Business email compromise cost US organizations $3.04 billion in reported losses in 2025 — the second-largest loss category the FBI tracks. When you discover a compromised mailbox, you're usually discovering it late: the attacker has typically been reading for weeks, and the fraudulent payment is how they cashed out. The response has two clocks running at once — the money clock and the access clock.
The money clock (hours)
If any payment moved or any payment details changed under the compromised mailbox's influence: run the wire fraud first-48-hours sequence now — bank recall, IC3 filing, insurer notice — and freeze every pending payment associated with threads the mailbox touched. Then audit 90 days of payment-detail changes and re-verify each by callback to numbers on file. Assume every instruction that passed through the mailbox is suspect until verified.
The access clock (same day)
- Preserve first: export the mailbox, sign-in logs, and rule configurations before touching anything — evidence dies in cleanup.
- Then evict: reset credentials, revoke all active sessions and app passwords, and re-enroll MFA. A password reset alone doesn't end active sessions.
- Hunt the persistence: attackers add hidden inbox rules (auto-forwarding to external addresses, auto-deleting security alerts and replies), mail delegates, and OAuth app grants. Remove all of them. The forwarding rule is how they'll watch your response.
- Scope sideways: check whether the account was used to phish colleagues or counterparties — BEC spreads through reply-chains, and your compromised thread is now someone else's incoming fraud.
The counterparty problem
If your mailbox was used to defraud a customer or vendor — fake invoices from your real domain — you have a notification decision with legal and commercial weight. Involve counsel early, but bias toward fast, factual notification of affected counterparties: they can still stop payments you can't, and silence ages terribly in later disputes about who could have prevented what.
The week-one review
Two questions determine whether this repeats. Technical: how did they get in, and is MFA enforced everywhere including legacy protocols? Procedural: why could an email — any email, from anyone — move money or change payment details without out-of-band verification? The first question is IT's. The second is the payment approval process, and it's the one that decides whether the next compromised mailbox costs you anything at all.