A Nine-Step Playbook for Verifying Vendor Bank-Detail Changes Before Money Moves
A fraudster needs only one unverified bank-detail change to redirect a quarter's payments. Here's a nine-step procedure that blocks it and gives your insurer the evidence it expects at claim time.
Treat every change request as unverified the moment it arrives
Vendor payment fraud doesn't succeed in the payment run. It succeeds days earlier, when someone in accounts payable updates a supplier's bank details because an email looked right. By the time money moves, the system treats the fraudster's account as the vendor's, and every downstream approval confirms the lie. That's why vendor bank account change verification needs its own procedure, separate from invoice approval.
Step 1: Freeze, log and route. Every request to change bank details, whether it arrives by email, portal, letter, phone or a vendor's sales rep, goes into a single intake queue. Nobody actions it from their inbox. The log should capture:
- Date and time received, channel, and the exact sender address or phone number
- The vendor's existing record ID and current bank details (masked)
- The requested new details, including bank country and currency
- Open invoices and scheduled payments to that vendor in the next 30 days
That last field matters because fraudsters time requests to land just before a large payment. A change that arrives with a big invoice due this week carries higher risk, and the log should flag it from the start.
Call back on a number you held before the request arrived
Step 2: The independent callback. This is the single control that stops most social engineering fraud, and it's the one most often done badly. A sound callback has three rules:
- Use a number from your own records, captured at onboarding or from a signed contract. Never use the email, the new invoice, the letterhead or a hurried web search. If you have no prior number, get one through a different channel, such as your buyer's known contact at the vendor, and record where it came from.
- Reach a known person, ideally someone in the vendor's finance team you've dealt with before. Ask them to read the new account details out to you. Don't read them out yourself.
- Record the call: who called, who answered, the number dialed, its source, the time, and what was confirmed.
A callback to the number in the request email is a call to the fraudster. And with voice cloning now cheap, "it sounded like Priya" isn't verification either. What counts is that you placed the call, to a number you trusted beforehand.
Match the account name with the bank where your market supports it
Step 3: Validate the account itself. A callback confirms intent. An account-name check confirms the destination belongs to the vendor. What's available depends on where the receiving account sits:
- UK: Confirmation of Payee checks the payee name against the account before payment.
- Eurozone: Verification of Payee is being phased in for euro credit transfers under EU instant payments rules. Check your bank for current status.
- India: "Penny-drop" validation can return the account holder's registered name, though banks and fintech providers differ in how they offer it.
- Australia: Confirmation of Payee has been rolling out across banks. Availability varies by institution.
- US and many other markets: there's no universal name-check scheme, so firms rely on bank account-validation services or third-party vendor-verification platforms.
For cross-border vendors, be wary when a domestic supplier suddenly asks to be paid into an account in another country. Treat that as a red flag even if the callback passes.
Dedicated validation platforms such as Trustpair and Eftsure automate these checks across countries. Typical 2026 market pricing for mid-market companies runs roughly USD 15,000 to 80,000 a year, depending on vendor volume, countries covered and ERP integration. Contracts and regions vary widely, so confirm with the vendors. For companies paying a few hundred suppliers across several countries, a platform usually beats manual checks on both cost and consistency.
Set dual approval thresholds by value and vendor risk
Step 4: Separate the person who verifies from the person who approves. Every bank-detail change needs a second approver. The question is how senior. Tie it to exposure, meaning the value of payments flowing to that vendor over the next 90 days, not the size of any single invoice.
| Exposure to vendor (next 90 days) | Verification required | Approver | Cooling-off period |
|---|---|---|---|
| Under USD 10,000 | Callback plus account-name check where available | AP supervisor | 2 business days |
| USD 10,000 to 100,000 | Callback by a second person plus written confirmation from the vendor's known contact | Finance manager or controller | 3 to 5 business days |
| Over USD 100,000, or any change of bank country | Two independent callbacks, account-name check, and procurement or relationship owner sign-off | Controller plus CFO or treasurer | 5 to 10 business days |
Treat these bands as a starting template. Convert them to local currency and scale them to your payment profile. A 40-person company might set its top band at USD 25,000. A multinational might start it at USD 500,000.
Lock vendor master data so no one person can change and pay
Step 5: Enforce it in the ERP, not just the policy. Written procedures fail under deadline pressure. System controls don't. Good vendor master data controls include:
- Segregated roles: users who can edit vendor bank fields can't create, approve or release payments. Audit this quarterly, because role creep is common.
- Sensitive-field locks: major ERPs such as SAP, Oracle, Microsoft Dynamics and NetSuite generally offer some way to treat bank fields as sensitive, so a change blocks the vendor for payment until a second user confirms it. Configuration and licensing differ, so check yours.
- Change reports: a daily or weekly report of every bank-detail change, reviewed by someone outside AP.
- No bulk edits to bank fields without a ticket and dual sign-off.
If your ERP can't enforce dual control natively, a payment-hub or validation tool sitting between the ERP and the bank can provide it. A spreadsheet sign-off sheet is not an acceptable permanent substitute.
Hold new details through a cooling-off period
Step 6: Delay the first payment on purpose. A cooling-off period means new bank details become payable only after a set number of business days. Use the windows in the table above. The delay catches fraud the callback missed, because real vendors often notice the attempt in that window. Their finance team asks why a change was confirmed that they never requested, or the legitimate contact replies to an email thread the fraudster failed to hijack.
Vendors will sometimes push back, especially if they're short on cash. Hold the line. A genuine supplier will accept a few days' delay on one payment cycle. One that won't, and keeps escalating urgency, is showing you a red flag. If a commercial exception is truly needed, it goes to the CFO in writing, with the reason recorded.
Send a controlled first payment and confirm receipt out of band
Step 7: Prove the new account works before the big money goes. For high-exposure vendors, either send a smaller first tranche, or pay one normal-sized invoice and wait for confirmation of receipt before releasing the next. Confirmation must come through the trusted channel used for the callback, not a reply to the email that requested the change.
Hypothetical example: say a 300-person manufacturer pays a logistics provider about USD 180,000 a month. A change request arrives on the 24th, with a USD 175,000 invoice due on the 30th. Under this playbook, the request is logged and flagged. Two people call back on the contract number, the account name is checked, and the controller and CFO approve. The details clear a five-day cooling-off period, and the first payment is released only once the provider's known finance contact confirms the account. That's slower than paying on the 30th. It's also how you avoid a USD 175,000 loss that your policy may cover only in part, if at all.
Escalate on these red flags and stop the payment cycle
Step 8: Know when to stop. Any one of the following should pause all payments to that vendor and escalate to the controller and security team:
- The sender's domain differs slightly from the vendor's known domain, or the reply-to address doesn't match the sender
- The new account is in a different country, or under a bank or account holder name that doesn't match the vendor's legal entity
- Pressure language: an audit, account freeze, CEO instruction, or "please don't call, I'm traveling"
- The request arrives just before a large payment, or asks for an overdue invoice to be paid to the new account
- The vendor contact refuses a callback or offers only a new phone number
- Several vendors request changes in the same week, which can point to a compromise of your own mailbox
If a fraudulent payment has already gone out, call your bank's fraud team immediately and ask for a recall. Then report to local law enforcement or your national cybercrime reporting channel, and notify your insurer. Recovery odds generally fall quickly as hours pass, so write this sequence down in advance, with phone numbers.
Build the evidence file crime and cyber insurers ask for
Step 9: Document as though you'll be filing a claim. Social engineering fraud cover usually sits in a crime policy or as an extension to a cyber policy. It's commonly sublimited well below the main limit, and many wordings make cover conditional on a verification procedure being followed, often a callback to a number obtained before the request. Some policies also treat voluntary transfers differently from hacking losses. Wordings differ by insurer and country, so have a broker or licensed adviser check yours against this playbook.
When a claim is filed, adjusters typically want:
- The original request, with full email headers
- The callback record: number dialed, its source and date captured, who answered
- Account-name check results
- ERP change logs showing who edited and who approved
- Evidence the cooling-off period applied, or the approved exception
- A timeline of discovery, bank recall request and police report
Store this evidence with each vendor's change ticket, keep it at least through the policy's discovery period, and test the process twice a year with a simulated change request. Your next step: pull last quarter's bank-detail changes from your ERP and count how many have a callback record from a pre-existing number. If the answer isn't all of them, fix that before your next renewal.
Quick answers
Does a callback still work now that voices can be cloned?
Yes, as long as you place the call to a number you held before the request arrived. Voice cloning can defeat inbound calls and calls to numbers supplied by the requester. It doesn't defeat you dialing a number from your contract file and reaching the vendor's real switchboard. For high-value changes, add written confirmation from a known contact.
How long should a cooling-off period be for new vendor bank details?
Most finance teams use 2 to 10 business days, scaled to how much money will flow to that vendor soon. Low-exposure vendors can clear in two days. A change of bank country or a large upcoming payment justifies a week or more. Exceptions should need written CFO approval.
Will our cyber or crime policy pay if an employee was tricked into changing bank details?
It might, but social engineering cover is often sublimited and frequently depends on documented verification, such as a callback to a pre-existing number. Wordings vary widely by insurer and country, and some treat voluntary transfers differently unless a specific extension is bought. Ask a licensed broker to map your procedure against the policy conditions before a loss happens.
Are vendor-validation platforms worth it for a mid-size company?
They tend to pay off when you have several hundred active suppliers, pay across multiple countries, or run an ERP that can't enforce dual control on bank fields. Typical 2026 market pricing is roughly USD 15,000 to 80,000 a year for mid-market contracts, varying by volume, countries and integration. Smaller firms with few vendors in one country can usually manage with a disciplined manual callback and their bank's name-check service.
Written by the Breach Underwrite Group editorial team. Prices, features and rules change often - confirm details with the vendor or a licensed professional before you decide.