Should My SaaS Become a Payfac?

Every published answer to this question comes from someone selling one of the outcomes. Payfac-as-a-service vendors publish high registration-cost estimates. Payfac-infrastructure vendors publish low ones. Processors selling managed platforms land in between. This post has a position too: Flint sells embedded payments with no revenue share and no payfac product, so we make the same money whether you register or not. What we can do is put the attributed numbers side by side and show you the arithmetic.

The short version: below roughly $50M in annual processing volume, every named source that publishes a threshold agrees the answer is no. Above that, the answer depends on an incremental-margin calculation this post works through, and the two ends of the published threshold range ($50M and $750M) are best explained by what each publisher sells.

Last reviewed 2026-08-02. Every figure below links to the party that published it.

The four options#

Register as a payment facilitator. You become a card-network-registered third party agent through a sponsoring acquirer, sign your merchants' agreements yourself, touch settlement funds, and keep the whole spread. Everything below is about what that costs.

Payfac-as-a-service. A vendor holds the registration and underwrites your merchants; you white-label the experience and take a revenue share or a spread over buy rates. The vendors differ more than their marketing suggests; we compared seven of them on sourced facts in the payfac-as-a-service reference.

ISO or referral. You hand merchants to a processor for a smaller cut and almost no operational involvement.

Embedded payments without spread monetization. Payments live inside your product, your merchants pay the processor's published rates directly, and you make money on your software. This is what Flint sells, so weigh our framing accordingly.

What "registered payfac" actually means#

The cleanest definitions come from the card networks' own rulebooks. An ISO solicits and services merchants who contract directly with the acquirer, and under Mastercard's rules a Service Provider "must not have access to any account for funds due to a Merchant," a prohibition with a payfac-shaped exception (Mastercard Rules, Rule 7.3). A payment facilitator signs the merchant agreement itself and receives settlement on the merchant's behalf, and Visa's Core Rules make it "responsible and financially liable for each Transaction processed on behalf of the Sponsored Merchant" (Visa Core Rules). Signing the agreement and touching the money is the whole difference, and it is where the liability lives; receiving and distributing funds is also what can put state money-transmitter licensing on the table (Venable LLP), though federally most payfacs fit FinCEN's payment-processor exemption (FIN-2014-R009) and the binding AML program arrives contractually from the sponsor bank.

Two mechanical rules worth knowing before any cost math. Sub-merchants that pass $1M in annual volume on a network must sign direct agreements with the acquirer, so your largest merchants graduate out of your paper (Venable). And PCI DSS Level 1 attaches at 300,000 annual Visa transactions, not at the payfac designation itself; below that, self-assessment can suffice (Visa TPA FAQ).

The cost stack, attributed#

Network registration. Visa publishes $5,000 for registration and up to $5,000 per year for renewal, paid through your acquirer. Mastercard does not publish its fees; Infinicept reports a $5,200 initial bundle, and Stripe's payfac guide uses $5,000 per network with $10,000+ annual renewals. Registration is the cheap part.

Sponsor bank. You cannot register yourself; only an acquirer can register you, after a risk and financial review of your underwriting, monitoring, and funds-handling program. Reserves are real money: Finix describes sponsor-bank reserves of one to two weeks of processing volume, which at $5M a month is $1.25M to $2.5M parked.

PCI DSS Level 1. Published assessment costs range from Thoropass's $30,000 to $200,000 for a full report on compliance, through Fiska's "at least $50,000," to Stripe's $50,000 to $500,000 with $200,000+ per year ongoing validation.

People. Stripe's guide prices the merchant-management build at four FTEs minimum, a compliance program at two more, and $500,000+ of underwriting policy work; Stax publishes similar per-function figures, and Tilled publishes the mid-range: at least 3 to 5 full-time employees for the payments operation, including at least one dedicated risk, underwriting, and compliance hire at $75,000 or more. Infinicept argues the opposite: below a few hundred million in volume, one person may be enough to start. Note who sells what: Stripe sells the alternative, Tilled sells payfac-as-a-service, Infinicept sells payfac enablement. Visa itself only requires that "competent risk management personnel" exist before operations begin (Visa risk guide); the headcount is your problem to size.

Money-transmitter licensing. Only if you control the funds flow, which is the point of registering. Stripe estimates $450,000 minimum over three years for 50 US states.

The totals, spread by publisher.

Every row is a party with a position, which is why the spread runs 14x. Two footnotes before you average anything: the Stax and Stripe tables share near-identical line items, so treat them as one estimate lineage rather than two independent ones, and Finix's published numbers have moved with its business model, from "$3M-$5M and two to three years" when it sold build-avoidance to "6 to 12 months" once it held a registration to sell access to. The defensible reading is not a midpoint. It is that the floor is low only if you already have payments operators on staff, and the celebrated totals include licensing you only need if you take custody of funds.

The revenue side#

What registration buys you is the spread between what your merchants pay and what processing costs you. The only vendor-published tiering we found is Worldpay's own ROI material: referral deals at 0 to 20 basis points of volume, integrated payments up to 40, payfac-as-a-service at 60 to 100, and full registered payfacs at 100 to 120. Treat these as one conflicted vendor's published ranges, not market data; they are, however, the ranges that vendor uses to sell the middle tier.

The arithmetic that matters is incremental: registration only earns you the step from the payfac-as-a-service tier to the full-payfac tier, roughly 20 to 40 basis points on Worldpay's numbers, because the middle tiers are available without registering.

Worked examples#

Using Worldpay's published ranges for revenue and the attributed cost spread above, with the arithmetic ours:

This is why the published thresholds disagree. Swipesum, Fiska, and Finix all put the consider-it line around $50M in annual volume; Finix sells the infrastructure you would register on. a16z, the closest thing to a party without a payfac product in this list, put it at roughly $50M to $100M in gross merchandise value in 2020 and revised it to $100M+ by 2023. Worldpay suggests registration below roughly $750M may not be justified, Rainforest says the math often fails even past $1B, and Tilled goes as far as $2 billion; all three sell the managed tier you would stay on. Each threshold marks where its publisher's product wins. The incremental math above puts the crossover in between, and it depends on your cost floor: a team that already runs risk operations crosses over far earlier than one hiring from zero.

You are not at the threshold yet if#

• Your annual processing volume is under $50M. No named source, on any side, says registration pays here.
• You cannot name the people who will own underwriting, risk monitoring, and chargeback operations as their job.
• Parking one to two weeks of your processing volume as a bank reserve would strain the balance sheet.
• The spread you would keep is under 20 basis points better than what a payfac-as-a-service deal already pays you, after the fees in your Schedule A.
• You do not want to be in the funds flow. Receiving and distributing settlement is what creates the transaction liability in the network rules and what puts state licensing on the table.

If none of those disqualify you, you are in genuine decision territory, and the comparison reference covers who to talk to, including the two vendors that sell a graduation path to your own registration.

Where Flint fits#

Flint cannot sell you any version of this. There is no revenue-share surface in the API, no fee-split field, and no payfac program; each business on your platform becomes its own Flint merchant, pays published rates directly, and your platform is never in the funds flow. That is the right shape when payments revenue is not your business model: you get merchant onboarding as an endpoint and commerce embedded in your product, and you monetize your software. If Stripe has declined your business, Flint runs on Stripe rails and cannot help either. And if your goal is the spread, Flint has nothing to sell you there: take the payfac-as-a-service tier now, and revisit registration when the incremental math above clears your own cost floor for two consecutive years.

Corrections: support@withflintpay.com. Vendor figures were retrieved 2026-08-02; treat every number here as perishable and follow the links before deciding.