Payment Processing · Industry

Ask most business owners who they process with and you get a brand name. Ask who holds their merchant agreement and the answer is usually the same brand name, and it is usually wrong. There are three distinct arrangements behind the checkout button, they are defined in the card network rules rather than in marketing copy, and which one you are in decides who is liable for your sales tax, who eats a chargeback, whose name a customer sees on their statement, and how difficult it will be to leave.
In the traditional model, an acquiring bank holds a merchant agreement directly with your business and issues you your own merchant identification number. An independent sales organisation may have sold you the account and may service it, but the agreement runs to the acquirer. You are the merchant.
In the payment facilitator model, a single company holds the acquiring relationship and you accept cards underneath it as a sponsored merchant, sometimes called a sub-merchant. Onboarding takes minutes instead of days because you are being added to somebody else's arrangement rather than underwritten into your own.
In the merchant of record model, you are not the merchant at all. Another company buys your product and resells it to your customer as the legal seller. It is the party to the sale, and your relationship with it is that of a supplier.
These are not loose commercial descriptions. The Visa Core Rules and Visa Product and Service Rules, in the edition effective 18 April 2026, set the structure out directly. Rule 1.5.2.1 states that an acquirer must have a merchant agreement with each of its merchants, and that "a Payment Facilitator must have a Merchant Agreement with each of its Sponsored Merchants." Your agreement, in the payfac model, is with the payfac.
The liability chain is stated just as plainly. Rule 5.3.1.2, headed Acquirer Liability for Payment Facilitators and Sponsored Merchants, provides that "a Sponsored Merchant will be treated as a Merchant of its Payment Facilitator's Acquirer," and that "the acts and omissions caused by a Sponsored Merchant will be treated as those of the Payment Facilitator and those caused by a Payment Facilitator or a Sponsored Merchant as those of the Acquirer." Read that from the merchant's side: your chargeback ratio, your fraud and your risk programme breaches are the payfac's problem, which is exactly why a payfac can close your account on its own judgement and without much explanation.
There is also a ceiling on the arrangement. Rule 5.3.1.4 requires an acquirer that contracts with a payment facilitator to enter into a direct merchant agreement with any sponsored merchant that exceeds one million US dollars in annual transaction volume: before processing at all if the merchant is new to the payfac, or within two years of crossing the threshold for an existing relationship. There are carve-outs, including for relationships more than two years old with regular reporting and continued acquirer oversight, and for a list of merchant category codes covering utilities, healthcare, schools, rent and court payments. The principle survives the exceptions: the sub-merchant model is designed for small merchants, and growing out of it is a contemplated event, not a failure.
The third model sits outside the acquiring structure entirely, because the merchant of record is the merchant. Paddle's own terms are unambiguous about what you are agreeing to: "You appoint Paddle as your non-exclusive reseller of the Product across all territories," and as merchant of record Paddle "reserves the right to set the price or licence fee at which the Product is offered for sale to Buyers." It is the seller. You are the supplier.
That has visible consequences. Paddle's own documentation states that a card charge appears on the customer's statement as PADDLE.NET followed by a descriptor of up to ten characters, and PayPal payments always show as PAYPAL *PADDLE.NET with no customisation available. FastSpring describes a merchant of record as the entity that sells to the customer and is the responsible party for the transaction, and lists tax calculation and remittance, chargebacks, fraud and regulatory compliance among what it takes on. Xsolla runs the same structure for game publishers.
The prices are not close, and only some of them are published. Stripe, operating as a payment facilitator, publishes 2.9% plus 30 cents for domestic online cards and 2.7% plus 5 cents in person, with 1.5% added for international cards and about 1% for currency conversion. Paddle publishes one number for the merchant of record service, 5% plus 50 cents per checkout transaction, with no monthly fee and tax handling, subscription billing, fraud tooling and buyer support included.
Roughly double, then, for the same card on the same rails. FastSpring and Xsolla publish no rate card at all; our reviews of both record third-party estimates in the same territory or higher, and present them as reported rather than established, because neither company confirms a number publicly. If you are comparing, the only figure that matters is a written quote for your own mix of markets and payment methods.
Two things, mainly, and both are real. The first is tax. Selling software or digital services across borders creates registration and remittance obligations in a growing list of jurisdictions, and since the Supreme Court's 2018 Wayfair decision most US states impose economic nexus thresholds that a modest online business can cross without any physical presence. A merchant of record becomes the legal seller and takes those obligations on. Paddle's terms describe exactly this: the reseller structure is what allows it to handle sales tax collection, reporting and remittance.
The second is underwriting. Because the merchant of record is the party to the transaction, you are not being underwritten as a merchant at all, which is why this model is common in software, games and other categories where conventional acquirers are cautious.
Read the liability language rather than the marketing summary, though. Paddle's help centre sets out what actually happens: because Paddle is the merchant of record the dispute is raised against Paddle, and Paddle then deducts both the transaction amount and the fee from your seller balance. The chargeback fee is 20 USD, GBP or EUR, or 40 CAD or AUD. If Paddle wins the dispute it returns the recovered transaction amount to your balance, but the fee is not refunded, because the payment provider's own processing fee is not refundable to Paddle. A pre-chargeback alert, where Paddle refunds proactively to stop a dispute being filed, still carries the fee. The master services agreement backs this up with a set-off right covering "liability for refunds and Chargebacks and associated costs and fees," and provides that where a balance is insufficient the supplier pays the shortfall immediately. The merchant of record absorbs the operational work of fighting disputes. It does not absorb the loss.
The customer relationship is the substantive cost, and it is not obvious on day one. Under a merchant of record, subscriptions, stored card credentials and billing history sit with the reseller because it is the party to those contracts. Under Paddle's terms the seller issues the refund and the supplier does not invoice the buyer directly. Leaving is therefore a migration project rather than a cancellation: our reviews of Paddle, FastSpring and Xsolla all make the same point independently, that the exit cost is structural rather than a fee, and that what is portable should be established in writing before you integrate rather than after.
The smaller terms deserve the same attention, because on a merchant of record they are set by the reseller rather than negotiated by you. Paddle's agreement, last updated in October 2025, applies a foreign exchange margin of 2% on major currencies, 2.5% on a short list of others and 3% on everything else where a sale needs converting into your payout currency, pays out on or before the 15th of the following month once your accrued balance passes 100 dollars, euros or pounds, and reserves the right to apply an alternative discount structure in place of the headline rate at its sole discretion.
The payfac model has a quieter version of the same problem. Because the arrangement is not underwritten to you, funds can be held or accounts closed on the facilitator's risk judgement, which is the mechanism behind most of the frozen-payout stories in this industry. We covered how that works, and what the contracts actually say, in our piece on reserves and holds.
For a sense of what published pricing looks like at the transparent end of this market, our review of Helcim sets out a complete interchange-plus rate card, and our review of Tilled covers a payfac-as-a-service provider that publishes its platform plans and revenue shares outright, which is rare enough in this industry to be worth using as a benchmark when someone quotes you a number over the phone.
None of these three models is the right answer in general. But you should know which one you are in, because the answer to "who do I chase when this goes wrong" is different in each, and the worst time to work it out is the week the money stops arriving.