Visa's Level 2 Sunset: The Interchange Math Rewrite CFOs Missed

Most finance teams treat card interchange the way they treat FX spreads. A fixed cost of doing business, published in a schedule somewhere, negotiated once, then ignored. The rebate lands quarterly. The line item on the P&L looks the same each period. Nobody asks how the sausage gets made.

That worked when Level 2 was the ceiling for commercial data enrichment. It stops working in April 2026, when Visa officially retires its Level 2 interchange program and replaces it with the Commercial Enhanced Data Program (Source: Visa, https://usa.visa.com/partner-with-us/payment-technology/commercial-enhanced-data-program.html). And the CFOs who read that as "a rate update" are the ones who will be surprised when their rebate curve bends the wrong way next fiscal year.

The rate change is the visible part. The invisible part, the part that will actually move revenue, is that your ERP has been quietly qualifying transactions for reduced interchange on templated data that a stricter enrichment standard is going to reject.

Reframe: Interchange Is a Data Contract, Not a Pricing Table

Here is the frame worth internalizing. Every commercial card transaction is a data contract between your ERP, the acquirer, and the network. You send fields. The network prices the transaction based on the completeness and quality of those fields. Level 2 was a loose contract, tax amount, customer code, a few basics. Level 3 was tighter, requiring line-item detail, unit costs, product codes, freight.

The Commercial Enhanced Data Program collapses that hierarchy and rewrites the enrichment standard. Starting in April 2026, Level 2 goes away, and effective January 24, 2026, Visa is already increasing rates for Product 3 business credits submitted with Level 2 data (Source: Visa, https://usa.visa.com/partner-with-us/payment-technology/commercial-enhanced-data-program.html). Translation: the runway is short, and the penalty for staying on the old contract has already started.

Once you see interchange as a data contract, a whole set of AP practices look different. That "customer code" field your team maps to a generic string? It's an input to a pricing engine. The tax field that defaults to zero on services invoices? It's telling the network something about the transaction. The line-item description that reads "Professional Services - See Attached"? That's the exact kind of templated data the new program is designed to catch.

The Templated Data Problem Nobody Audited

Here is the part that will make some readers uncomfortable. In practice, large volumes of commercial card transactions have been qualifying for enhanced interchange on data that would not survive a real audit.

The pattern shows up like this. An AP team implements a virtual card program. The middleware between the ERP and the card platform gets configured once, usually during onboarding, usually by someone who no longer works there. Default values get assigned to required fields. Tax amount defaults to a computed percentage. Customer code defaults to the invoice number. Line-item detail defaults to whatever text is in the invoice description field, truncated to fit the network's character limit.

None of that is fraud. It's just how ERP-to-card integrations get built when the goal is "make the payment go through" instead of "make the payment go through with defensible enrichment." And for years it has worked, because Level 2 checks were largely format checks. Fields present, fields in valid format, transaction qualifies.

The Commercial Enhanced Data Program is not going to be a format check. Forrester's 2026 analysis notes that AI adoption in AP is no longer limited to data extraction or coding assistance, with vendors now deploying agentic capabilities for exception handling, fraud detection, and supplier management (Source: Forrester, https://www.forrester.com/blogs/). The same pattern applies on the network side. Enrichment validation is moving from "did you send the field" to "does the field make semantic sense given the merchant category, the amount, and the counterparty."

Templated data does not make semantic sense. It survives a format check. It fails a reasonableness check.

Why This Hits Rebate Revenue Twice

CFOs running mature virtual card programs need to understand this hits the P&L in two places, not one.

The first hit is prospective. Transactions that would have qualified for reduced interchange under Level 2 will not qualify under the enhanced program if the enrichment data cannot pass validation. Your effective rebate rate compresses. That is the hit everyone is modeling.

The second hit is the one nobody is modeling. Retroactive disqualification. When network-side validation identifies transactions that qualified on templated data, the remediation path is a chargeback or a downgrade of the transaction after the fact. The rebate you already booked, the rebate your treasury team already forecast into working capital, becomes a claw-back.

If you run a program of any real volume, that is not a rounding error. That is a multi-quarter revenue restatement risk sitting inside a line item nobody in finance thinks of as risky.

This is the part of the story that gets missed when people read the Visa announcement as a rate change. The rate change is a headline. The audit posture change is the actual event.

The Operational Playbook Before April 2026

If you run a commercial card program, here is the sequenced work that needs to happen between now and April.

Step one: audit your enrichment pipeline as a distinct system. Not the card program. Not the ERP. The middleware and mapping logic that sits between them. Identify every field that gets populated with a default, a computed value, or a truncation. That inventory is your risk surface.

Step two: sample transactions by merchant category and re-score them against a semantic standard. Pick a quarter of transactions. For each one, ask: if a human auditor read the line-item data and the merchant category, would the enrichment be defensible? A services invoice with line items reading "Consulting" and a tax amount of zero may be legitimate. It may also be templated. You need to know which.

Step three: separate your card program economics from your card program orchestration. This is where most AP leaders get stuck. The team that negotiates the card program is usually not the team that owns the ERP integration. The team that owns the ERP integration is not the team that reads Visa's operating bulletins. Nobody owns the enrichment contract end to end. Someone has to.

Step four: model the rebate curve under two scenarios. Best case, all your current volume qualifies under the new program. Worst case, transactions dependent on templated fields downgrade. The gap between those two numbers is your exposure. If it's material, it belongs in your 2026 planning conversation with the audit committee, not buried in an AP operations review.

Step five: pressure test your card partner. Ask them directly how they are preparing for the Commercial Enhanced Data Program. Ask them how their enrichment engine handles services categories, indirect spend, and non-PO invoices. Ask them what happens to a transaction that fails semantic validation post-settlement. If the answers are vague, that is your answer.

Where Finexio Sits in This

The reason Finexio has spent over ten years and more than $75M in investment building an orchestration layer, not just a card issuance product, is that this exact scenario is where orchestration matters and pure issuance does not.

The three-party model we run, Finexio as the orchestrator, J.P. Morgan Chase as the issuing bank, Mastercard and Visa as the networks, means enrichment data quality is owned by an operator whose economics depend on the transaction qualifying correctly the first time. Not by a bank whose incentive stops at issuance. Not by a network whose incentive is enforcement.

When Visa moves the enrichment standard, orchestrators absorb the operational lift of remapping fields, validating semantics, and reconciling downgrades. That is what AP payments as a service is supposed to mean. Not "we send the payment." It means "we own the data contract with the network on your behalf."

For programs already running on templated data, the migration path is not glamorous. It is data hygiene, supplier record cleanup, and rebuilding the enrichment mapping with the new standard in mind. It is exactly the kind of work most in-house teams do not have the bandwidth to lead in the same quarter they are closing the books.

And on the fraud side, Finexio Shield's $2M guarantee is worth mentioning here because the same enrichment infrastructure that survives an audit also produces the transaction signals that make fraud detection work. Clean data is not just a rebate story. It is a control story.

FAQ

Does the Commercial Enhanced Data Program affect virtual card and physical card equally?

Both are in scope, but virtual card programs are more exposed to the templated data problem because they typically process higher volumes of services and indirect spend, where enrichment fields default more often than in physical card use cases like T&E.

If our rebate is already forecast into 2026 revenue, what should we tell the audit committee?

At minimum, flag the enrichment standard change as a program risk with a range of outcomes rather than a point estimate. The honest answer for most programs is that the downside is unquantified until an enrichment audit is done. Doing that audit is the risk disclosure.

Is this only a Visa issue, or is Mastercard heading the same direction?

Mastercard is moving in a parallel direction with its own commercial data initiatives. Assume the trajectory is universal. Enrichment quality is becoming the underwriting question for commercial card interchange across networks.

The Work That Actually Matters

The Visa Level 2 sunset is not a rate announcement. It is a shift in what counts as a qualifying transaction. Programs built on templated data will discover that quietly, through compressed rebates and post-settlement downgrades, over the course of 2026 and 2027.

Programs built on real enrichment will keep their economics. Some will improve them.

If you want to know which one your program is before April tells you, Book a Consultation with the Finexio team. We will pressure test your enrichment pipeline against the new standard and tell you what the exposure looks like in your specific volume mix.

Sources

- Visa Commercial Enhanced Data Program: https://usa.visa.com/partner-with-us/payment-technology/commercial-enhanced-data-program.html - Forrester AP Automation Analysis 2026: https://www.forrester.com/blogs/