B2B Payment Processing Companies

Why paying a business is a different problem from paying a consumer

A consumer payment is finished at the moment of authorization. Somebody taps a card, it approves or declines, and the entire commercial relationship it settles begins and ends inside that second. A business-to-business payment almost never behaves that way. It starts as a purchase order, becomes a delivery, becomes an invoice, sits in an approval queue, waits out terms two finance teams negotiated months earlier, and only then does money move. The transfer is the last step of the process, not the process.

Three features of that description have no consumer equivalent, and each generates a category of software.

• Net terms. When a supplier invoices net 30, it has extended unsecured credit for thirty days. Every accounts receivable ledger is a small loan book, and days sales outstanding is the interest rate on it.
• Approval workflow. The person who receives an invoice is rarely the person who may authorize it, and neither is the person who can release funds. Segregation of duties is a control requirement, which means the payment instruction has to survive a queue.
• Remittance data. A supplier receiving $84,320 needs to know which forty-one invoices it covers, at what discount, with which credit notes applied. Without that, cash has arrived and nothing has been collected.

The mistake most businesses make here is buying for the wrong half of the problem. A company frustrated by slow collections buys payables software, which does nothing for receivables. A company drowning in cash application buys a faster rail, when the rail was never the bottleneck — the missing remittance detail was.

Check, ACH, card and wire: what each rail actually buys

The United States still settles an enormous share of business payments on paper. Checks persist not out of nostalgia but because they are universally accepted, require no onboarding of the counterparty, and carry a remittance stub stapled to the money. Every electronic alternative has to beat that combination, and most beat only two parts of it.

Read that as a set of trades, not a ranking. Cards move cost from the payer to the supplier and give the payer float and rewards. Bank debit also lands cost on the supplier but hands it control of timing, which is why recurring-invoice businesses favour it. Instant rails remove settlement risk by removing recourse — the wrong trade when paying a supplier you have never dealt with, the right one when releasing a payout you have already reconciled.

Level 2 and Level 3 data, and why most suppliers never claim it

When a business pays by commercial card, the interchange the supplier absorbs is not a single number. Interchange — the portion of a card fee that goes to the card-issuing bank, set by the card network and not negotiable by the merchant, the processor or the acquirer — is a table of hundreds of programs, and commercial cards have programs that reward richer data.

The commercial consequence runs in both directions, which is why this argument keeps recurring in B2B. The buyer's card rewards are funded out of interchange; the supplier's acceptance cost is that same interchange. Enhanced data is the one lever that lowers the supplier's side without the buyer giving anything up.

Two facts decide whether this is worth an engineering sprint. First, enhanced data does nothing at all for a consumer card — not a smaller saving, no saving. It qualifies only commercial products: purchasing cards, corporate cards and government purchase cards. Second, the prize scales with card volume and average ticket. A distributor running several million dollars a year of purchasing-card receivables recovers a meaningful, recurring amount; a firm running a few thousand a month will never repay the certification testing.

The reason so few merchants capture it is mundane. A gateway advertising Level 3 support means only that its API accepts the fields; whether anything populates them depends on an ERP or billing system upstream that was usually never wired to pass line items down. Populated is not valid, either — placeholder customer codes, unit-of-measure values outside the accepted code set and line totals that do not reconcile all cause a silent downgrade. Nobody gets an alert; the transaction simply settles at a worse rate. Test with a real purchasing card on a real invoice and read the qualification on the settlement report rather than the feature page.

Virtual cards, and the argument they start

A virtual card is a single-use card number issued for one specific payment, generally with the amount, the supplier and the validity window locked down before it is released. Accounts payable platforms and commercial card issuers push them hard, and the reason is visible in the economics.

The buyer gets three things: extended float, because the payment settles on a card billing cycle rather than out of the bank account today; control, because a number that works once for one amount cannot be reused by a compromised supplier mailbox; and a rebate, because the issuer shares part of the interchange the supplier pays. Corpay built a large business on exactly this, and virtual card issuance sits in the standard payment mix at BILL and most of its competitors.

The supplier's side of the same transaction is less cheerful. An invoice that would have arrived as an ACH credit for a small flat fee now arrives as a card transaction costing a percentage of face value, frequently as a PDF that somebody has to key into a terminal. The honest description is that a virtual card program transfers margin from suppliers to buyers and pays for part of the buyer's AP software out of the supplier's gross profit.

The AP automation category, and what actually separates the vendors

Accounts payable automation is the label for software that captures supplier invoices, codes them to the general ledger, routes them for approval, executes the payment and writes the result back to the accounting system. The workflow half has largely commoditised. What differs between vendors is the payment half: which rails they reach, whose money they hold while a payment is in flight, and who earns the interest on it.

• BILL serves small and mid-sized US businesses and the accounting firms that run bill pay for them. It executes payments over ACH, virtual card and check, holds customer funds as custodian, and states plainly in its terms that customers do not receive interest on balances held with it.
• Tipalti is built for the opposite shape of payables: thousands of international suppliers, creators and partners, where collecting valid tax documentation and screening every payee against sanctions lists is as much of the job as sending money.
• Melio targets very small businesses, letting a payer fund a bill by card even when the supplier does not accept cards — the supplier receives an ACH transfer or a mailed check. It is now owned by Xero and embedded in several accounting products.
• Corpay sells to mid-market and enterprise, combining AP automation with commercial card issuing, cross-border payments and FX hedging. An FTC action filed in 2019 alleging hidden fees against its predecessor FLEETCOR remained pending as of mid-2026 — a reason to read the fee schedule line by line, not a reason to exclude the company.
• Cass Information Systems audits and pays freight and facility invoices for large shippers, disbursing through a bank it owns. Its published financials show interest on customer funds contributing a substantial share of net revenue.

That last point generalises. Whenever a provider holds your money between funding and delivery, somebody earns on the balance. Ask who, ask how many days the funds sit, and treat the answer as part of the price.

Receivables: the half most companies never automate

Almost every finance team buys payables software before receivables software, and almost every finance team has a larger problem on the receivables side. Collections still run on emailed PDFs, a remit-to address and hope.

There are three coherent approaches, suiting different businesses.

Pull the money on a mandate. Bank debit — Bacs and SEPA in Europe, ACH debit in the United States — lets a supplier collect on a schedule it controls, removing card expiry and decline as sources of churn. GoCardless is the specialist and reaches several domestic schemes through one integration. The trade is reversal risk: the UK Direct Debit Guarantee gives payers an immediate refund on request with no time limit, and SEPA Core allows an unconditional refund inside eight weeks. That profile is worse than cards, not better; businesses adopt bank debit for cost and control, not safety.

Accept cards properly. If customers want to pay invoices by card, then acceptance with working Level 2 and Level 3 transmission, an invoice-linked payment page and automatic write-back to the ledger is a legitimate decision. Stripe and Adyen both serve this pattern; the question to ask either is not whether cards are supported but whether enhanced data actually reaches the network on your transactions.

Build it. Platforms moving money on their users' behalf — lenders, insurers, marketplaces, property managers — need something below the application layer. Modern Treasury provides bank connectivity, initiation controls and a double-entry ledger across ACH, wires, RTP, FedNow and checks; Dwolla, now owned by NMI, offers a similar US bank-payments API with transfers performed by partner financial institutions. Neither is an AP tool and neither acquires cards.

Reconciliation is the deliverable, and remittance data is what makes it possible

Cash application — matching received funds to open invoices — is where the cost of a badly chosen rail actually lands, and it lands as headcount. A supplier receiving one ACH credit covering forty-one invoices, with nothing in the addenda beyond a truncated company name, reconstructs the allocation by hand.

The mechanics explain the failure. Corporate ACH credits carry either a single 80-character addendum record or, in the CTX format, a full EDI 820 remittance document. CTX solves the problem outright — but it requires the payer's bank to originate it, the receiving bank to deliver rather than strip it, and the supplier's ERP to parse it, three conditions that frequently fail in sequence. Real-time rails carry structured ISO 20022 messaging with room for remittance detail, and the same delivery question applies at the receiving bank.

The practical questions are narrow. Does structured remittance travel with the payment, or in a separate email? Can the supplier retrieve it without a portal login? Does the format match what its accounting system can ingest? A vendor that answers "we send a remittance advice" has told you almost nothing — a PDF emailed to a shared inbox is a remittance advice, and it is the version that requires a person.

What to check before signing with any B2B payments provider

The differences that matter in this category are rarely on the comparison page. These are the questions that expose them.

• Who holds the funds in transit, and who earns on them? Get it in writing, including how many days a typical payment sits.
• What can be held, and on what standard? Several major platforms reserve broad rights to delay, review or refuse a payment at their discretion. If a deadline-driven payment cannot tolerate a hold, that clause is a functional limitation.
• What is on the restricted-business list, and can it change unilaterally? Most of these agreements incorporate an acceptable use policy by reference and reserve the right to amend it by posting.
• Which rails, with which cutoffs? Same-day ACH windows, wire cutoffs, check mail dates, and whether RTP or FedNow is genuinely available rather than roadmapped.
• What happens on exit? Vendor master records, bank details, mandates and payment history should be exportable in a usable format. Direct debit mandates in particular are hard to move and easy to lose.

Then do the exercise that makes the decision obvious. Take one month of payables and one of receivables, count how many payments travelled each rail, and put a fully loaded cost against each — fees, float given up, and the hours spent keying, chasing and reconciling. Most companies discover the expensive rail is not the one with the visible fee.

Frequently asked questions