The Complete Guide to Business Email Compromise
How BEC works, the four variants that cause most losses, why it slips past filters and careful staff, and the payment controls that actually stop it.

A payment leaves accounts payable on a Thursday afternoon. The invoice was real, the amount matched the purchase order, the supplier was one the company had used for years, and the person who approved it had done the job for a decade. The only thing wrong was the account number, changed three weeks earlier by an email that looked exactly like all the others in the thread.
That is business email compromise. There is no malware to find, no exploit to patch and usually no alert to investigate. The attack is a conversation, and the loss is an ordinary payment made by an employee doing their job properly. It is one of the reasons BEC consistently appears among the costliest categories in the FBI's annual Internet Crime Report, well ahead of attacks that get far more attention.
What business email compromise actually is
BEC is fraud that uses email to get an organization to send money or data to the attacker, by impersonating someone the organization already trusts. The impersonated party is usually an executive, a supplier, an employee or a professional adviser. The request is usually a payment, a change to payment details, or a file of personal data.
Two things separate it from ordinary phishing. The first is that it is targeted: the attacker has done enough homework to know who approves payments, who the suppliers are and roughly what a normal request looks like. The second is that the money moves through legitimate channels. Nobody is tricked into installing anything. A real employee logs into the real banking portal and makes a real payment, which is why the transaction clears every automated control the business has.
The attacker's work is therefore all in the setup. Everything they do before the request is aimed at making one email unremarkable.
The four stages of a BEC attack
1. Research
Attackers build a picture of the organization from public sources: the website, job adverts, press releases, filings and employee profiles on professional networks. They learn the finance structure, the names of senior staff, the format of internal email addresses, which suppliers are named in case studies and when the company is busy. A merger, a funding round or a seasonal peak all create plausible reasons for an unusual payment.
2. Getting into position
There are two routes, and they call for different defenses.
- Impersonation from outside. The attacker registers a domain that looks like the target's or the supplier's, swapping a letter, adding a hyphen or changing the top-level domain. Alternatively they use a free email account with the right display name, which is enough on a phone where only the name shows.
- Account takeover. The attacker gets into a real mailbox, usually through credential phishing or a reused password, and then uses it as a trusted position. MITRE ATT&CK catalogs this as the use of valid accounts, and it is the version that causes the largest losses, because every signal a recipient might check is genuine.
3. Observation and preparation
An attacker inside a mailbox rarely acts immediately. They read. They search for terms like invoice, remittance, bank and wire to find live payment conversations, a behavior ATT&CK describes as email collection. They often create a mail rule that quietly moves messages containing those words into an obscure folder, so the real account holder never sees the thread the attacker is now handling. Some set up auto-forwarding to an external address so they keep visibility even after a password change.
4. The request and the cash-out
When a genuine payment is due, the attacker makes the ask. If they hold the mailbox, they reply inside the existing thread, sometimes after deleting the real messages. If they are impersonating from outside, they start a new thread or reply to a copied one. The receiving account is typically a mule account at a legitimate bank in the same country as the victim, which avoids the friction of an international transfer. Funds are withdrawn or moved on within hours, which is why response time decides whether anything is recovered.
The main variants
CEO fraud and executive impersonation
A message appearing to come from a senior executive asks a finance employee to make an urgent payment outside the normal process, with a reason not to discuss it. The executive is usually travelling, in a meeting or otherwise uncontactable. The amount is often chosen to sit just below a threshold that would trigger extra approval. The real mechanism is authority, not technology, which is covered in more detail in our article on whaling and executive impersonation.
Vendor and invoice fraud
This is the variant that produces the largest single losses, because the amounts are already large and legitimate. The attacker either compromises the supplier's mailbox or imitates it, then asks for future invoices to be paid to a new account. Sometimes they intercept a real invoice, change the bank details and forward it on. Because the invoice, the contract and the goods are all genuine, the fraud is often discovered weeks later when the supplier chases payment. We cover the mechanics in vendor email compromise and the specific case of an invoice with changed bank details.
Payroll diversion
An email that appears to come from an employee asks HR or payroll to update their bank details before the next pay run. The amounts are small compared with invoice fraud, so the request attracts less scrutiny, and the fraud is usually only discovered when the employee reports a missing salary. Attackers often time these for the few days before payroll closes. Our guide to payroll diversion fraud goes through the controls in detail.
Attorney and adviser impersonation
Someone claiming to be a lawyer, auditor or corporate adviser contacts a finance employee about a confidential matter, often an acquisition or a settlement, that must not be discussed internally. The secrecy is the attack: it is a built-in reason to bypass every normal check. These requests often arrive late in the day or just before a holiday.
Data-theft BEC
Not every BEC asks for money. A request appearing to come from an executive or HR may ask for a list of employee payroll records or tax forms. The data is then used for identity fraud, tax fraud or to make the next round of impersonation more convincing. Treat a bulk request for personal data with the same suspicion as a payment request.
Why BEC defeats normal checks
Most security advice about email assumes there is something to spot. BEC is built so there is not.
- There is nothing malicious in the message. No attachment, no link, no code. Secure email gateways are largely looking for things these emails do not contain.
- The writing is good. The old advice about spelling mistakes and odd grammar has not been reliable for years. Attackers copy tone and signature blocks from real messages in the thread.
- The sender can be genuine. When the mail comes from a compromised supplier mailbox, SPF, DKIM and DMARC all pass, because the mail really is from that domain. Authentication proves the domain, not the intent.
- Mobile email hides the evidence. Phone clients show the display name and often truncate the address, so a lookalike domain is invisible at a glance.
- The context is real. Thread hijacking means the attacker inherits all the trust the previous twenty messages built. Our article on reply chain phishing explains why this is so effective.
- The pressure is structural. Urgency, authority and confidentiality are not tells that something is fake. They are the reasons a reasonable employee skips a step, and an attacker uses all three at once.
Attackers have also started adding a second channel to close the gap. An email asking for an urgent transfer is followed by a phone call or a voice note that sounds like the executive, produced from recordings of them speaking publicly. The call is not there to convey information. It is there to answer the one objection a careful employee has, which is that the request only exists in writing. Our article on deepfake voice call fraud explains why an unexpected inbound call confirming an unexpected email verifies nothing at all.
The practical consequence is important: you cannot train people to detect BEC by inspection. What works is a process that does not depend on the message being judged correctly.
Detection: what to actually monitor
Because the fraud is a legitimate payment, detection has to happen either in the mailbox or in the payment process.
In the mailbox
- New inbox rules and auto-forwarding. Alert on the creation of any rule that forwards, redirects or moves mail to an external address or an unusual folder. This is one of the highest-value detections available, because almost every mailbox takeover involves one.
- Sign-ins that do not fit. Logins from unexpected countries, two sign-ins from distant locations within a short window, sign-ins from hosting providers or anonymizing services, and successful authentication after a run of failures.
- Legacy authentication attempts. Older mail protocols that bypass modern sign-in checks are a favored route. Block them and alert on attempts.
- New OAuth app consents. An attacker who cannot beat MFA may instead persuade a user to grant a malicious application access to their mail. See consent phishing for how that works.
- Mass deletions or searches. An attacker tidying a thread leaves traces in audit logs.
In the payment process
- Any change to supplier or employee bank details should generate a record, a verification step and a notification to someone other than the person who made the change.
- First payments to a new payee deserve more scrutiny than routine ones, especially if they are large or urgent.
- Invoices with new remittance details but otherwise identical formatting to previous ones.
- Requests to change a payment after approval but before release.
The payment controls that actually stop it
These are the controls worth arguing for. They work whether or not anyone spots the email.
Callback verification on a number you already hold
Before any new payee is set up or any existing bank details are changed, call the supplier or employee on a phone number from your own records, not one in the email, the invoice or the signature block. Speak to a named person you can identify. The IC3 makes the same point in its guidance: use a secondary channel to verify changes to account information. An attacker can control the email and the documents. They cannot answer a number you already had.
Dual authorization for new and changed payees
Setting up a payee and releasing a payment should require two different people, and the second person's job is to confirm the verification happened, not simply to click approve. Set the threshold low enough to catch the payments that would genuinely hurt, rather than at a level only the largest transfers reach.
A short delay on first payments to new accounts
A twenty-four hour hold before the first payment to a new beneficiary costs almost nothing and removes the urgency the attacker depends on. Many frauds collapse during that window because the real supplier gets in touch about something else.
Use the bank's own checks
Ask your bank what account-name verification, beneficiary confirmation or positive-pay style services it offers, and turn them on. Where a payee name can be checked against the account name before a transfer, mismatches become visible before the money moves.
Make escalation socially safe
Write down, and have leadership say out loud, that no payment instruction will ever arrive by email alone, that nobody will be criticized for checking, and that an employee who delays a payment to verify it has done the right thing. This costs nothing and disables the lever most BEC attempts rely on. The NCSC's board toolkit frames this as a governance responsibility rather than a training issue, and that is the right place for it.
Close the supplier onboarding gap
Most verification failures trace back to a missing phone number. When a supplier is first set up, the only contact details on file are often the ones that arrived in the supplier's own email, which means a verification call later goes straight to whoever sent it. Capture an independently sourced phone number at onboarding, from the supplier's published website or the contract, record who verified it and when, and treat the field as mandatory rather than optional. Review the existing supplier master file for records that have no verified number, because those are the accounts a fraudster will find easiest to change.
Protect the process in writing
One page is enough: who can approve what, how changes to bank details are verified, what happens when a request is urgent, and who to call when something feels wrong. Our one-page security policy template covers the payment section.
Technical controls that reduce your exposure
Technical measures will not stop a determined impersonation, but they remove the easy routes and make attacks noisier.
- Phishing-resistant multi-factor authentication on every mailbox, starting with finance, payroll, procurement and administrators. Hardware keys or passkeys resist the real-time proxy attacks that defeat one-time codes. CISA's MFA guidance is a reasonable baseline to point at internally.
- SPF, DKIM and DMARC with an enforcing policy on your own domains, so attackers cannot send mail that appears to come from you. Our email security guide explains how the three records work together and where they stop helping.
- External sender warnings and display-name impersonation detection, which catch the cheap lookalike attempts.
- Register obvious lookalike domains for your own brand, and monitor new registrations that resemble it.
- Restrict who can create forwarding rules to external addresses, and review existing ones.
- Conditional access that limits mailbox access to managed devices or known locations where that is workable.
If money has already gone
The first few hours decide the outcome. Work in parallel rather than in sequence.
- Call the sending bank immediately and ask them to attempt a recall and to request a hold harmless letter or letter of indemnity from the receiving bank. Use the fraud line, not general customer service, and get a reference number.
- Report it to your national cybercrime or fraud body. In the United States that is the IC3, which asks for full banking details in the complaint because they feed the recovery process. Elsewhere, report to the national police cybercrime unit or fraud reporting service.
- Secure the mailboxes involved. Reset passwords, revoke active sessions and tokens, remove forwarding rules and malicious app consents, and check the audit log for what was read. Our guide to responding to business payment fraud sets out the sequence.
- Tell the real counterparty by phone. If a supplier's mailbox was compromised, they have a problem too, and other customers may be targeted.
- Preserve evidence before anyone tidies up: full message headers, the original emails, audit logs, the invoice files and a timeline of who did what.
- Notify your insurer within the policy window, even if cover is uncertain.
- Assess data exposure. If a mailbox was accessible, assume its contents were read, and consider whether that triggers a notification obligation. Our article on data breach notification covers the questions to ask.
Not sure about a payment email? Paste the message into CyberWatch AI Scan for a free check, then verify the request by calling a number you already hold.
Testing whether your controls work
Policies fail quietly. The only way to know yours holds is to test it against the scenarios that matter.
- Tabletop the supplier bank change. Walk a real accounts payable employee through a changed-details request and see exactly which number they would call and where they would find it. Most organizations discover their only contact number came from the supplier's email signature.
- Measure verification, not awareness. Track what proportion of new payees in the last quarter had a documented callback. That number tells you more than any quiz score.
- Run realistic simulations for the money-handling teams. A generic password-reset test teaches finance nothing. A simulated supplier bank change, sent to the people who would receive a real one, shows whether the process survives pressure. Our guides to phishing simulation and training staff on payment fraud cover doing this without eroding trust.
- Review after every near miss. An attempt that was caught is free intelligence about what the attacker knew.
Who owns this, and a first month of work
BEC falls between departments, which is how it survives. Finance owns the payment process, IT owns the mailboxes, HR owns payroll records and nobody owns the attack. Name one person accountable for the end-to-end path from a request arriving to money leaving, and give them the authority to change the process rather than only to report on it.
If you are starting from nothing, the following order gets the largest reduction in risk for the least effort:
- Week one. Turn on multi-factor authentication for every mailbox, prioritizing finance, payroll, procurement and administrators. Alert on the creation of external forwarding rules. Block legacy authentication protocols.
- Week two. Write the payment verification rule in one paragraph, get leadership to endorse it in their own words, and circulate it. Audit the supplier master file for missing verified phone numbers.
- Week three. Introduce dual authorization for new and changed payees, plus a short hold on first payments to new accounts. Agree the escalation path for a request that breaks the rules.
- Week four. Tabletop a supplier bank change with the people who would actually receive it, then fix whatever the exercise exposes. Confirm with your insurer and your bank what cover and what recall support you have.
None of these steps requires new software, and together they address the two things every BEC attack needs: a mailbox it can reach and a payment nobody checks by voice.
Where CyberWatch AI fits
CyberWatch AI lets an organization send its own staff realistic practice emails, including executive payment requests and supplier bank-detail changes, and see who verifies and who pays. The people who need it get short, specific training on that pattern instead of a generic annual module, and anyone unsure about a real message can check it and report it to the security team in one step.
The short version
BEC is not a technology problem with a technology fix. It is a process problem wearing an email. Assume that at some point your team will receive a message that is indistinguishable from a genuine request, and build a payment process that stops it anyway: verification by phone on a number you already hold, two people for new payees, a short delay on first payments, and a culture where checking is the expected behavior rather than an insult.
For the wider program this sits inside, see our guides to business cybersecurity, phishing and how to defend against it and vendor impersonation fraud.
Frequently asked questions
What is business email compromise?
Business email compromise is a fraud in which an attacker uses email, usually impersonating a trusted colleague or supplier, to get an organization to send money or sensitive data to the wrong place. It relies on social engineering and legitimate-looking messages rather than malware.
How is BEC different from phishing?
Phishing usually tries to harvest credentials or deliver malware at scale. BEC is targeted and often malware-free: the goal is a single authorized payment or data transfer made by a real employee following what looks like a real instruction.
Does multi-factor authentication stop BEC?
It stops a large share of mailbox takeovers, which removes the attacker's best position. It does not stop impersonation from a lookalike domain or a compromised supplier, so payment verification controls are still needed.
Can money sent in a BEC payment be recovered?
Sometimes, if you act within hours. Contact the sending bank immediately to request a recall and a hold harmless letter, and report the fraud to your national cybercrime or fraud body. Recovery becomes unlikely once funds are withdrawn or moved abroad.
Which teams do attackers target?
Anyone who can move money or change payment records: accounts payable, payroll, procurement, finance leadership and executive assistants. Attackers also target IT administrators to reach those mailboxes.
Why do filters not catch these emails?
A well-written BEC email contains no attachment, no link and no malicious code. When it is sent from a compromised real mailbox inside an existing thread, there is nothing technically wrong with it to detect.
What single control gives the best protection?
An out-of-band callback to a phone number you already hold on file before any new or changed payee is paid, with no exception for urgency or seniority. Dual authorization on new payees is a close second.
Should staff be allowed to question an executive request?
Yes, and leaders should say so explicitly. Most BEC attempts depend on a junior employee feeling unable to challenge a senior name, so an organization where checking is expected and praised removes the attacker's main lever.
Is BEC covered by cyber insurance?
It depends entirely on the policy. Many cyber policies exclude fraudulent funds transfer unless social engineering or crime cover is added, and insurers often require specific verification controls. Check the wording with your broker before you need it.
Sources
- Business Email Compromise (BEC), FBI Internet Crime Complaint Center
- Internet Crime Report 2024, FBI Internet Crime Complaint Center
- Industry Alerts, FBI Internet Crime Complaint Center
- Phishing attacks: defending your organisation, UK National Cyber Security Centre
- Avoiding Social Engineering and Phishing Attacks, CISA
- Phishing, Technique T1566 - Enterprise, MITRE ATT&CK
- Email Collection, Technique T1114 - Enterprise, MITRE ATT&CK
- Valid Accounts, Technique T1078 - Enterprise, MITRE ATT&CK
- Multi-Factor Authentication (MFA), CISA
- Cyber Security Toolkit for Boards, UK National Cyber Security Centre


