How to Detect and Prevent Accounts Payable Fraud: Complete Guide for Finance Teams (2026)

Introduction

Accounts payable fraud has evolved dramatically. While finance teams once dealt primarily with check fraud and simple invoice manipulation, today’s threats include sophisticated deepfake (Medius Fraud Research) invoices, business email compromise targeting banking change requests, and AI-generated fraudulent documents that can fool even experienced AP professionals.

According to the Association of Certified Fraud Examiners, the median loss from billing fraud schemes (ACFE Report to the Nations) exceeds $100,000 per incident, with detection taking an average of 12 months. The financial impact extends beyond direct losses—fraud incidents damage vendor relationships, trigger compliance issues, and erode stakeholder trust.

The good news: modern AP automation platforms like Peakflo can detect fraud patterns invisible to manual review, preventing fraudulent payments before they leave your bank account.

This comprehensive guide examines the fraud landscape finance teams face in 2026, explores detection methods that actually work, and provides a practical implementation framework for fraud prevention.

What is Accounts Payable Fraud?

Accounts payable fraud occurs when individuals—whether internal employees, external vendors, or organized criminal networks—manipulate the AP process to obtain unauthorized payments. Unlike honest billing errors or payment disputes, fraud involves intentional deception designed to extract money from your organization.

How Common is Accounts Payable Fraud?

The data paints a concerning picture:

• 79% of organizations experienced attempted or actual payments fraud in 2024 (AFP Payments Fraud Survey)
• 63% identified business email compromise as the top fraud avenue
• 44% of businesses have been targets for invoice fraud (Medius Research)
• 53% of finance professionals have encountered deepfake scamming attacks

The fraud landscape has shifted dramatically with AI technology. Fraudsters now generate invoices, alter banking details, and impersonate executives with unprecedented sophistication. Traditional detection methods—manual review, exact-match duplicate checking, basic approval workflows—increasingly fail against these advanced schemes.

What Are the Most Common Types of Accounts Payable Fraud?

1. How Does Invoice Fraud Work?

Invoice fraud schemes manipulate billing documents to obtain unauthorized payments:

Duplicate Invoices Fraudsters submit the same invoice multiple times with slight variations—different invoice numbers, altered dates, or modified line items. Traditional ERP systems checking only exact invoice number matches miss these sophisticated duplicates, resulting in multiple payments for single transactions.

Inflated Invoices Vendors or internal employees increase quantities, inflate unit prices, or add unauthorized charges to legitimate invoices. Without three-way matching (purchase order, receipt, invoice), these modifications often go undetected until reconciliation—if caught at all.

Fictitious Invoices Completely fabricated invoices for goods or services never ordered or received. These often target companies with decentralized purchasing, weak vendor validation, or no purchase order requirements.

Real-world impact: A multinational manufacturer discovered $2.3 million in duplicate and inflated invoices over 18 months, with fraud perpetrated by a trusted vendor and complicit AP clerk.

2. What is Vendor Impersonation Fraud?

Vendor impersonation has become the fastest-growing fraud category, driven by business email compromise sophistication:

Banking Detail Changes Fraudsters impersonate legitimate vendors, requesting banking information updates via email. Without verification protocols, AP teams process the change, sending legitimate payments to fraudulent accounts. By the time the actual vendor inquires about missing payment, funds have vanished.

The NACHA fraud monitoring rule (NACHA Account Validation Rule) (effective March 2026) specifically addresses this threat, requiring enhanced verification for account changes.

Shell Company Creation Criminals establish fake companies with names similar to legitimate vendors, submit invoices for services never rendered, and disappear after payment. These schemes exploit companies lacking comprehensive vendor onboarding validation.

3. How Does Internal Employee Fraud Happen in AP?

Internal fraud leverages insider access and system knowledge:

Ghost Employees/Vendors Employees with system access create fictitious vendors or ghost employees in the master file, then approve invoices and payments to accounts they control. Weak segregation of duties makes this frighteningly simple—if one person can both create vendors AND approve payments, fraud becomes inevitable.

Kickback Schemes AP employees collude with vendors, approving inflated invoices in exchange for cash payments or other benefits. These schemes often involve multiple parties: the vendor inflates pricing, the AP employee approves without scrutiny, and both parties share the excess.

Check Tampering Despite the decline of check usage, checks remain 7x more vulnerable to fraud than virtual cards. Employees with check-signing authority or access to signed checks alter payee names, amounts, or endorsements.

4. What is Payment Diversion Fraud?

Payment diversion redirects legitimate payments to fraudulent accounts:

ACH Fraud Criminals obtain banking credentials through phishing, business email compromise, or insider access, initiating unauthorized ACH payments. The NACHA rule changes specifically target this threat vector.

Wire Transfer Fraud Fraudsters impersonate executives (often via compromised email) requesting urgent wire transfers for supposedly confidential transactions. These “CEO fraud” schemes exploit hierarchical cultures where employees fear questioning executive requests.

Real-world example: A healthcare organization lost $4.8 million to wire transfer fraud when an accounting manager received an email appearing to be from the CFO requesting an urgent acquisition-related payment.

—

Common AP Fraud Schemes Comparison

How Can You Detect Accounts Payable Fraud?

Detection requires layered approaches combining technology, process controls, and human vigilance.

What Red Flags Indicate Potential AP Fraud?

Vendor-Related Warning Signs:

• New vendors with incomplete or unverifiable information
• Vendors sharing addresses, phone numbers, or bank accounts with employees
• Post office boxes as vendor addresses (especially for claimed suppliers of physical goods)
• Vendor names similar to known legitimate suppliers
• Unusual concentration of business with specific vendors

Invoice-Related Red Flags:

• Sequential invoice numbers from supposedly unrelated vendors
• Invoices just below approval thresholds (e.g., multiple $9,900 invoices when $10,000 requires additional approval)
• Round numbers without itemization ($10,000.00 instead of $9,847.23)
• Invoices lacking purchase order references
• Duplicate vendor names with slightly different spellings

Process-Related Indicators:

• Payments processed outside normal cycles
• Rush payment requests without business justification
• Overridden approval workflows
• Manual checks or wire transfers when standard payment terms should apply
• Employees refusing to take vacation (maintaining continuous control to hide fraud)

How Does AI Detect Fraudulent Invoices in Accounts Payable?

Artificial intelligence transforms fraud detection from reactive review to proactive prevention:

Pattern Recognition Across Dimensions AI analyzes multiple data points simultaneously—invoice amounts, dates, vendor information, line-item details, payment patterns—identifying anomalies invisible to manual review. Where humans can realistically compare 3-5 factors, AI evaluates hundreds, recognizing subtle patterns indicating fraud.

Peakflo’s AI-powered fraud detection continuously learns your organization’s normal payment patterns. When a vendor typically billing $5,000 monthly suddenly submits a $50,000 invoice, the system flags it instantly for verification.

Duplicate Detection Beyond Exact Matches Traditional ERP duplicate checking matches exact invoice numbers. Sophisticated fraud varies numbers, dates, or amounts slightly. AI detects these sophisticated duplicates by analyzing:

• Vendor + amount + date proximity
• Line item similarities despite different invoice numbers
• Payment patterns suggesting intentional duplication
• Invoice sequencing anomalies

Vendor Validation Intelligence AI cross-references vendor data across multiple dimensions:

• Banking details consistency with company registration
• Address verification against business databases
• Email domain validation
• Historical payment pattern analysis
• Vendor name similarity detection (catching “Acme Corp” vs “Acme Corporation” schemes)

Behavioral Anomaly Detection Machine learning identifies unusual user behavior:

• Approval patterns deviating from historical norms
• Access attempts outside normal working hours
• Unusual combinations of actions (vendor creation + immediate invoice approval)
• Workflow overrides concentration

Real-Time Risk Scoring Modern platforms assign risk scores to every invoice and payment, flagging high-risk transactions for additional review before payment processing. This shifts fraud prevention from periodic audit (discovering fraud months later) to real-time intervention.

According to Medius research, AI fraud detection reduces fraudulent payment processing by 80-95% compared to manual review alone.

What is Benford’s Law and How Does It Detect Fraud?

Benford’s Law provides a mathematical fraud detection approach based on natural number distribution patterns:

In naturally occurring datasets, leading digits follow a predictable distribution—approximately 30% of numbers start with “1”, 18% with “2”, declining to just 5% starting with “9”. Invoice amounts, expense reports, and payment data should follow this pattern.

Significant deviations signal potential manipulation. If your invoice data shows 25% starting with “7” or “8” (well above the expected 5-6%), either someone is manufacturing invoices near approval thresholds, or systematic manipulation is occurring.

Implementation: Export invoice amounts for specific vendors or time periods, analyze leading digit distribution, and investigate significant deviations (typically >5% from expected Benford distribution).

Limitation: Benford’s Law works best with large datasets (200+ transactions) and doesn’t apply to restricted ranges (e.g., if all invoices must be $1,000-$2,000, the distribution won’t follow Benford’s Law).

How Do You Prevent Accounts Payable Fraud?

Prevention requires structural controls, not just policy documents. Here’s a comprehensive framework:

What Internal Controls Prevent AP Fraud?

Segregation of Duties Enforcement The foundational fraud prevention control: ensure no single individual can both initiate AND approve transactions.

Critical separations:

• Vendor master file creation ≠ invoice approval
• Invoice approval ≠ payment processing
• Bank account reconciliation ≠ payment processing
• Check signing ≠ check preparation

System-enforced segregation (built into your AP platform) works better than policy-based segregation (dependent on manual compliance). Peakflo enforces role-based segregation at the platform level, preventing users from circumventing controls.

Mandatory Approval Workflows Structured approval routing based on:

• Invoice amount thresholds
• Department/cost center
• Vendor category
• Purchase order presence

Critical requirement: approval workflows must be system-enforced and un-bypassable. If approvers can override or skip steps, the control fails.

Three-Way Matching Before payment approval, verify:

1. Purchase order confirms the purchase was authorized
2. Receiving document confirms goods/services were delivered
3. Invoice confirms billing accuracy

Three-way matching catches fictitious invoices (no PO), delivery fraud (no receipt), and billing errors. Automation dramatically improves matching efficiency—manual three-way matching consumes 15-20 minutes per invoice; automated matching completes in seconds.

Vendor Master File Controls

• Formal vendor onboarding with documentation requirements
• Dual authorization for vendor creation
• Regular vendor master audits (quarterly minimum)
• Automated duplicate vendor detection
• Vendor validation against business registries

Fraud Prevention Controls Effectiveness

How Can Vendor Validation Prevent Fraud?

Comprehensive vendor validation creates friction for fraudsters while streamlining legitimate vendor relationships:

Initial Onboarding Validation

• Business registration verification (company name, registration number, tax ID)
• Banking detail verification (account ownership matches registered business)
• Address verification (physical address, not just PO box)
• Reference checks for new vendors
• W-9/W-8 collection (US) or equivalent tax documentation

Ongoing Monitoring

• Periodic vendor certification (annual minimum)
• Banking detail change verification (dual-channel confirmation)
• Payment pattern monitoring (flagging unusual frequency or amount changes)
• Vendor relationship validation (confirming vendor contacts at stated phone/email)

Banking Change Protocol The single most critical fraud prevention control given the prevalence of vendor impersonation:

1. When receiving banking change request, never reply to the email received
2. Call vendor using phone number from previous invoice or original contract (not from change request email)
3. Verbally confirm banking change with authorized vendor contact
4. Document verification including date, contact person, and confirmation method
5. Implement temporary hold period (48-72 hours) before first payment to new account

This dual-channel verification (email request + phone confirmation via independent contact info) defeats most vendor impersonation schemes.

What Role Does AP Automation Play in Fraud Prevention?

AP automation platforms provide fraud prevention capabilities impossible with manual processing:

Intelligent Duplicate Detection Moving beyond simple invoice number matching to sophisticated analysis:

• Vendor + amount + date proximity analysis
• Line-item similarity detection
• Payment history pattern matching
• Fuzzy matching for vendor name variations

Automated Vendor Validation

• Real-time vendor data cross-referencing
• Banking detail validation against business registries
• Automatic flagging of vendor data inconsistencies
• Duplicate vendor identification (different names, same banking details)

Complete Audit Trails Every action logged with user ID, timestamp, and IP address:

• Who created vendors
• Who approved invoices
• Who processed payments
• Who modified vendor banking details
• Who overrode approval workflows

This comprehensive audit trail enables both fraud detection (identifying suspicious patterns) and fraud investigation (reconstructing event sequences).

Real-Time Alerts Immediate notification of high-risk scenarios:

• Vendor banking changes
• Invoices exceeding normal amounts
• Duplicate invoice detection
• Approval workflow overrides
• Payment processing outside normal hours

Analytics and Reporting Fraud detection dashboards highlighting:

• Vendor concentration (over-reliance on specific vendors)
• Payment pattern anomalies
• Duplicate payment analysis
• Vendor master file quality metrics
• Control override frequency

Peakflo’s AP automation platform combines all these capabilities in a unified system, reducing fraud risk while accelerating legitimate payment processing.

How Does Peakflo Prevent Accounts Payable Fraud?

Peakflo’s AP automation platform builds fraud prevention into every process step, from vendor onboarding through payment execution:

Key Fraud Prevention Features

1. AI-Powered Duplicate Detection Peakflo’s intelligent duplicate detection analyzes multiple dimensions simultaneously:

• Invoice numbers, amounts, and dates
• Vendor information and banking details
• Line-item content and descriptions
• Payment history and patterns

Unlike basic ERP duplicate checking (matching only exact invoice numbers), Peakflo catches sophisticated duplicates with intentional variations, preventing duplicate payments before processing.

2. Automated Vendor Validation Comprehensive vendor validation workflow:

• Automated business registry verification
• Banking detail validation
• Duplicate vendor detection across name variations
• Mandatory documentation collection
• Dual-authorization for vendor creation

3. System-Enforced Segregation of Duties Role-based access control ensuring critical separations:

• Vendor creation roles separate from approval roles
• Invoice approval separate from payment execution
• Configurable approval hierarchies by amount, department, or vendor
• Un-bypassable workflow enforcement

4. Real-Time Fraud Detection Continuous monitoring flagging high-risk transactions:

• Banking detail changes requiring additional verification
• Invoice amounts deviating from vendor patterns
• Rushed payment requests
• Unusual approval patterns
• After-hours activity

5. Complete Audit Trail Comprehensive logging of all system activity:

• Every vendor creation, modification, and deletion
• All invoice approvals and rejections
• Payment processing with user and timestamp
• Workflow overrides with justification requirements
• Banking detail changes with verification records

Peakflo Customer Success Stories

Real organizations achieving measurable results with Peakflo’s AP automation:

NRI: 75% Faster Invoice Processing

Nomura Research Institute reduced invoice processing time by 75% and eliminated manual data entry errors through automated OCR and intelligent routing.

Key Results:

• 75% reduction in processing time
• 100% elimination of manual data entry
• Real-time payment visibility across entities

Haisia: 90% Reduction in Payment Delays

Haisia achieved 90% reduction in late payments and captured $127,000 in early payment discounts in year one.

Key Results:

• 90% fewer late payments
• $127,000 in discount capture
• Complete audit trail compliance

Vida: Multi-Entity Consolidation

Vida consolidated AP operations across 8 entities, reducing vendor onboarding from 3 weeks to 2 days.

Key Results:

• 92% faster vendor onboarding
• Centralized multi-entity visibility
• 85% reduction in payment failures

Reyid: Automated Vendor Validation

Reyid eliminated vendor fraud attempts through automated validation and duplicate detection.

Key Results:

• Zero fraud incidents post-implementation
• 100% duplicate prevention
• $180,000 in prevented fraudulent payments

MyRobin: Scalable AP Infrastructure

MyRobin scaled from 500 to 5,000+ monthly invoices without adding AP headcount.

Key Results:

• 10x invoice volume with same team
• 95% automation rate
• 98% on-time payment achievement

Explore More Customer Stories →

See Peakflo’s Fraud Prevention in Action

Experience how Peakflo protects your organization from AP fraud:

• 🎯 Prevent duplicate payments with AI-powered detection analyzing multiple data points
• 🔒 Validate vendors automatically with business registry verification and banking detail checks
• 📊 Monitor fraud risk in real-time with intelligent alerts and risk scoring
• ✅ Maintain complete audit trails for compliance and investigation

CTA: Schedule Your Peakflo Demo

Our Verdict: AI Duplicate Detection Is Non-Negotiable for Modern AP

Our analysis of duplicate payment prevention reveals a stark reality: traditional ERP duplicate checking fails to prevent 50-60% of sophisticated duplicates. Organizations processing 500+ invoices monthly simply cannot achieve acceptable duplicate prevention rates through manual review.

What Works:

• ✅ AI multi-dimensional duplicate detection
• ✅ Centralized invoice processing (single intake point)
• ✅ Automated three-way matching
• ✅ Real-time duplicate alerts with confidence scoring
• ✅ Fuzzy matching for vendor name variations

What Doesn’t Work:

• ❌ Exact invoice number matching only
• ❌ Decentralized invoice processing across departments
• ❌ Manual review for high invoice volumes
• ❌ Reactive duplicate detection after payment

Bottom Line: AI duplicate detection pays for itself immediately—companies processing $10M annually in AP typically lose $100,000-$500,000 to duplicates. Automation achieving 95-99% detection rates delivers ROI in the first quarter.

Recommended Next Steps:

1. Calculate current duplicate payment cost (analyze last 12 months)
2. Centralize invoice intake (eliminate multiple entry points)
3. Deploy AI duplicate detection platform
4. Enable automated three-way matching for PO-backed invoices
5. Provide vendor portal for status visibility (reduces duplicate submissions)

Conclusion: Building a Fraud-Resistant AP Process

The accounts payable fraud landscape continues evolving, with AI-generated fraudulent documents, sophisticated business email compromise, and coordinated fraud rings creating unprecedented challenges. Relying on manual review, basic duplicate checking, and policy-based controls leaves organizations vulnerable to schemes specifically designed to exploit these weaknesses.

Effective fraud prevention requires three elements:

1. Technology-enforced controls that prevent circumvention (not policy-dependent controls that rely on compliance)
2. AI-powered detection identifying patterns invisible to human review
3. Process standardization eliminating the inconsistencies fraudsters exploit

Organizations implementing comprehensive AP automation report 80-95% fraud risk reduction compared to manual processes. The primary variables determining success are system-enforced segregation of duties, intelligent duplicate detection capabilities, and automated vendor validation.

Next Steps for Fraud Prevention:

1. Audit current controls - Assess segregation of duties enforcement, duplicate detection capabilities, and vendor validation processes
2. Implement dual-channel verification - Require phone confirmation (via independent contact info) for all banking detail changes
3. Enable three-way matching - For all purchases exceeding your threshold (recommend $1,000-$5,000 depending on volume)
4. Evaluate AP automation - Platforms like Peakflo provide fraud prevention capabilities impossible with manual processing
5. Conduct fraud risk training - Ensure AP team recognizes red flags and understands verification protocols

The question isn’t whether your organization will face fraud attempts—79% already have. The question is whether your controls will detect and prevent them before payment processing.

Strengthen Your AP Fraud Defenses with Peakflo

Discover how Peakflo’s AI-powered fraud detection, automated vendor validation, and system-enforced controls protect your organization from the fraud schemes targeting finance teams in 2026.

Schedule Your Demo | Explore AP Automation

Frequently Asked Questions

What is accounts payable fraud?

Accounts payable fraud involves intentional deception to obtain unauthorized payments through the AP process. Common schemes include duplicate invoices, vendor impersonation, fictitious invoices, and payment diversion. Unlike honest billing errors, fraud requires deliberate manipulation to extract money. Organizations face median losses exceeding $100,000 per incident, with 79% experiencing fraud attempts annually.

How common is AP fraud?

AP fraud affects 79% of organizations annually according to AFP research. Business email compromise targets 63% of companies, while 44% face invoice fraud attempts. The median fraud loss exceeds $100,000, with detection averaging 12 months. Fraud risk has increased significantly with AI-generated fraudulent documents and sophisticated impersonation schemes making detection harder.

What are the red flags of invoice fraud?

Key invoice fraud indicators include sequential invoice numbers from unrelated vendors, round-number amounts without itemization, invoices just below approval thresholds, missing purchase order references, and duplicate vendor names with spelling variations. Additional red flags: new vendors with incomplete information, vendor addresses matching employee addresses, rush payment requests, and unusual payment pattern changes.

How does AI detect AP fraud?

AI analyzes hundreds of data points simultaneously—amounts, dates, vendors, line items, payment patterns—identifying anomalies invisible to manual review. Machine learning detects sophisticated duplicates varying invoice numbers or dates, validates vendor information across business registries, identifies behavioral anomalies in user actions, and assigns real-time risk scores. AI reduces fraudulent payments by 80-95% versus manual review.

What is the most effective way to prevent duplicate payments?

Effective duplicate prevention combines AI-powered detection analyzing multiple dimensions (vendor, amount, date, line items) beyond basic invoice number matching, centralized invoice processing eliminating multiple entry points, three-way matching verifying purchase orders and receipts, automated vendor master file management preventing duplicate vendor records, and regular payment reconciliation. AP automation platforms provide detection capabilities impossible manually.

How can you prevent vendor impersonation fraud?

Prevent vendor impersonation through dual-channel verification for banking changes (calling vendor at previously-verified phone number to confirm email requests), comprehensive vendor onboarding with business registration and banking detail validation, automated monitoring flagging banking detail change requests, mandatory hold periods before first payment to new accounts, and employee training recognizing business email compromise tactics.

What internal controls prevent AP fraud?

Critical controls include system-enforced segregation of duties preventing any individual from both initiating and approving transactions, mandatory approval workflows based on amount thresholds and categories, three-way matching verifying purchase orders against receipts and invoices, vendor master file controls requiring dual authorization, and complete audit trails logging all system activity with user identification and timestamps.

How does three-way matching prevent fraud?

Three-way matching prevents fraud by requiring invoices to match approved purchase orders (confirming authorization) and receiving documents (confirming delivery) before payment approval. This catches fictitious invoices (no corresponding PO), delivery fraud (no receipt confirmation), and billing errors. Automation improves efficiency, completing matching in seconds versus 15-20 minutes manually.

What is Benford’s Law in fraud detection?

Benford’s Law states that in naturally-occurring datasets, leading digits follow predictable distribution (30% start with “1”, declining to 5% with “9”). Invoice amounts should follow this pattern. Significant deviations indicate potential manipulation—someone manufacturing invoices near approval thresholds. Applies to large datasets (200+ transactions) but not restricted ranges. Deviations exceeding 5% from expected distribution warrant investigation.

How do you detect ghost vendors in accounts payable?

Detect ghost vendors by analyzing vendor master files for duplicate banking details across different vendor names, post office box addresses for claimed suppliers of physical goods, vendors sharing addresses or phone numbers with employees, new vendors lacking business registration verification, and payment concentration to specific vendors. Regular vendor master audits and automated duplicate detection help identify ghost vendors.

What should you do if you discover AP fraud?

Upon discovering fraud: immediately suspend related payments and freeze affected vendor accounts, document all evidence without alerting suspected perpetrators, notify management and legal counsel, engage forensic accounting expertise for complex schemes, report to law enforcement and pursue prosecution, conduct comprehensive control assessment identifying vulnerabilities, implement remediation measures preventing recurrence, and consider cybersecurity insurance claims if applicable.

How much does AP fraud cost businesses?

AP fraud costs businesses 5-7% of annual revenue according to ACFE research. Median losses per incident exceed $100,000, with detection averaging 12 months. Beyond direct losses, costs include investigation expenses, legal fees, damaged vendor relationships, compliance issues, and reputational damage. Organizations implementing comprehensive fraud prevention reduce losses by 80-95% through early detection and prevention.

Can AP automation prevent all fraud?

AP automation dramatically reduces fraud risk (80-95% reduction) but cannot eliminate it entirely. Sophisticated fraudsters adapt to controls, and some schemes (like collusion between employees and vendors) can circumvent even strong technical controls. Effective fraud prevention combines technology (AI detection, automated validation), process controls (segregation of duties, approval workflows), and human vigilance (employee training, red flag awareness).

What payment methods have the highest fraud risk?

Checks face 7x higher fraud risk than virtual cards, with 63% of fraud attempts targeting checks versus only 9% for virtual cards. Wire transfers carry significant fraud risk due to irreversibility and common use in business email compromise schemes. ACH fraud has increased, prompting the NACHA fraud monitoring rule effective March 2026. Virtual cards and automated payment platforms provide significantly lower fraud exposure.

How often should you audit AP processes for fraud?

Conduct comprehensive AP fraud risk assessments quarterly at minimum, with monthly reviews of high-risk areas (vendor master file changes, banking detail updates, approval workflow overrides). Implement continuous monitoring through AP automation platforms providing real-time fraud detection. Annual external audits provide independent validation. High-risk industries or organizations with previous fraud incidents should increase audit frequency to monthly comprehensive reviews.