What if your organization suddenly lost power? Or your ERP becomes infected with malware? As the old joke goes, “Nobody expects the Spanish Inquisition!” Stuff happens. Organizations and their operations can suddenly experience local or regional shocks relating to weather, geologic events or computer ransomware. We don’t like to think about it but large-scale events such as 9/11, the 2008 financial crisis or the 2020 Covid pandemic can bring serious operational challenge to everyone at once.
A principal effect in all kinds of unexpected events is inaccessibility to or the operation of computer systems. The world runs on computers. This is as true for financial operations as for production, sales and marketing. And a disruption to payables and receivable operations is deleterious to a business and though accounts receivable is outside the scope here, coordination of AP with AR for cash flow is vital.
Consequently, your control framework should include a disbursement continuity plan. When your ERP, payment platform or the power fails, your organization still must pay suppliers and employees. How it does that decides whether the outage becomes a fraud loss, a duplicate-payment mess or a supplier crisis.
Most continuity planning asks one question: How do we get systems back up? Disbursement continuity asks another important question: How do we keep paying the right party, the right amount, exactly once, while those systems are down? That is a controls question and it belongs to finance.
This article is written for CFOs, controllers and directors of financial operations. It covers the outage scenarios worth planning for, which controls can flex in degraded mode and which cannot, how to build and run the plan, and how to reconcile cleanly once systems return.
The Concentration Trade-off
Automation has been a boon to disbursement control. It has brought enforced approval workflows, system-based segregation of duties, automated three-way match, duplicate detection and a complete audit trail. It also moved the whole payment process onto a small number of systems, though the AP automation vendors that promote it rarely emphasize that.
That concentration, though, is what attackers exploit. In one survey of companies hit by ransomware, 88 percent said their ERP systems were affected and 62 percent saw at least 24 hours of downtime (the survey was run by an ERP security vendor). Recent incidents show what follows when the payment engine stops:
Jaguar Land Rover (2025). The September attack took down production at all] JLR's plants worldwide, along with its financial and supplier-invoicing systems. JLR settled invoices manually while automated payment systems were offline, then offered critical suppliers majority prepayment at order and reimbursed their financing costs. It reportedly had no cyber insurance and relied on a £1.5 billion government-backed loan.
CDK Global (2024). An attack on one dealer-software provider forced car dealerships onto manual processes, with administrative paperwork stacking up. Larger dealer groups severed their connections to the provider as part of incident response.
ChangeHealthcare (2024). The clearinghouse outage was estimated to be costing hospitals $2 billion a week in cash flow. One health system noted that its large cash reserves held off bigger problems. Liquidity is part of continuity.
The lesson is not to abandon automation. It is that every control that lives inside a system needs a defined fallback and the fallback must be designed, documented and tested before it is needed.
Three Scenarios, Three Plans
A single "systems down" procedure is not enough. What fails and what still works differ sharply by scenario, so the response differs too.
Scenario | What fails | What usually still works | Primary disbursement response | Biggest control risk |
Internal compromise(ransomware or intrusion in your ERP, AP platform, email or identity provider) | ERP and AP workflow; possibly email and single sign-on; data may be encrypted or altered | Bank portals (if credentials are independent of corporate SSO); phones; vendors and banks | Pay critical vendors from an offline vendor master and payment schedule through pre-approved bank channels | Attacker still inside; vendor master or bank details may have been tampered with; email cannot be trusted |
Third-party outage (payment processor, AP automation provider, payroll provider, bank platform) | The outsourced payment rail or workflow | Your own ERP, email and internal data | Switch to a secondary bank or alternate payment method; issue payments directly | Duplicate payments when the provider recovers and releases queued items |
Regional power or telecom outage | Everything local: offices, networks, phones; possibly your bank and vendors too | Staff and systems in other regions; Little works in the affected area | Defer non-critical payments; shift work to staff or banks outside the region; communicate expected delays | Pressure to make exceptions by phone; unverifiable requests once service returns |
The Iberian blackout of April 2025 shows the third scenario at full scale. Power was out across Spain and Portugal for about ten hours, and most businesses closed and banking stopped functioning. In that situation, the right plan is often an orderly pause plus clear communication, not heroic manual payments.
The BridgePay attack illustrates the second scenario: one processor's ransomware incident disrupted public services as well as merchant businesses across the U.S., even without card data being exposed.
Which Controls Flex, which Must Never Bend
Outage events are an opportune time for fraud. Approvals move to email and phone, staff are under pressure to get suppliers paid, and requests to change bank details sound plausible. Business email compromise alone produced about $3.05 billion in reported U.S. losses in 2025 and 74% of organizations experienced BEC that year. A degraded-mode plan that relaxes the wrong controls can invite exactly this kind of fraud.
The answer is to decide in advance, in writing, which controls may change form during a system outage, and which may not change at all. For example:
Control | Normal operation | Degraded mode | Can it flex? |
Vendor bank account changes | Verified per policy, system-logged | Frozen. No changes accepted during an outage except for life-safety or regulatory necessity, and then only with call-back to an out-of-band number already on file | Never |
Call-back verification | To a phone number from the vendor master, not the request | Same, using the offline vendor master copy | Never |
Payee and amount | Matched to approved invoice or PO | Matched to the offline payment schedule or a documented source record | Never |
Segregation of duties | Enforced by system roles | Enforced by named people on a pre-approved emergency roster; preparer and releaser are always different people | Form changes, principle does not |
Approval authority | System workflow and thresholds | Pre-authorized emergency approvers and lower thresholds for dual approval | Form changes, principle does not |
Approval evidence | Electronic workflow log | Signed paper log or out-of-band confirmation; never a single email or video call alone | Form changes, principle does not |
Three-way match | Automated | Waived only for Tier 1 vendors with recurring, predictable amounts; documented for post-outage review | Yes, with documentation |
Payment timing | Standard terms and payment runs | Prioritized by vendor tier; non-critical payments deferred | Yes |
Payment method | Preferred method per vendor | Any pre-approved emergency method (bank portal ACH or wire, checks from secure stock) | Yes, within the pre-approved list |
The fake identity/virtual reality question needs emphasis. Visual impersonation once seemed the stuff of “Mission Impossible,” but AI-generated voices and video can now defeat verification habits that were reasonable a few years ago. In the now-notorious case at Arup, a finance employee who had doubts set them aside thanks to a video call with deepfakes of the CFO and colleagues and disbursed $25.6 million to a fraudster account. Continuity guidance makes a related point: if the identity provider itself is compromised, the control plan must name an alternate authentication method, emergency access controls, manual verification and who can activate them. A crisis is precisely when a convincing executive request should get more scrutiny.
Build a Plan: Six Components
Companies that kept paying through an attack had these pieces ready beforehand. In one insurer's case study, a recovery team stood up a limited disaster-recovery environment and made manual payments using prior records, so payroll for 85,000 contractors went out on time. In another, a payments provider worked from an external export of vendors with upcoming payments to issue checks and electronic payments after ransomware shut down the customer's accounting system.
1. Critical vendor tiering
Prioritize vendors. Decide in calm conditions who you will pay first. A three-tier model works for most organizations:
Tier 1 – Pay regardless. Payroll and payroll taxes, utilities, critical-path suppliers, insurance premiums, debt service, regulatory obligations.
Tier 2 – Pay within the outage window if possible. Suppliers whose non-payment would disrupt operations within 30 days, and small suppliers heavily dependent on your business.
Tier 3 – Defer. Delay everything else—but communicate with the payees.
Your organization’s leadership should review and approve the tiering annually. Prioritization is a business decision, not an AP judgment call in the middle of a crisis.
2. Offline payment data
If your ERP is encrypted, the data you need is inside it. Maintain an offline copy, refreshed at least weekly and after each payment run, of:
• the vendor master for Tier 1 and Tier 2 vendors (legal name, remit-to, verified bank details, verified phone contact);
• the open payables and scheduled payments for the next 30 to 45 days;
• recurring payment amounts and due dates (rent, leases, debt service, payroll funding).
Store it encrypted, outside the production network and accessible to named individuals, and without corporate SSO (single sign-on). Treat it with the same access controls as the live vendor master.
3. Emergency payment channels
Document which channels will be used and confirm that they work independently of corporate systems:
• bank portal access for ACH and wire with credentials and tokens not tied to corporate SSO;
• a secondary banking relationship able to originate payments if the primary bank or processor is down;
• secure check stock and a manual check procedure, with positive pay files that can be submitted through the bank portal;
• confirmed bank contacts for each relationship, with phone numbers kept offline.
4. Emergency authority and roster
Name, in a board- or CFO-approved document:
• who can declare a disbursement continuity event and who can end it;
• emergency approvers and their limits, with at least two backups per role;
• the preparer and releaser roster that preserves segregation of duties when system roles are unavailable.
The Segregation of Duties Matrix and Payment System Access Control policy on this site can serve as the baseline the emergency roster maps back to.
5. Liquidity
An outage can stop collections at the same moment you choose to pay Tier 1 suppliers early or manually. ChangeHealthcare showed how quickly cash flow gaps open. Treasury should know how many weeks of Tier 1 payments the organization can fund with receipts disrupted, and which credit facilities can be drawn on short notice. JLR's approach, prepaying critical suppliers and covering their financing costs, is a supplier-liquidity tool worth having in the plan as an option.
6. Communications
Prepare the following templates before you need them.
• A supplier notice stating that payments may be delayed, that the company will not accept bank detail changes during the incident, and how vendors can verify any request claiming to come from you
• An internal notice reminding staff that payment requests arriving by email, text or video during the incident follow the degraded-mode rules
• A bank notice asking relationship banks to apply heightened scrutiny to unusual payment instructions
The supplier notice is also a fraud control. Criminals monitor public incident disclosures and send fake "updated payment instructions" to your vendors and to you.
Running Degraded Mode
Once systems fail, disbursements should follow a defined sequence. Each step has an owner and leaves a written record.
1. Stop and hold. Suspend all scheduled payment runs and any queued files at banks and processors until the scope of the incident is known. If compromise is suspected, assume recent changes to vendor master data are untrustworthy.
2. Declare. An authorized person must formally declare a disbursement continuity event, record the time and activate the emergency roster and degraded-mode rules.
3. Freeze changes. Freeze bank detail changes and new vendor setups. Notify relationship banks and send the supplier notice.
4. Establish the clean baseline. Retrieve the offline vendor master and payment schedule. Compare them against the most recent known-good backup if one is available; where they differ, the offline copy governs until verified.
5. Triage. Build the payment list for the outage window from Tier 1 obligations first, then Tier 2. Every item needs a source record: an invoice copy, contract, recurring schedule or payroll register.
6. Prepare, approve, release. Build each payment from the baseline. An emergency approver authorizes it and records the approval on the signed log. A different person releases it through the pre-approved channel.
7. Log everything. Make a manual log entry for each payment with payee, amount, method, date, source document, preparer, approver and releaser. This log is the bridge back to the system of record.
8. Review daily. The controller or designee reviews the log and bank activity daily, reconciles released payments to bank confirmations and resets priorities as the outage continues.
This sequence aligns with the Fraud Incident Response flowchart on this site. For an internal compromise, run both: incident response contains the attack, while degraded mode keeps essential payments moving under control.
Recovery and Reconciliation
The riskiest moment for duplicate payments is often the day systems come back. Queued payment files release, invoices paid manually are still open in the ERP and staff are eager to clear the backlog. JLR, for example, had to increase its invoice-processing capacity to work through the supplier payment backlog as systems were restored.
Before normal payment runs resume:
• Purge or review queued files. Confirm with every bank and processor which files and items were pending when the outage began and cancel or hold any that were already paid manually.
• Post every manual payment. Enter each degraded-mode payment from the manual log into the ERP against the original invoice, so the invoice closes and cannot be paid again.
• Run duplicate detection deliberately. Run duplicate checks across the full outage window plus a buffer on each side, including near matches on amount, date and invoice number. The Duplicate Payment policy on this site sets out the matching rules.
• Verify the vendor master before trusting it. After an internal compromise, compare restored vendor master records, especially bank details, against the offline baseline and the change log. Investigate every difference before the first automated run.
• Reconcile to the bank. Reconcile every account used during the outage, line by line, against the manual log.
• Close the event formally. The authorized person ends the continuity event in writing. Internal audit or the controller reviews the manual log, waived controls (such as three-way match exceptions) and any incidents, and records lessons learned.
Suppliers also need closure. Confirm what was paid, what remains open and when standard terms resume.
Testing the Plan
A plan that has never been tried is an assumption. Tabletop exercises are discussion-based simulations that help organizations prepare for various emergencies. They allow teams to practice their responses in a low-pressure environment, identifying gaps in emergency plans and improving overall preparedness.
Security experts consistently recommend tabletop and technical exercises, including encryption-only ransomware scenarios. Recovery objectives that exist only on paper don't help during a live incident. Finance should run its own exercises at least annually, alongside IT's.
Suggested scenarios, each run as a two-hour tabletop:
Scenario | Inject partway through | What it tests |
ERP encrypted two days before payroll and a major supplier payment run | A message that appears to be from the CFO asks for an urgent wire to a new account | Offline data, emergency roster, segregation of duties without system roles, refusal of executive pressure |
Primary payment processor down for five business days | The processor recovers on day three and releases its queue | Secondary channels, supplier communication, duplicate-payment prevention |
Regional power and telecom outage for 48 hours | A Tier 1 vendor calls a personal mobile asking to be paid to a different account | Deferral decisions, staff outside the region, the bank-change freeze |
Email compromise discovered during quarter-end close | Evidence that vendor master bank fields were edited last week | Vendor master verification against the offline baseline, bank notification, payment holds |
After each exercise, record what failed. Common findings include bank tokens tied to an employee's laptop, an offline file that hasn't been refreshed in months, and emergency approvers who don't know they are on the list. Assign owners and dates to each gap.