Auditvault Pro.
Security Controls

Eight Business Email Compromise Mistakes That Leave Your Payments Exposed

Most BEC losses come from controls that were never finished, not from clever attackers. Here are eight gaps, the fix for each, and how each one shows up on your cyber insurance application.

Auditvault Pro editorial team · November 2026 · 8 min read

Why BEC Losses Usually Come From Unfinished Controls

If your finance team can release a six-figure wire because of one email, you're going to be targeted. What you don't know yet is which control will hold when it happens. Business email compromise protection rarely fails because an attacker was brilliant. It fails because a control was half-built: a DMARC record that only monitors, a callback rule nobody wrote down, an approval step a senior executive can skip.

Those gaps cost you at renewal too. Cyber underwriters in most major markets now ask pointed questions about email authentication, payment verification and MFA. Weak answers can mean a higher premium, a smaller social engineering sublimit, or an exclusion you find out about only after a loss.

Advertisement

The eight mistakes below are the ones that most often undo BEC controls. Each one comes with its fix and what it means for your insurance. You can run the list as a gap assessment in a single meeting with your CFO, your IT lead and your broker.

Mistakes One and Two: Email Authentication That Only Watches

Mistake 1: DMARC left at p=none indefinitely

A DMARC policy of p=none collects reports and nothing more. It doesn't stop anyone from sending mail that claims to come from your exact domain. Plenty of companies publish p=none at the start of a project, find a few unknown senders (a marketing platform, an HR system, an invoicing tool) and stall. Two years later, attackers can still spoof the CFO's address to your suppliers.

What to do instead: Run DMARC enforcement as a project with a named owner and a deadline. A realistic path for a mid-size company:

  1. Weeks 1 to 4: publish p=none with reporting and inventory every legitimate sending service.
  2. Weeks 4 to 8: fix SPF and DKIM alignment for each sender and retire services nobody owns.
  3. Weeks 8 to 12: move to p=quarantine. If you're nervous, use the percentage tag to apply it to only part of your mail at first.
  4. Months 4 to 6: reach p=reject on the primary domain and on parked domains that never send mail.

If you have more than a handful of sending services, pay for a managed DMARC service. Most projects stall because someone has to read the raw aggregate reports by hand. Vendors such as dmarcian, Valimail, Red Sift OnDMARC and EasyDMARC typically charge from around $1,000 a year up to the low tens of thousands for small and mid-size organizations. The price depends on domain count, mail volume and features, and enterprise contracts run higher. These are typical 2026 market ranges. They vary by contract and region, so confirm with vendors. Major mailbox providers now require authentication from bulk senders, so this work protects your deliverability as well.

Insurance impact: Many applications ask directly whether DMARC is enforced. If you answer "yes" while you're still at p=none, you risk a misrepresentation finding. If you're mid-rollout, say so and give a target date for enforcement.

Mistake 2: Assuming DMARC stops lookalike domains

DMARC protects your domain and nothing else. It does nothing about auditvau1tpro.com or a supplier's domain with one letter swapped. A large share of sophisticated BEC arrives from exactly those lookalikes, or from genuine supplier mailboxes that have been compromised. Your DMARC record doesn't touch either one.

What to do instead: Turn on external-sender banners and first-time-sender warnings in Microsoft 365 or Google Workspace. Add lookalike-domain monitoring, which many email security gateways and DMARC vendors bundle in. Then give finance one rule: if a payment email carries a warning banner, verify by phone before acting.

Mistakes Three to Five: Payment Verification That Trusts the Attacker

Mistake 3: Calling back the number in the suspicious email

A callback works only if you dial a number the attacker can't control. If the number came from the new invoice, the email signature or an "updated contact sheet," you're calling the fraudster. This is the most common way callback procedures fail. It's also the most dangerous, because the person who made the call believes they followed the rules.

What to do instead: Put the callback procedure in writing. Numbers can come only from the vendor master file as it stood before the change request, from a signed contract, or from a contact you've already verified. Record who called, which number they used, who answered and when. If the only number available came from the request itself, the payment waits.

Mistake 4: No out-of-band check on bank-detail changes

Fraudsters don't need to fake an invoice if they can change where a real one gets paid. Bank-detail changes need their own controlled workflow. A clerk editing a field in the ERP doesn't count as one.

What to do instead:

  • Accept change requests only through a supplier portal or on a signed form. Never act on a plain email instruction.
  • Verify through a second channel: a callback to a known contact and, where you can, confirmation from a second person at the vendor.
  • Use bank-side payee-name checks where they exist. Examples include Confirmation of Payee in the UK, Verification of Payee for euro transfers in the EU, account-validation ("penny drop") services in India, and similar services from banks in other markets.
  • Hold the first payment to new details for a few business days and route it to a second approver. The short delay costs far less than one redirected payment.

Mistake 5: Letting executives bypass the process

"The CEO needs this paid today, confidentially" is the oldest BEC script there is. It works because the control has an unwritten exception for seniority.

What to do instead: Have the board or the CFO approve a written policy that exempts no one from dual approval or callback, including the CEO. Then announce it internally. Once staff know the CEO will never ask them to skip a step, any urgent request to skip one is itself a red flag.

Mistakes Six and Seven: Compromised Mailboxes and Slow Recall

Mistake 6: Weak MFA and no mailbox-rule monitoring

The most damaging BEC attacks run from inside a real mailbox, either yours or a supplier's. Once inside, attackers set hidden forwarding or deletion rules and read invoice threads for weeks. SMS codes and push approvals can be beaten by adversary-in-the-middle phishing kits and by MFA fatigue attacks.

What to do instead: Move finance, executives and IT admins to phishing-resistant MFA (FIDO2 security keys or passkeys), backed by conditional access policies. Start with those roles, because they hold the payment authority and admin rights attackers are after. Set alerts for new inbox rules that forward mail externally, or that move or delete messages containing words like "invoice" or "payment." Microsoft 365 and Google Workspace both offer alerting on suspicious mailbox rules, with richer detection at higher license tiers, and most MDR providers cover it too.

Mistake 7: No rehearsed fraud-recall plan

Recovery depends on speed. The sooner your bank requests a recall, the better your chances. Once funds have passed through mule accounts, they're usually gone. Recall tools differ by payment rail and country. Examples include SWIFT recall requests for cross-border payments, Pix's special return mechanism in Brazil, and domestic fraud-reporting channels in many other markets.

What to do instead: Keep a one-page playbook in treasury. It should list:

  • the bank's fraud desk number, not your relationship manager's
  • the details the bank needs to request a recall
  • the local police or cybercrime reporting channel
  • your insurer's breach hotline

Rehearse the playbook once a year. Notify the insurer early, because many policies require prompt notice and the use of pre-approved response vendors.

Mistake Eight: Assuming Your Cyber Policy Pays for BEC

Many buyers learn after a loss that sending funds to a fraudster voluntarily often isn't covered under the core cyber insuring agreements. Cover may sit in a social engineering or funds-transfer fraud extension with a sublimit far below the main limit. For smaller companies that sublimit commonly falls somewhere between the tens of thousands and the low hundreds of thousands of dollars, even when the main limit runs to several million. Treat that as an indicative range only, because it varies by insurer, market and loss history. Some wordings make a documented callback a condition of cover. Others push the loss to a crime or fidelity policy. Ask a licensed broker to map exactly where a BEC loss would land under your program.

Use this table to set priorities before renewal:

GapFixTypical effortInsurability effect
DMARC at p=noneStaged move to p=reject3 to 6 monthsLets you answer a common application question cleanly
Callbacks to supplied numbersWritten procedure using master-file numbers2 to 4 weeksMay be a condition of social engineering cover
Unverified bank changesOut-of-band check, payee-name check, holding period1 to 2 monthsStrengthens the case for a higher sublimit
Executive bypassBoard-approved no-exception policyDaysRemoves a common ground for claim disputes
Weak MFAPhishing-resistant MFA for high-risk roles1 to 3 monthsOften a minimum underwriting requirement
No recall planTreasury playbook, annual drill2 weeksImproves recovery odds and lowers net loss

Hypothetical example: Say a 60-person distributor in Dubai carries a $2 million cyber limit with a $100,000 social engineering sublimit. A spoofed supplier email redirects a $180,000 payment. Even if the insurer accepts the claim, $80,000 is uninsured. If finance can't produce a callback record, the whole amount may be disputed. Fixing mistakes 3 and 4 costs a fraction of either outcome.

Next step: Score yourself on all eight items before your next renewal. Close the two cheapest gaps, the written callback procedure and the no-exception policy, within 30 days. Then take the evidence to your broker and ask what it would take to raise your social engineering sublimit.

Questions buyers ask

How long does it take to move DMARC from p=none to p=reject?

Most mid-size organizations need three to six months. Most of that time goes into finding every legitimate service that sends mail as your domain and fixing its SPF and DKIM alignment. Companies with many business units or legacy systems often take longer. A managed DMARC service usually shortens the timeline because it does the report analysis for you.

Does cyber insurance cover money lost to business email compromise?

Sometimes, but often only under a social engineering or funds-transfer fraud extension with a sublimit well below the main policy limit. Some policies require a documented verification step, such as a callback, before they pay. Others send these losses to a separate crime policy. Wording varies by insurer and jurisdiction, so have a licensed broker confirm how your policy would respond.

What counts as a valid callback for a vendor bank-detail change?

A valid callback is a phone call to a number you held before the request arrived, such as one from the vendor master file, a signed contract or a previously verified contact. Never use a number from the change request, the email signature or an attached invoice. Record who called, the number used, who confirmed the change and the time, and keep that record in the payment file.

Written by the Auditvault Pro editorial team. Prices, features and rules change often - confirm details with the vendor or a licensed professional before you decide.