Invoice Matching: 2-Way, 3-Way, and 4-Way Explained

Invoice matching is a verification control in accounts payable that compares a supplier invoice against supporting documents before authorizing payment. Those supporting documents typically include the purchase order, goods receipt, and inspection report. The process catches overpayments, flags billing errors, and deters fraud before money leaves the organization.

Without this verification step, organizations expose themselves to a straightforward but costly problem: paying for goods they never ordered, quantities they never received, or prices they never agreed to. A single transposed digit on an invoice line item can result in thousands of dollars in overpayment. Multiply that across hundreds or thousands of invoices per month, and the cumulative overpayment can reach five or six figures annually.

This guide covers the three main types of invoice matching: 2-way, 3-way, and 4-way. Beyond defining each type, it explains when each is appropriate based on transaction risk and volume. It also provides practical guidance on setting tolerance thresholds for price and quantity variances, diagnosing and resolving common matching failures, and understanding how upstream data quality directly affects matching accuracy.

---

Why Invoice Matching Matters for Accounts Payable

Invoice matching is a front-line internal control that stands between your organization and three specific categories of financial risk.

Billing fraud is the most direct threat. A supplier submits an invoice for goods or services that were never delivered, or inflates quantities and prices beyond what was agreed. According to research published in ACFE's Fraud Magazine, billing schemes are the most common type of asset misappropriation, with a median loss of $100,000 per case and nearly 29% of victim organizations lacking adequate controls to prevent the fraud. Without a structured matching process, these schemes can persist for months before detection.

Overpayment through pricing errors, duplicate invoices, and quantity discrepancies is routine in high-volume accounts payable environments. A supplier may bill at a rate that differs from the purchase order, or submit the same invoice twice under slightly different reference numbers. Invoice verification before payment catches these discrepancies at the point where correction is cheapest, before funds leave the organization rather than after, when recovery requires lengthy dispute processes.

Compliance failure during audits creates a different kind of exposure. Auditors expect organizations to demonstrate that every payment was verified against supporting documentation before it was authorized. When that verification trail does not exist, the finding is not just a procedural gap. It signals a control environment where unauthorized or inaccurate payments could go undetected, which raises questions about the reliability of the entire payable process.

Beyond these defensive functions, matching is also a fundamental step in how invoice approval workflows control payment authorization. Matching verifies that the invoice data is accurate and supported by documentation. The approval step that follows then authorizes the actual payment. These are distinct controls: one confirms the facts, the other authorizes the action. Collapsing them into a single step weakens both.

Matching is sometimes confused with two related but distinct processes. Validation confirms that an invoice meets formal requirements: correct format, valid vendor, proper authorization, and complete fields. Reconciliation compares payment records against bank statements and ledger entries after payment has occurred. Matching sits between these two steps: it verifies that the invoice data is accurate against procurement documents before payment is authorized, catching errors that validation cannot detect and reconciliation would only surface after the money has already left.

Matching also affects financial reporting accuracy. When invoices are paid without verification, the underlying expense records and accruals may reflect incorrect amounts. Over time, these inaccuracies compound, distorting period-end financials and creating reconciliation problems that consume far more effort than the original verification would have required.

---

Comparing 2-Way, 3-Way, and 4-Way Matching

The three matching types differ in how many documents are cross-referenced during the invoice matching process and, consequently, what level of verification each provides before a payment is approved.

Matching Type	Documents Compared	Best Used For	Verification Level
2-Way	Invoice + Purchase Order	Services, subscriptions, recurring purchases	Basic
3-Way	Invoice + Purchase Order + Goods Receipt	Physical goods, standard procurement	Standard
4-Way	Invoice + Purchase Order + Goods Receipt + Inspection Report	High-value goods, regulated industries, quality-critical purchases	Maximum

Three supporting documents make up the foundation of accounts payable matching. A purchase order is the buyer's formal authorization to acquire goods or services at agreed-upon prices and quantities. A goods receipt note confirms that the ordered items physically arrived at the expected location and records the quantities actually received. An inspection report goes one step further, documenting whether the delivered goods meet the quality standards, specifications, or regulatory requirements defined at the time of purchase.

The key decision principle is straightforward: each additional document adds a layer of verification, but it also adds complexity and creates more potential for exceptions that require manual review. Choosing the right matching type depends on the transaction itself. Low-risk recurring services rarely justify four-document verification, while a shipment of pharmaceutical ingredients almost certainly does. Effective matching programs often apply different levels to different purchase categories rather than enforcing a single policy across all spend.

These matching types all assume a purchase order exists. Not every business operates this way. Sole traders, freelancers, and small businesses that purchase on an ad-hoc basis without formal POs may not need structured matching at all. For these organizations, invoice verification against a bank statement or payment record may be sufficient. The matching framework described here becomes relevant when procurement volume and transaction value justify formal purchase orders.

---

2-Way Matching: Invoice Against Purchase Order

Two-way matching is the process of comparing a supplier invoice against the original purchase order to verify that the billed amounts and terms match what was agreed before the transaction took place. It is the most fundamental form of invoice matching and serves as the baseline control for accounts payable verification.

In a 2-way match, the AP clerk checks the following data points between the invoice and the purchase order:

• Vendor name: the supplier on the invoice must match the supplier on the PO
• PO number reference: the invoice must cite a valid, open purchase order
• Line items and unit prices: each billed item and its per-unit cost must align with what was authorized on the PO
• Invoice total vs. PO total: the overall billed amount must fall within the approved order value
• Payment terms: net-30, net-60, or other agreed terms must be consistent across both documents

When 2-Way Matching Is Sufficient

Purchase order matching at the 2-way level works well for transactions where no physical goods change hands. Common examples include:

• Service-based purchases such as consulting engagements, janitorial or cleaning contracts, and IT support agreements
• Software subscriptions billed monthly or annually
• Recurring fixed-cost purchases where the deliverable is not a tangible product

The reason 2-way matching is sufficient here is straightforward: there is no physical delivery to verify. Adding a goods receipt check to a consulting invoice would introduce additional process steps without providing any meaningful control value, because nothing was shipped or received in a warehouse.

Worked Example

A mid-size company contracts with an IT support provider for ongoing helpdesk coverage at $2,000 per month. The procurement team issues PO-4510 authorizing 12 months of service at that rate.

At the end of January, the IT provider sends an invoice for $2,000 referencing PO-4510. The AP clerk pulls up the purchase order and runs through the comparison:

Data Point	Purchase Order (PO-4510)	Supplier Invoice
Vendor	Apex IT Solutions	Apex IT Solutions
PO reference	PO-4510	PO-4510
Service description	Monthly IT helpdesk support	Monthly IT helpdesk support
Amount	$2,000	$2,000
Payment terms	Net-30	Net-30

Every field aligns. The invoice passes the 2-way match and moves forward to payment approval.

The Limitation of 2-Way Matching

Because PO matching at this level only compares two documents, it cannot detect problems that occur between the order and the invoice. If a company orders 500 units of a product but the supplier ships only 400, a 2-way match will not catch the discrepancy. The invoice may perfectly reflect the original PO, yet the organization pays for goods it never received. Similarly, if the wrong items were shipped or a delivery arrived damaged, nothing in the invoice-to-PO comparison would surface those issues.

For any transaction involving physical goods, organizations typically need an additional verification step: confirming what was actually delivered. That confirmation comes from the goods receipt, and incorporating it into the matching process is what defines 3-way matching.

---

3-Way Matching: Adding the Goods Receipt

Three-way matching is the process of comparing three documents before authorizing a supplier payment: the purchase order, the goods receipt (also called a goods received note, or GRN), and the supplier invoice. It is the most widely used matching type in accounts payable and is considered the standard control for most organizations that purchase physical goods.

The goods receipt is the document that separates 3-way matching from 2-way matching. Where 2-way matching only confirms that an invoice aligns with what was ordered, the goods receipt confirms that the ordered goods were physically received, in the correct quantity, and in acceptable condition. This closes a significant control gap. Without it, an organization could pay for goods that never arrived, arrived short, or arrived damaged.

What a 3-Way Match Compares

Three specific comparisons take place during a 3-way match:

1. Invoice vs. purchase order: Unit prices, extended amounts, payment terms, and line item descriptions on the invoice are checked against the original PO to confirm the supplier is billing according to the agreed terms.
2. Invoice vs. goods receipt: Quantities on the invoice are compared to quantities recorded on the goods receipt to verify the supplier is only billing for what was actually delivered.
3. Purchase order vs. goods receipt: The goods receipt is compared to the PO to confirm whether the order has been fulfilled in full, partially fulfilled, or overshipped.

All three comparisons must pass within acceptable tolerance thresholds before payment is released.

When to Use 3-Way Matching

Three-way matching is appropriate for any procurement transaction where physical goods or materials are shipped and received. This includes standard business purchases, raw materials for manufacturing, inventory replenishment, office supplies ordered in bulk, and equipment acquisitions. If something is delivered to a dock, a warehouse, or a receiving desk, a goods receipt should exist, and 3-way matching should apply.

Worked Example

A manufacturer issues PO-7823 for 500 units of Part A at $12.00 per unit, totaling $6,000. The shipment arrives at the warehouse, and the receiving team counts 480 units. They log a goods receipt for 480 units in acceptable condition.

The supplier sends an invoice for 500 units at $12.00 each, totaling $6,000.

The 3-way match runs and fails on the quantity check. The invoice claims 500 units, but the goods receipt confirms only 480 were received. The discrepancy is 20 units, or $240.

Document	Quantity	Unit Price	Total
Purchase order (PO-7823)	500	$12.00	$6,000.00
Goods receipt	480	N/A	N/A
Supplier invoice	500	$12.00	$6,000.00

The AP department holds payment and contacts the supplier with two options: adjust the invoice to $5,760 (480 units at $12.00) to reflect what was actually received, or arrange delivery of the remaining 20 units before the original invoice is paid.

This is exactly the kind of overpayment that 2-way matching alone would miss. The PO and the invoice agree perfectly. Only the goods receipt reveals the shortfall.

For organizations handling high-value purchases or operating in regulated industries, a fourth document can be added to the match: an inspection report that confirms quality and specifications before payment is authorized.

---

4-Way Matching: Including the Inspection Report

Four-way matching is the most rigorous level of invoice verification in accounts payable. It compares four documents before authorizing payment: the supplier invoice, the purchase order, the goods receipt, and an inspection or quality acceptance report.

The first three documents serve the same function as in 3-way matching. The inspection report adds a fourth layer: verification that the received goods meet quality specifications, contractual standards, or regulatory requirements. Where the goods receipt confirms that items arrived in the correct quantity, the inspection report confirms that those items are actually fit for their intended use. This distinction matters because a shipment can be complete and on time while still failing to meet the quality standards the buyer requires.

When 4-Way Matching Is Justified

Four-way matching makes sense in procurement categories where quality failures carry serious consequences:

• High-value capital equipment purchases where defective components could cause operational shutdowns or safety hazards
• Regulated industries such as pharmaceuticals, aerospace, and food manufacturing, where quality certification is mandatory before materials enter production
• Raw materials where defects cause downstream production failures, making it cheaper to catch problems at receiving than after processing has begun
• Government contracts with formal acceptance testing requirements, where payment authorization depends on documented proof that deliverables meet contractual specifications

In each of these cases, a structured inspection process already exists independently of the matching workflow. The 4-way match ties that inspection outcome into the payment approval chain.

When 4-Way Matching Is Not Appropriate

Adding a fourth matching point where no genuine inspection process exists creates paperwork without adding control value. Four-way matching is unnecessary for low-value purchases, standard office supplies, routine services, or any procurement category where quality inspection is not formally performed. If no one is conducting an actual inspection and generating a documented report, requiring one solely to satisfy a matching rule introduces delay and administrative cost with no corresponding reduction in risk. Organizations should reserve 4-way matching for the transaction categories that genuinely warrant it.

Worked Example: Pharmaceutical Raw Material Purchase

A pharmaceutical company orders 10,000 units of a chemical compound at $50 per unit under PO-3301, for a total commitment of $500,000. The warehouse receives all 10,000 units and records a goods receipt confirming the delivery quantity. The quality lab pulls samples from the shipment, runs purity and composition tests, and issues an inspection report confirming the compound meets the required specifications. The supplier then submits an invoice for $500,000.

The AP team runs the 4-way match:

Document	Quantity	Unit Price	Total	Quality Status
Purchase Order (PO-3301)	10,000 units	$50.00	$500,000	N/A
Goods Receipt	10,000 units	N/A	N/A	Received
Inspection Report	10,000 units	N/A	N/A	Passed
Supplier Invoice	10,000 units	$50.00	$500,000	N/A

All four documents align. The PO confirms the agreed quantity and price. The goods receipt confirms delivery. The inspection report confirms the materials meet quality standards. The invoice matches on both quantity and amount. Payment is authorized.

Had the inspection report flagged a purity deficiency in 2,000 of the 10,000 units, the match would have failed on the quality dimension even though the quantity and price were correct. The AP team would hold payment until the quality issue was resolved, whether through a return, replacement, or negotiated credit, preventing the organization from paying full price for materials that do not meet specification.

---

How to Set Tolerance Thresholds for Price and Quantity Variances

A tolerance threshold is a pre-defined acceptable variance between documents that allows a match to pass without manual review. Tolerances exist because minor discrepancies are normal in business transactions. Rounding differences, small shipping adjustments, and tax calculation variations occur routinely and do not warrant holding every invoice for human intervention.

Two main types of tolerance govern most invoice matching configurations: price tolerance and quantity tolerance.

Price Tolerance

Price tolerance defines how much the invoiced unit price can differ from the purchase order price before the system flags an exception. Organizations express this as a percentage, an absolute dollar amount, or both.

Common ranges for price tolerance:

• Standard purchases: 1-5% is typical for routine, moderate-value transactions. A 2-3% threshold catches genuine pricing errors while allowing normal variation from rounding, currency conversion, or minor contractual adjustments.
• High-value or regulated purchases: 0-1% is appropriate when individual line items represent significant spend or when regulatory compliance demands tight controls. Capital equipment, pharmaceutical ingredients, and government contract purchases often fall here.
• Combined thresholds: Some organizations apply both a percentage and an absolute cap, using whichever is lower. For example, 3% OR $500, whichever is lower. This prevents a small percentage from masking a large absolute overpayment on high-value line items.

Quantity Tolerance

Quantity tolerance defines the acceptable difference between the quantity invoiced and the quantity received (or the quantity ordered, in 2-way matching).

Common ranges for quantity tolerance:

• Standard goods: 0-5% covers most scenarios where items are counted in discrete units. A 2% threshold is a reasonable starting point for general office supplies, components, and packaged goods.
• Bulk materials: Industries dealing with commodities sold by weight, volume, or length often accept 5-10% variance. Raw materials like steel, chemicals, or agricultural products are inherently difficult to deliver in exact quantities.
• Serialized or regulated items: Zero tolerance is standard when every unit must be individually tracked. Pharmaceuticals, firearms, medical devices, and other items subject to serial number tracking should match exactly.

Deciding on the Right Threshold Levels

Four factors should drive your threshold decisions:

Transaction value. Higher-value transactions justify tighter thresholds. A 5% variance on a $200 purchase order is $10. The same 5% on a $200,000 purchase order is $10,000. Scale your controls to the financial exposure.

Industry and regulatory requirements. Regulated industries (healthcare, defense, financial services) often mandate tighter controls as part of compliance frameworks. Check whether your industry has specific audit requirements that constrain how much variance you can accept.

Supplier relationship and track record. An established supplier with years of accurate invoicing may warrant a wider threshold than a new vendor with no history. Some organizations maintain supplier-specific tolerance profiles, tightening thresholds for vendors with a pattern of billing discrepancies.

Operational cost of exception handling. Every exception that triggers manual review costs time. If your AP team spends hours each week resolving exceptions that turn out to be immaterial rounding differences, your thresholds are too tight. Calculate the average cost to resolve an exception (staff time, processing delay) and weigh that against the risk of the variances you would be letting through.

The Two Failure Modes

Getting tolerance wrong creates problems in either direction.

Thresholds set too loose allow significant overpayments to pass through undetected. A 10% price tolerance on a high-volume supply contract could mean thousands of dollars in excess payments accumulating over months before anyone notices. Loose thresholds effectively disable the control that matching is supposed to provide.

Thresholds set too tight generate a flood of false-positive exceptions. When the AP team must manually review dozens of immaterial variances every day, the process slows to a crawl. Staff begin rubber-stamping approvals to clear the queue, which defeats the purpose of the control entirely. Worse, genuine problems get buried in the noise.

The goal is a threshold calibration where exceptions are infrequent enough to receive genuine attention and significant enough to justify the review effort. Periodic analysis of your exception data, looking at how many exceptions are resolved without any change versus how many reveal actual errors, tells you whether your thresholds need adjustment.

---

Common Invoice Matching Failures and How to Resolve Them

Matching exceptions are a normal part of any invoice matching process. They do not mean the process is broken. They mean the system is doing its job by catching discrepancies before payment goes out. The goal is not to eliminate exceptions entirely but to diagnose them quickly and resolve them efficiently.

Here are the most common failure types, what causes them, and how to fix them.

Price Discrepancy

A price discrepancy occurs when the unit price on the invoice differs from the unit price on the purchase order.

Common causes:

• The supplier increased their price, but the PO was never updated to reflect the new rate.
• Currency conversion differences created a gap between the PO amount and the invoiced amount.
• A volume discount was negotiated but not applied to the invoice or not recorded on the PO.

Resolution: Pull the original PO terms and compare them against the supplier's current price list or contract. If the price increase is legitimate and was agreed upon, issue a PO amendment to reflect the updated price and re-run the match. If the increase was not authorized, send the invoice back to the supplier with documentation of the agreed terms.

Quantity Discrepancy

A quantity discrepancy occurs when the invoiced quantity does not match the quantity recorded on the goods receipt.

Common causes:

• The supplier shipped a partial order, but the invoice covers the full PO quantity.
• Counting errors at the warehouse resulted in an inaccurate goods receipt.
• Goods were physically received but the receipt has not yet been logged in the system.

Resolution: Start by verifying the physical receipt against the warehouse records. Check whether any goods receipt entries are still pending or sitting in a queue. If the shipment was partial, contact the supplier to confirm the delivery schedule for the remaining items and request a corrected invoice that reflects only what was delivered.

Missing Documents

A missing document exception occurs when one of the required documents, whether the purchase order, goods receipt, or inspection report, is not available in the system at the time of matching.

Common causes:

• The purchase order was placed verbally or sent by email but never entered into the procurement system.
• Goods were received at the warehouse but the receipt was not logged.
• An inspection is still pending or the inspection report has not been uploaded.

Resolution: Trace the procurement chain from requisition to receipt. If a PO exists outside the system, enter it formally and re-match. If the goods receipt is missing, coordinate with the warehouse to log it. If the purchase was made without a PO entirely, flag it as a non-PO purchase and route it through an alternative approval workflow.

Duplicate Invoices

A duplicate invoice exception occurs when the same invoice is submitted more than once.

Common causes:

• The supplier sent the invoice by email and also by post, and both were entered separately.
• Different departments within the organization each submitted the same invoice for processing.

Resolution: Match on a combination of invoice number, vendor ID, and invoice date to detect duplicates before they enter the matching process. When a duplicate is caught, void the redundant entry and confirm that only one instance proceeds to matching and payment.

Reducing Exception Volume

Many matching failures originate not from procurement problems but from data entry mistakes made when invoices or purchase orders are first recorded. Transposed digits, miskeyed quantities, and inconsistent vendor names all create exceptions that would not exist if the source data were accurate. Improving invoice processing accuracy to reduce errors upstream directly reduces the frequency of matching exceptions downstream.

One of the most overlooked causes of matching failures is data quality. When the data itself is wrong or inconsistent before matching even begins, no amount of process refinement will prevent exceptions from piling up.

---

How Data Quality Affects Matching Accuracy

Matching algorithms compare data, not documents. If the data extracted from an invoice or entered into a purchase order is wrong, the match will fail regardless of whether the underlying documents actually agree. This is the single most overlooked cause of high exception rates in accounts payable, and understanding it changes how organizations diagnose and fix matching problems.

Consider a straightforward example. A vendor submits an invoice under the name "Acme Corp." The purchase order was created with the vendor listed as "ACME Corporation." A second invoice from the same vendor arrives with the name recorded as "Acme Corp" without the trailing period. These are the same vendor, but the matching system sees three different strings and flags each one as a mismatch. The same problem occurs with dates: an invoice dated "03/04/2025" could be interpreted as March 4 or April 3 depending on regional format conventions, producing a date mismatch against the PO even when both documents reference the same transaction.

These are not edge cases. They represent the most common categories of data quality issues that drive matching failures:

• Inconsistent vendor naming across documents. Abbreviations, punctuation differences, legal suffixes (LLC, Inc., Ltd.), and alternate trade names create string mismatches between invoices, purchase orders, and receiving reports.
• Manual transcription errors in PO entry. Transposed digits in unit prices, incorrect quantities, or misspelled item descriptions introduced during purchase order creation propagate downstream into every match attempt.
• OCR and scanning errors when digitizing paper invoices. Poor scan quality turns "1" into "l," drops decimal points, or misreads handwritten annotations, corrupting the extracted data before it ever reaches the matching engine.
• Date and number format inconsistencies. Different date formats (DD/MM/YYYY vs. MM/DD/YYYY), currency symbols, thousand separators (commas vs. periods), and unit-of-measure abbreviations create systematic mismatches across document sets.
• Missing or truncated PO reference numbers on invoices. When a vendor omits the PO number, includes only a partial reference, or uses their own internal order number instead, the system cannot link the invoice to the correct purchase order for comparison.

The problem compounds in PDF-heavy workflows where every invoice must be digitized before matching can begin. Most small and mid-market organizations still receive the majority of their invoices as PDFs or scanned paper rather than through EDI or structured electronic formats. In these environments, data extraction is not an optional preprocessing step; it is the first link in the matching chain, and errors introduced at that stage propagate into every downstream comparison. Extraction quality directly determines matching outcomes. Understanding how OCR and AI extract structured data from invoice documents is directly relevant here: when extraction technology standardizes vendor names, normalizes date formats, and accurately reads line-item details, the data entering the matching system is consistent and reliable. Exception rates drop not because the matching rules changed, but because the inputs improved.

This matters for anyone managing invoice processing workflows accountants use across client portfolios, where data quality issues multiply across dozens of vendors, document formats, and entry points. The compounding effect of small data inconsistencies across a high volume of transactions turns a manageable exception rate into a persistent bottleneck.

Organizations experiencing high matching exception rates should look upstream at their data capture processes before tightening tolerance thresholds or adding more manual reviewers. Adjusting matching rules to compensate for bad data creates a cycle of increasingly complex workarounds that never address the root cause. Often the matching logic itself is sound. The data feeding it is the problem.

The principle is direct: matching is only as good as the data it receives. Clean, standardized, accurately captured data at the point of entry is the foundation that every matching method, whether 2-way, 3-way, or 4-way, depends on to function as designed.

---

Building an Effective Matching Process

The concepts covered throughout this article, from matching types and tolerance thresholds to failure diagnosis and data quality, all connect to a single practical question: how do you build a process that actually works for your organization?

1. Classify your transactions by matching type. Use 2-way matching for services, subscriptions, and consulting fees where no physical delivery occurs. Use 3-way matching for physical goods, materials, and inventory purchases. Reserve 4-way matching for high-value or regulated purchases with genuine inspection processes.

2. Set appropriate tolerance thresholds. Start with established industry ranges (1 to 5 percent for price, 0 to 5 percent for quantity) and adjust based on transaction value, supplier track record, and your team's exception handling capacity.

3. Document your matching policy. Capture which matching type applies to each transaction category, your tolerance thresholds, who owns exception resolution at each level, and how discrepancies above certain dollar amounts escalate. This gives your AP team consistent daily guidance and provides auditors with clear evidence that your controls are intentional and consistently applied.

4. Address data quality upstream. When exceptions spike, the temptation is to relax matching rules or widen tolerances, but first look at what is feeding the process. Vendor name inconsistencies, transposed numbers, and duplicate entries typically cause more matching failures than overly strict criteria, so improving data capture accuracy often yields better results than loosening the rules.

5. Review and refine on a regular cycle. Establish a monthly cadence to examine your matching metrics. Track exception rates by supplier, by transaction type, and by matching level. When a specific supplier or category generates a disproportionate share of exceptions, investigate the root cause. The problem may be a matching rule that does not fit that transaction type, a tolerance threshold that needs adjustment, or a data quality issue that should be resolved at the source.

Invoice matching does not need to be complicated. Start with the matching type that fits each transaction category. Set tolerance thresholds that balance financial control with operational efficiency. Invest in data quality upstream rather than chasing exceptions downstream. The goal is not to eliminate every exception. It is to build a process that reliably catches genuine errors, whether overbilling, short shipments, or unauthorized charges, while allowing clean invoices to move through to payment without unnecessary delay.