Why social engineering cover usually stops at your own bank account
An attacker gets into a mailbox at your company and reads your billing threads for a few weeks. Then he emails a client from your real address: our bank details have changed. The client pays the next invoice into the attacker's account. Nobody notices until your receivables team chases the overdue payment. The client says it already paid, and its lawyers want to know what you plan to do about it.
Most buyers assume the social engineering fraud endorsement on their crime or cyber policy covers this. Usually it doesn't. That's why client impersonation fraud coverage now comes up as its own item in broker conversations.
The standard social engineering insuring agreement covers loss of your money or securities. Typically that means funds your employee, acting in good faith, transfers on a fraudulent instruction that appears to come from a vendor, client or senior colleague. The trigger is your employee being tricked into moving your money. In the scenario above, nobody at your company moved anything. Your client's accounts payable clerk was deceived, and the money left your client's account.
Read the operative clause and you'll usually find three things that block a claim:
- Ownership: covered property is money owned by the insured, or held by the insured for others. Your client's operating funds are neither.
- Who was deceived: the wording requires that an employee of the insured was deceived. Here it was a client employee.
- Who transferred the funds: the transfer must be made by or on behalf of the insured. Here the client made it.
This isn't a drafting flaw. The cover was designed for a different risk. The one you actually face, sometimes called reverse social engineering or invoice manipulation, has to be picked up elsewhere in your program, if at all. Wordings differ by carrier and country, so treat the above as the typical pattern, not a guarantee.
Looking to streamline vendor RFP selection?
Access pre-negotiated tier discounts through our independent institutional network.
The two losses a client impersonation fraud actually creates
Buyers often treat this as one loss. It's two, and different policy sections respond to each.
Loss one is your unpaid receivable. The client paid the criminal and won't pay you again. You've delivered the goods or services, and the invoice sits uncollected. That's a first-party loss.
Loss two is the client's claim against you. If the client pays you a second time, it may then try to recover what it lost. Its argument will be that weak email security let the attacker in, and that the fraudulent instruction came from your genuine domain. That's a third-party liability claim, with defense costs on top of any settlement.
A hypothetical example shows how this plays out. Say a 40-person engineering consultancy has a client that pays a USD 180,000 milestone invoice to an attacker after a compromised project manager's mailbox sent out changed bank details. Three outcomes are realistic:
- The client refuses to pay again. The consultancy writes off USD 180,000 or sues its own client, which is commercially awkward.
- The client pays again to protect the relationship, then brings a negligence or breach-of-contract claim for the same amount plus costs.
- The two sides negotiate a split, often close to even. Each then looks to its own insurers for its share.
Under a typical cyber policy, outcome one might fall under an invoice manipulation extension. Outcome two would go to network security liability or professional indemnity, subject to exclusions. Outcome three could touch both. Incident response costs (forensics to confirm how the mailbox was compromised, plus legal advice on notification) usually sit under the cyber policy's breach response section either way. That's often the one part that pays without a fight.
Which policy sections may respond, and what blocks them
No single standard section was built for this loss. Cover comes from a patchwork, and each piece has its own trigger and common exclusion. Use the table to map your current program before you talk to your broker.
| Policy section | Which loss it targets | Typical trigger | Common blocker |
|---|---|---|---|
| Crime: social engineering fraud endorsement | Your own funds | Your employee transfers your money on a fraudulent instruction | Client's funds aren't owned or held by you; client employee was deceived |
| Crime: client property or third-party extension | Client's funds | Theft of client property by your employees | Usually requires employee dishonesty, not an outside hacker |
| Cyber: invoice manipulation or reverse social engineering cover | Your unpaid receivable | Breach of your systems leads a client to pay a fraudulent invoice | Low sublimits; often limited to net cost or margin; may require proof the client won't pay |
| Cyber: network security liability | Client's claim against you | Claim alleging your network security failure caused harm | Exclusions for loss or transfer of funds, monetary value of transactions, or contractual liability |
| Professional indemnity / E&O | Client's claim against you | Negligent act in rendering professional services | Cyber exclusions; arguments that email security isn't a professional service; funds exclusions |
Four points the table can't hold.
Invoice manipulation cover is the most direct fix for loss one, but it's narrow. Some cyber carriers, including Coalition, offer an insuring agreement that reimburses an insured that can't collect a receivable because a breach of its systems led a client to pay a fraudster. Sublimits are commonly in the low-to-mid six figures in USD for small and mid-size buyers. Some forms pay only your direct cost of the goods or services, not invoice value including margin. Check that the trigger is a breach of your systems. If the client's own mailbox was the one compromised, this cover typically won't apply.
Network security liability is the strongest route for loss two, and the funds exclusion is where it fails. Many cyber liability sections exclude the monetary value of electronic fund transfers or transactions that are lost or diminished. Carriers wrote that clause to keep trading and transfer losses out of liability cover, and a client's misdirected payment fits it neatly. Some wordings carve it back for third-party claims arising from a security failure. Many don't.
PI responds best where handling client money is the professional service. Law firms, conveyancers, accountants, escrow agents and property managers have a credible argument that a misdirected payment instruction arose from their services. A software developer or manufacturer usually doesn't, because sending an invoice isn't the service the policy insures.
Crime client-property extensions rarely help. They are built for dishonest employees, not outside criminals, so don't count on them here.
How wordings differ across markets
Market practice varies a lot. A program that works in one country may leave gaps in another, so multinationals and companies billing clients abroad should check every local placement.
North America. Crime forms built on standard industry templates dominate. Social engineering cover is usually an endorsement with a sublimit, often well below the main crime limit. Invoice manipulation cover is most widely available from specialist cyber carriers and managing general agents, sometimes bundled, sometimes optional.
London market and the UK. Lloyd's required policies to state clearly whether they affirm or exclude cyber cover, so many PI and property wordings now carry explicit cyber clauses. That improved clarity but can reduce cover: a PI policy that once responded silently to a misdirected-payment claim may now exclude it. Regulated professions such as solicitors buy PI under regulator-set minimum terms, which affects how claims from defrauded clients are handled. Confirm with your broker how top-up layers and any separate cyber policy interact with those terms. The UK's authorized push payment reimbursement rules bind payment providers and protect only certain smaller payers. They won't usually help a mid-size corporate client, and they don't change your liability exposure.
Continental Europe. In Germany, Austria and Switzerland, trust damage insurance (Vertrauensschadenversicherung) often covers losses from outside fraudsters, including fake-president and payment diversion schemes. Some forms extend to third-party damage, but only where the wording says so expressly. Elsewhere in the EU, cyber wordings vary by insurer, and data-protection exposure from the mailbox breach is usually better covered than the funds loss itself.
Middle East, India, Southeast Asia and Africa. Many commercial cyber and crime policies here follow London wordings, sometimes through local fronting insurers. Invoice manipulation cover exists but is offered less consistently. Some regional SME cyber products advertise phishing cover that turns out, on reading, to be first-party only. In Latin America, crime and cyber are often placed separately, and social engineering cover can bring extra underwriting questions on payment controls.
Pricing follows the same pattern. As a typical 2026 market range, adding social engineering or invoice manipulation cover with a sublimit in the low six figures might cost from a few hundred to a few thousand USD a year for a small or mid-size company. Higher sublimits, regulated professions and weak payment controls push that up sharply. Local pricing differs widely, so confirm with your broker and the carriers they approach.
Questions to put to your broker before renewal
Don't ask your broker whether you're covered for social engineering. The answer will be yes, and it won't address this risk. Ask about the specific scenario, in writing, and request answers that cite clause numbers.
- If a client pays a fraudulent invoice sent from our compromised email, which section covers our uncollected receivable, and at what sublimit? Ask whether it pays full invoice value or only your cost.
- If that client sues us, which liability section responds? Have the funds-transfer and monetary-value exclusions quoted back to you, along with any carve-backs.
- Does our PI policy carry a cyber exclusion, and does it remove claims arising from a mailbox compromise? This matters most for professional firms that handle client money.
- What happens if both the cyber and PI policies could respond? Overlapping other-insurance clauses can leave two insurers pointing at each other while defense costs build.
- Are there conditions precedent on controls? Some extensions require multi-factor authentication on email or documented callback procedures. Breaching a condition can void the claim, not just reduce it.
Red flags: a social engineering sublimit on the schedule with no definition of whose funds are covered; a cyber policy that advertises funds transfer fraud cover but defines it only as fraudulent instructions to your bank; and a PI renewal that quietly adds a broad cyber exclusion with no new cyber cover to replace it.
Controls cut both claim frequency and premium. Put MFA on every mailbox and enforce DMARC on your domain. Add a standing notice to invoices saying you'll never change bank details by email, and that clients should verify any change by phone using a number they already hold. That notice also helps your defense if a client later alleges negligence.
The decision rule: if a single invoice can exceed your social engineering sublimit, or you handle client money as part of your services, get written confirmation of cover for both losses, the receivable and the liability claim. If your broker can't confirm both, price an invoice manipulation extension and a carve-back to the funds exclusion before renewal. Have a licensed broker or coverage counsel review the final wordings, since only the policy language decides the outcome.