inputs and sources.

See it on your own data

A short walkthrough using your sector and your assets.

Part of our guide to
Cyber Risk Quantification

Translating measured likelihood into financial exposure in dollars.

References

  1. [1]IBM. IBM and Ponemon Institute, Cost of a Data Breach Report 2026
  2. [2]Cyentia Institute. Information Risk Insights Study (IRIS) 2025

See your own cyber risk score

Xcigence computes a standardized 300–850 cyber risk score for your organization and every vendor in your ecosystem — no agents, no questionnaires.

We use cookies to improve your experience, analyze traffic, and support our marketing. See our Privacy Policy.