Audit type · PSA

Phishing Simulation Audit

A phishing simulation audit measures the defensive chain, not the people in it. One bait, up to three variants, one named mailbox. Every result is a measurement with a header line behind it — never a judgement about a colleague.

Four questions, four answers with evidence

01

Does it arrive?

Delivery, bounces, greylisting, ESP events.

02

Where does it land?

Inbox, spam, category tab or quarantine — read from the mailbox, not guessed.

03

Do the technical layers hold?

SPF, DKIM, DMARC alignment and the gateway verdict, taken from the received headers.

04

What happens on click?

Link rewriting, warning interstitial, block page or nothing at all.

Variant catalogue

IDClassWhat it proves
V1Authenticated brand lureWhether content filtering catches a technically valid message
V2Forged senderWhether the domain policy (DMARC) is actually enforced
V3Cousin domain / display nameWhether lookalike senders and reply-to mismatches are seen

V4–V6 (internal sender, thread hijack, QR lure) require separate approval because they touch staff or devices.

What is measured

IDMeasurementSource
M1DeliveryESP event or SMTP 250
M2Mailbox placementlabelIds / InferenceClassification
M3–M5SPF, DKIM, DMARC resultAuthentication-Results, Received-SPF
M6Link handlingOriginal HTML vs. delivered HTML
M7ClickLanding page event, no input captured
M8ReportingQualitative, only if a process exists

Phishing Resilience Score

Fixed weights so campaigns stay comparable: authentication 30 · mailbox placement 25 · filter verdict 20 · link handling 15 · human response 10. Placement is scored in the correct direction per variant — inbox is right for a valid message, spam is right for a forgery.

ScoreMeaning
85–100The chain holds; remaining risk sits with the human
70–84Technically solid, gaps in link protection or DMARC
50–69At least one layer is ineffective
below 50Phishing is deliverable today and it works

Guardrails

  • No credential capture — no login form, no password field, ever.
  • One recipient per message, at most three messages per campaign.
  • No attachments, no macros, no QR codes for standard variants.
  • No third-party brands, no real person names, no authorities.
  • Landing page: no cookie, no third-party script, noindex, tokenised.
  • Raw evidence stays internal; the report carries hashes, not lures.

What you receive

  • Campaign document with authorisation, scope and abort criteria
  • Chronological operator log
  • The exact payload of every variant
  • Raw evidence with SHA-256 list
  • Report in house format, closing with attacker impact

A simulation is only legitimate on domains the client owns and mailboxes the client named. Remediation and hardening are a separate, commissioned engagement — the audit shows the exposure, not the cure.