Best Payment Platforms for Your Business (June 2026)

The founder's full review of payment processors, merchants of record, and the agentic commerce shift, as of June 2026

US merchants paid a record $187.2 billion in card processing fees in 2024, up roughly 70% since the pandemic, which works out to about $1,200 a year passed on to the average American family - Merchants Payments Coalition. For a small business, payments are usually the third largest cost line after people and goods, and unlike rent, this one silently scales with every dollar of success.

Here is the problem: the platform you pick in week one quietly decides your margins, your tax liability, and even whether you can access your own money. The differences are not cosmetic. A SaaS founder selling globally on a do-it-yourself processor stack commonly ends up at a 4.5-6.5% effective rate once billing, tax, fraud, and cross-border surcharges stack on top of the advertised 2.9% - Dodo Payments analysis. A coffee shop on the wrong plan pays a 14% higher online rate than it did in December 2025 because of one quiet repricing. And in the worst documented case of the past year, more than 3,200 organizations lost access to $29 million when a payment intermediary collapsed - Oakland Voices.

This guide is a full review of the payment platforms that matter for founders and small businesses in June 2026. It covers exact verified pricing for 16 platforms (every number was checked against the official pricing page this week), the merchant of record question that decides whether you ever think about VAT, the account freeze risk nobody prices in, and the agentic commerce shift that started changing how customers even reach your checkout. If you are at the very beginning of this journey, our guide to starting a company in 2026 covers the steps that come before payments; this one goes deep on the money layer itself.

Contents

1. How to Think About Payment Platforms (First Principles)
2. The Real Cost of Taking a Payment
3. The Direct Processors: Stripe, PayPal, Square, Helcim, Mollie, Airwallex
4. The Merchant of Record Route: Paddle, Lemon Squeezy, Polar, FastSpring, Dodo, Creem
5. Bank Debit and B2B Payments: GoCardless and ACH
6. Commerce Platforms and the Enterprise Tier: Shopify, Adyen, Checkout.com
7. Agentic Commerce and Stablecoins: What Is Actually Real in June 2026
8. Account Freezes and Holds: The Risk Nobody Prices In
9. The Decision Framework and Integration Playbook
10. Future Outlook: Where Payments Go From Here

The Master Comparison: All 16 Platforms Scored

Before the detailed profiles, here is the full field side by side. Four criteria, weighted for what a founder or small business actually experiences: Total Cost (30%) is the realistic all-in rate including add-ons and surcharges, not the headline number. Speed to Live (25%) is how fast a real business takes its first payment, including underwriting friction. Capabilities (25%) covers subscriptions, tax handling, fraud tooling, invoicing, and agentic readiness. Trust & Cash Flow (20%) weighs documented hold and freeze behavior, payout speed, and counterparty maturity. Each cell shows the score and the reason for it.

Total Cost (30%) weighs the realistic effective rate for a typical small business on that platform, including the add-ons most businesses end up using. Speed to Live (25%) measures days from signup to first real payment, which is dominated by underwriting models, not APIs. Capabilities (25%) rewards platforms that replace other software you would otherwise buy. Trust & Cash Flow (20%) is scored from documented hold policies, review-site patterns, regulatory actions, and company maturity. Scores reflect the June 2026 state of each platform and the use case it is actually built for, which is why an excellent enterprise processor like Adyen can rank mid-table in a guide for founders.

1. How to Think About Payment Platforms (First Principles)

Strip away the marketing and every payment platform answers the same three questions: who is the legal seller, who carries the risk, and who owns the checkout. Everything else (pricing, features, payout schedules) follows from those answers. The reason the market looks confusing is that 16 well-funded companies give 16 different answers, and each answer is correct for a different kind of business. Start from the structure and the right choice usually becomes obvious within minutes.

The first structural split is between a payment service provider (PSP) and a merchant of record (MoR). With a PSP like Stripe, Square, or Mollie, you are the legal seller: the processor moves money, but sales tax registration, VAT remittance, refund policy, and chargeback liability are yours. With a merchant of record like Paddle, Polar, or Creem, the platform itself is the legal seller of your product: it charges the customer, collects and remits tax in every jurisdiction, eats the compliance burden, and pays you out the net. The MoR charges roughly 2-3 percentage points more per transaction for taking on that liability - Paddle. That premium is the single most important pricing decision in this guide, and we dedicate section 4 to when it is worth paying.

The second structural split is underwriting philosophy, and it explains almost every horror story you will read in founder forums. Aggregators like Square and Stripe approve you in minutes by skipping upfront underwriting, then manage risk afterward with automated holds, rolling reserves, and sudden terminations. Traditional-style acquirers like Adyen and Checkout.com underwrite you for weeks before a single payment, then rarely surprise you afterward. Fast onboarding and payout certainty are two ends of the same trade, and no platform escapes it. Understanding this before you sign up prevents the shock that hits merchants who discover the trade-off only when their first sales spike triggers a review, a pattern we document with cases in section 8.

The third split is newer: who brings the customer to checkout. Until 2025 this question did not belong in a payments guide. Then AI assistants started referring real traffic: agent-referred visits to US retail sites grew 393% year over year in Q1 2026 and now convert better than regular traffic - TechCrunch. Platforms now compete on whether your products are discoverable and purchasable by AI agents, and the major processors have spent eighteen months wiring themselves into agent protocols. Section 7 separates what is real from what is a press release.

2. The Real Cost of Taking a Payment

Every card payment you accept funnels through the same cost pipeline, and knowing its anatomy is the difference between comparing platforms and being fooled by them. The bottom layer is interchange: the fee set by card networks and paid to the customer's bank, averaging 2.35% on US credit transactions in 2024 - Merchants Payments Coalition. On top sit scheme fees (the network's cut) and finally the processor's markup. Flat-rate platforms like Stripe and Square blend all three into one simple number and pocket the difference when your customers use cheap cards. Interchange-plus platforms like Helcim and Adyen pass interchange through at cost and charge a visible markup, which is cheaper at volume but harder to predict.

The trap for founders is the gap between headline and effective rate. Stripe's famous 2.9% + 30¢ is real, but it is the price of bare card processing only. Add Stripe Billing for subscriptions (0.7% of volume), Stripe Tax (0.5% in registered jurisdictions), +1.5% for international cards, and +1% for currency conversion, and a globally selling SaaS realistically lands between 4.5% and 6.5% all-in - Swipesum. None of this is hidden, exactly. It is just never in the comparison table you read before signing up. The same dynamic plays out everywhere: PayPal's branded checkout is 3.49% + 49¢ before a 1.5% cross-border surcharge and a 3-4% currency spread, and Lemon Squeezy's flat 5% quietly becomes 12%+ on a small international subscription once surcharges stack.

Fee changes also only ever move in one direction, and 2025-2026 provided three clean examples. Square raised its free-plan online rate from 2.9% + 30¢ to 3.3% + 30¢ on January 13, 2026, a 14% increase for its smallest merchants - Beacon Payments. Stripe Billing consolidated its 0.5% starter tier into a flat 0.7%, a 40% increase for early subscribers when the grandfather rate expired in June 2025 - Stripe support. Polar repriced its free tier from 4% + 40¢ to 5% + 50¢ in May 2026. None of these required your consent, which is worth internalizing: you are not buying a price, you are buying a pricing philosophy.

It helps to watch one real transaction travel through the stack. Take a $25 monthly subscription sold by a US SaaS through Stripe. A domestic customer costs $0.725 in percentage fee plus the $0.30 fixed fee, plus $0.175 for Billing and $0.125 for Tax: about $1.33, or 5.3%. The same subscription bought by a German customer on a German card adds $0.375 for the international surcharge and $0.25 for currency conversion: about $1.95, or 7.8% of the sale. Notice two things in that math. The fixed fee alone is 1.2% of a $25 ticket, which is why low-priced products bleed disproportionately on every platform with a fixed component. And the blended rate of a business with a typical 70/30 domestic-international mix lands squarely in the 5-6% range, double the number on the pricing page. Run this exercise with your own average ticket before you choose anything; it takes ten minutes and reorders the comparison table more often than any feature list.

There is also a compliance cost that never appears on pricing pages: PCI scope. If you use a hosted checkout, a payment link, or a redirect (Stripe Checkout, Paddle's overlay, Square Online), you stay in SAQ A territory, the lightest self-assessment with roughly 30 requirements, many of which collapse if processing is fully outsourced - Secureframe. Build your own card form against a raw API and you slide toward SAQ A-EP, which carries 140-190 requirements including vulnerability scanning and penetration testing. The March 2025 PCI DSS v4.0.1 update tightened eligibility further: merchants embedding payment iframes must now confirm their site is not susceptible to script attacks, either through their own controls or written confirmation from the processor - PCI Security Standards Council. The practical translation for a small team: use the hosted checkout. The conversion difference is now negligible, and the compliance difference is a hundred-plus controls you do not have staff to own.

Chargebacks deserve their own line in your cost model because 2025 made them more expensive. Stripe charges $15 per dispute received, plus an additional $15 to contest one (refunded only if you win), so losing a contested dispute now costs $30 in fees on top of the lost sale - Signifyd. Visa added new tiered dispute fees in April 2025, and merchants whose dispute rates cross network thresholds enter monitoring programs with monthly fines. A useful rule from the data: a chargeback rate approaching 1% is not just a cost problem, it is the leading trigger for the account freezes covered in section 8. The cheapest chargeback is the one your clear refund policy prevented.

3. The Direct Processors: Stripe, PayPal, Square, Helcim, Mollie, Airwallex

Direct processors are the default starting point because they are cheap, fast to set up, and leave you in full control of the customer relationship. The trade is that you remain the legal seller, which means tax compliance, fraud strategy, and dispute handling stay on your desk. For a US-centric business, or any business with simple tax exposure, that trade is overwhelmingly worth it: you keep 2 to 3 points of margin that a merchant of record would absorb.

The chart below shows the published online card rates side by side. Note how tightly the headline numbers cluster: the real differences between these six platforms live in the add-ons, the international surcharges, and the underwriting behavior, not in the headline rate.

3.1 Stripe: the default, and still the deepest stack

Stripe remains the platform every other platform is compared against, and in 2026 the comparison got harder to win. The core price is unchanged at 2.9% + 30¢ per US online card transaction with no monthly fee - Stripe pricing. Around that core sits the broadest product surface in the industry: Stripe Billing for subscriptions at 0.7% of volume, Stripe Tax at 0.5% per transaction in jurisdictions where you are registered, Radar fraud screening included on standard pricing, ACH at 0.8% capped at $5, and Instant Payouts at 1.5%. At Stripe Sessions 2026 in April, the company announced 288 launches in one keynote, including the expanded Agentic Commerce Suite with Meta and Google and an agent wallet for Link, its 250-million-user consumer wallet - Stripe newsroom.

The 2026 twist is that Stripe now sells both sides of the merchant of record debate. Stripe Managed Payments, built by the Lemon Squeezy team after the 2024 acquisition, makes Stripe the legal seller and handles indirect tax in 80+ countries for +3.5% on top of standard processing, roughly 6.4% + 30¢ all-in on a US card - Stripe support. It reached general availability in 39 countries on April 22, 2026 and can be enabled on a payment link. For a non-technical founder the practical path is simpler than the product list suggests: create a Payment Link from the dashboard and you can take a real payment the same day, then graduate to hosted Checkout and plugins as you grow.

Pricing - Stripe:

Two operational details matter more than they look. First, geography changes the price: Stripe accounts domiciled in the EU pay just 1.5% + €0.25 on standard European consumer cards thanks to regulated interchange, so a European entity processing European customers pays nearly half the US rate for the same product - Stripe Spain pricing. Second, cash flow has a calendar: a new account's first payout takes roughly a week, the standard US schedule then rolls at about two business days, and Instant Payouts to a debit card cost 1.5% with a $0.50 minimum. Fraud screening through base Radar is included on standard pricing, with the Fraud Teams tier at $0.02 per screened transaction for rules, lists, and review queues. None of these change the buy decision alone, but together they explain why two businesses on identical Stripe pricing report very different experiences of it.

The honest weaknesses: fee stacking makes the real rate of a global business roughly double the headline, and Stripe's automated risk system produces well-documented freezes, including a 2025 case where $130,000 was frozen over a dispute-rate spike the hold itself worsened - Terms.law. Best for: almost everyone as a starting point, and specifically founders who want one stack that can grow from a payment link to agentic checkout without replatforming.

3.2 PayPal: conversion juice at a premium

PayPal is the only processor in this guide that is also a consumer brand, and that is precisely what you are paying for. Branded PayPal and Venmo buttons cost 3.49% + 49¢ per US transaction, the unbranded card fields cost 2.89% + 29¢, and Pay Later transactions run 4.99% + 49¢ - PayPal fees. That branded premium buys real conversion: Fastlane, PayPal's accelerated guest checkout, converts over 80% of the time per PayPal's own claims, and the wallet's reach still rescues sales from customers who refuse to type a card number into an unfamiliar site.

Most founders forget PayPal also owns the quieter, cheaper gateway underneath: Braintree. It prices cards and digital wallets at 2.89% + 29¢ with ACH at 0.75% capped at $5, and offers interchange-plus and custom rates to established businesses, which makes it the negotiation path inside the PayPal family once volume justifies a conversation - Braintree fees. On the small-business side, the bundle keeps widening: invoicing with ACH at 1% capped at $10, in-person acceptance through Zettle at 2.29% + 9¢, and QR payments from the same account. The blended picture is a platform with at least five distinct online rates, which is why modeling your actual payment mix matters more here than anywhere else in this guide: a store whose customers click the PayPal button 60% of the time has materially different economics from one where it is 15%.

PayPal also moved early and aggressively on agentic commerce. It became the first wallet inside ChatGPT's checkout flow, launched Agent Ready (existing merchants accept agent-initiated payments with no new integration), and powers Perplexity's shopping agent, currently with no merchant transaction fee on that channel - PayPal newsroom. Its stablecoin PYUSD passed a $4 billion market cap and rolled out to 70 markets in March 2026. The structural weakness is unchanged for a decade: PayPal's user agreement permits holding funds up to 180 days, and a class action filed in January 2025 alleges accounts are frozen without warning or explanation - Top Class Actions. Best for: consumer-facing businesses where the PayPal and Venmo buttons demonstrably lift conversion enough to cover the premium, used alongside a primary card processor rather than as the only rail.

3.3 Square: the in-person default

Square wins on the physical side of small business the way Stripe wins online. In-person card payments cost 2.6% + 15¢ on the free plan, the POS software is genuinely free, and hardware runs from a $59 reader to a $799 register, with Tap to Pay on iPhone needing no hardware at all - Square fees. There is no upfront underwriting: a coffee cart can sign up at breakfast and take cards by lunch. The whole back office (invoices, online store, customer directory, and now Managerbot, the AI business agent that opened to beta in April 2026) hangs off the same free account - Square press.

The caution flags are recent and specific. The January 2026 online repricing took the free plan to 3.3% + 30¢, making Square's online rate the highest among mainstream processors, and the keyed-in rate of 3.5% + 15¢ punishes phone orders. Square's aggregator model also means risk is managed after approval: merchant reports of 90-180 day fund holds following volume spikes are common enough that Fox Business covered a petition by 1,600+ sellers over withheld reserves - Fox Business. Best for: cafes, retail, salons, and service businesses that are primarily in-person and want POS, payments, and back office from one free account.

3.4 Helcim: the spreadsheet operator's choice

Helcim is what payments look like when a processor competes on transparency instead of brand. Pricing is pure interchange-plus: in-person costs interchange + 0.40% + 8¢ falling automatically to +0.15% + 6¢ at higher volume, online costs interchange + 0.50% + 25¢ with the same automatic volume discounts, and there are no monthly, setup, PCI, or cancellation fees at all - Helcim pricing. For a business doing $10K+ a month in card volume, the math reliably beats flat-rate processors. Two features punch far above their weight for small businesses: ACH at 0.5% + 25¢ capped at $6 makes a $10,000 invoice cost $6 instead of $290, and Fee Saver passes processing costs to customers in a card-brand-compliant surcharge.

The limits are clear-cut: Helcim onboards US and Canadian businesses only, settles only in USD and CAD, and its integration ecosystem is a fraction of Stripe's (WooCommerce, QuickBooks, and a January 2026 browser extension that bolts payments onto 20+ web tools). Interchange-plus statements also take more effort to audit than a flat rate. Best for: established North American small businesses, especially invoice-heavy B2B, that can shift big payments to capped ACH and want the lowest honest card cost without negotiating.

3.5 Mollie: the European default

Mollie is the processor European founders pick when Stripe feels American. European consumer cards cost 1.80% + €0.25 (genuinely cheaper than US pricing thanks to regulated EU interchange), iDEAL costs a flat €0.32, SEPA transfers €0.25, and there are no monthly fees or contracts - Mollie pricing. The platform serves 250,000+ European businesses, offers one-click plugins for every major cart, and provides Mollie Capital cash advances with 24-hour decisions. It is also moving with the market: Mollie was among the first European PSPs compatible with OpenAI's agentic checkout protocol and is a Principal Member of the European Payments Initiative as iDEAL migrates into Wero through 2026-2027 - Mollie news.

The big 2026 event is consolidation: Mollie agreed to acquire GoCardless for roughly €1.05 billion in December 2025, creating a combined platform of 350,000+ businesses across cards, local methods, and bank debit, expected to close in the second half of 2026 - Mollie. The caveats: Mollie onboards European businesses only, and its excellent 4.5/5 Trustpilot average still contains a steady pattern of risk-review locks with reserves up to 180 days. Best for: EU and UK businesses, especially Benelux and DACH e-commerce, that want local payment methods and EU-localized support at the lowest card rates in this guide.

3.6 Airwallex: payments plus the bank layer

Airwallex sells a different bundle: payment acceptance is almost a feature inside a global financial account. US domestic cards cost 2.8% + 30¢, FX conversion runs 0.5-1% above interbank, payouts to 120+ countries on local rails are free, and multi-currency accounts replace the bank + Wise + processor stack that cross-border businesses otherwise duct-tape together - Airwallex pricing. The company raised a $330M Series G at an $8B valuation in December 2025 and made San Francisco a dual headquarters, with annualized revenue passing $1 billion - Airwallex newsroom.

The trade-offs are the inverse of its strengths: international card acceptance costs a steep 4.3% + 30¢, onboarding requires KYB approval over 2-5 business days, and account freezes dominate its negative reviews, with roughly a 3.4/5 Trustpilot average. Best for: founders who genuinely operate across currencies (global e-commerce, agencies paying overseas contractors, SaaS billing in multiple regions) and value the FX and payout savings more than the cheapest acceptance rate.

4. The Merchant of Record Route: Paddle, Lemon Squeezy, Polar, FastSpring, Dodo, Creem

The merchant of record model answers one question with brutal efficiency: what if you never had to think about sales tax, VAT, or GST again? When an MoR is the legal seller, the tax liability is structurally theirs: registration, calculation, filing, remittance, and audit exposure in every jurisdiction your customers live in. This is the one thing no PSP add-on can replicate, because tools like Stripe Tax calculate and file but leave you as the taxpayer of record - Quaderno. For a two-person SaaS selling into 40 countries, that difference is not 2 points of margin, it is whether compliance is a line item or a job.

The economics have a crossover point, and honest analysis requires showing both sides. One 2025 audit of a $50K MRR SaaS selling into the EU, UK, Australia, and multiple US states put the unbundled Stripe stack at $3,232 per month versus Paddle's all-in $2,625, making the MoR roughly $7,000 a year cheaper in that international scenario - Lumscope. The same source's counter-case: past roughly $85K MRR with a domestic-only customer base, the 5% becomes pure overhead and DIY wins clearly. The model also concentrates new risks: your revenue sits in the MoR's account on net payout terms, their name appears on customer statements, and a compliance failure at the platform hits every seller on it, as the FTC's $5 million settlement with Paddle in June 2025 demonstrated - FTC.

The hidden half of the DIY comparison is what tax compliance costs when you run it yourself, because the processor fee is only the visible part. Stripe Tax calculates at 0.5% per transaction, but filing moved to a subscription: the $90 plan includes four filings a year across two US states, with each additional jurisdiction billed separately and foreign filings priced individually - Stripe support. Third-party filing services price the same work à la carte: Numeral charges $75 per state filing and $150 per state registration, which puts a SaaS with nexus in ten states filing quarterly at roughly $3,000 a year in filings before a single audit question - Sphere. And after paying all of it, the liability is still yours: calculation and filing tools do not transfer audit exposure, which is the one item in the MoR bundle money cannot buy piecemeal. When founders say the MoR premium "bought back their weekends," this stack of registrations, thresholds, and filing calendars is what they mean.

A structural note before the profiles: the MoR category is being squeezed from above and below simultaneously. From above, Stripe Managed Payments went GA in April 2026 and turned merchant-of-record into a one-parameter option on the world's largest processor. From below, a wave of 2024-founded startups (Creem, Dodo) undercut the incumbent 5% with 3.9-4% rates. The app builders joined in too: Lovable made Paddle its built-in MoR in April 2026, the first native payments integration in an AI app builder, a pattern we mapped across the tooling landscape in our ranking of AI app builders.

4.1 Paddle: the institutional choice

Paddle is the MoR with the most complete machine around the checkout. The price is a single all-inclusive 5% + 50¢ per transaction covering processing, global tax compliance, fraud, billing, and churn recovery, with no monthly fee - Paddle pricing. What justifies it beyond tax: Retain, the AI dunning system from the $200M ProfitWell acquisition, is included free and recovers failed payments with retry timing trained on millions of failures, and the November 2025 release expanded Paddle's MCP server to 80+ tools so AI agents can drive the full billing API. Paddle serves 6,000+ software companies, runs a free AI Launchpad accelerator with 265+ startups incubated (a niche complement to the mainstream programs in our US accelerator ranking), and after the Epic v. Apple ruling positioned itself as the web-checkout escape from Apple's 15-30% commission.

The picture below is the product in one frame: Paddle's overlay checkout drops onto any site with a few lines of script, while Paddle remains the seller of record behind it.

Mobile is Paddle's newest front, opened by the courts. After the April 2025 Epic v. Apple contempt ruling forced Apple to allow US apps to steer users to external payment, Paddle and RevenueCat shipped a joint web checkout that redirects in-app purchase intents to a prefilled Safari checkout with Apple Pay, at Paddle's standard rate of roughly 5-6% all-in versus Apple's 15-30% commission - Tomorrow's Publisher. Case studies from the first wave are striking: one app developer, HubX, recovered $100K+ in 72 days of web sales that would previously have carried Apple's commission. For any founder whose product lives in an app store, this single regulatory change may matter more than every processing-fee difference in this guide.

The weaknesses are material. Onboarding verification can take 48 hours to 2 weeks and rejects businesses for undocumented requirements like unconditional refund policies - dev.to. Payouts are monthly by default, so revenue earned early in a month can sit six weeks. And the June 2025 FTC settlement, with its $5M payment and permanent ban from processing for tech-support telemarketers, showed regulators treat the MoR as the seller in every sense. Best for: SaaS and digital product companies with global customers and no finance team, who want the most battle-tested MoR and will tolerate slower cash flow for it.

4.2 Lemon Squeezy and Stripe Managed Payments: one story now

Lemon Squeezy deserves an unusual review because its own roadmap points away from itself. The product still works as advertised: 5% + 50¢ merchant of record with a no-code storefront, license keys, and same-day checkout links - Lemon Squeezy pricing. But since Stripe acquired it in July 2024, the team has been building Stripe Managed Payments, and the CEO's January 2026 update openly apologized for "slower support responses and less frequent product updates" while promising an easy migration path to the Stripe product - Lemon Squeezy blog. Surcharges stack hard on the headline rate: +1.5% international cards, +1.5% PayPal, +0.5% subscriptions, which pushes a small international subscription toward an effective 12-13%.

The successor is compelling in a different way. Stripe Managed Payments costs more all-in (about 6.4% + 30¢ on a US card) but applies MoR status at the transaction level, meaning a growing business can move only its international volume onto MoR while processing US sales at standard rates, something no standalone MoR offers. It went GA in 39 countries on April 22, 2026 - Stripe changelog. Best for: existing Lemon Squeezy stores that should stay put until the migration tooling matures, and new projects that want MoR-as-a-feature inside the Stripe stack rather than a separate platform.

4.3 Polar: the developer's MoR, rebuilt for AI billing

Polar is the open source entry, and in 2026 it found its niche: billing for AI products. New organizations pay 5% + 50¢ on the free Starter plan, with paid tiers ($20, $100, $400 per month) cutting the rate to 3.8%, 3.6%, and 3.4% respectively - Polar pricing. The differentiator is usage-based billing for LLM products: Polar's ingestion strategy wraps any OpenAI-compatible model client and meters tokens automatically into invoices, which is why its homepage now reads "a billing platform for the intelligence era." A $10M Accel-led seed round in June 2025 came with angels from Vercel, Supabase, Shopify, and Lovable - Polar blog, and 14+ framework adapters make a working integration a half-day job.

The risks are stage-appropriate: a mandatory pre-payout review of up to 14 days, payout holds that can extend to the 120-day chargeback window, and support complaints documented through early 2026. The May 2026 repricing also stung early users, since Early Member organizations permanently lose their grandfathered 4% + 40¢ rate the moment they touch a paid plan. Best for: technical indie founders and AI startups that bill by usage or tokens and want MoR compliance handled in a few lines of code.

4.4 FastSpring, Dodo, and Creem: specialists at the edges

FastSpring is the 20-year veteran, strongest where its age is an asset: games and established software companies. It no longer publishes pricing (commonly reported at 5.9% + 95¢), supports payment methods in 200+ regions, handles buyer-side support 24/7, and after the Epic v. Apple ruling built a real business in done-for-you game web shops as the alternative to 30% app store fees - FastSpring. The $0.95 fixed fee makes effective rates exceed 10-12% on low-ticket subscriptions, and a $150/year risk-verification fee for sellers under $5,000 a year is openly hostile to hobbyists. Best for: game studios and established software businesses with meaningful volume and a sales conversation's worth of patience.

Dodo Payments attacks the market the incumbents underserve: India and emerging-market founders selling globally. Pricing is 4% + 40¢ with UPI support at 4% + 15¢ domestically, plus surcharges (+1.5% international, +3% PayPal and BNPL) - Dodo pricing. The platform claims 25,000+ builders, ships Sentra, an AI billing agent that wires products and webhooks from prompts inside Cursor or VS Code, and added stablecoin acceptance (USDC, one-time payments) in May 2026. The counterweight: a disclosed funding base of just $1.1M pre-seed, account-closure complaints on review sites, and a fee schedule complex enough that a rival published a teardown of it. Best for: founders in India and emerging markets who need an MoR that treats UPI and their local reality as first-class.

Creem is the price leader and the most AI-native onboarding in the category: a flat 3.9% + 40¢ with international card fees included, tax compliance in 50+ countries, and a CLI plus SKILL.md flow that lets an AI coding agent set up your entire store - Creem pricing. The structural caution is concentration of trust: Creem is an Estonian pre-seed company (EUR 1.8M raised August 2025) that started as a two-person team, holds payments 7-12 days for risk assessment, pays out twice monthly with a $50 minimum, charges $7 or 1% per bank payout, and checks out customers in USD and EUR only. Best for: indie hackers and small AI SaaS teams optimizing for the lowest MoR rate and AI-driven setup, who size the counterparty risk consciously and keep balances swept.

5. Bank Debit and B2B Payments: GoCardless and ACH

Cards are the wrong rail for a meaningful slice of small-business revenue, and the businesses that learn this earliest save the most. Card fees scale linearly and never cap: a $20,000 retainer invoice costs roughly $580 at 2.9%. Bank debit fees cap: the same invoice via GoCardless costs $5, and via Helcim's ACH it costs $6 - GoCardless pricing. For B2B services, agencies, memberships, and any business that invoices the same clients monthly, moving even half of volume to bank rails is frequently worth more than every processor negotiation combined.

GoCardless is the specialist here, built entirely around pull-based bank payments: ACH in the US at 0.5% + 5¢ capped at $5 on the Standard plan, Bacs in the UK, SEPA in the eurozone, across 30+ countries and 100,000+ businesses. Its structural advantage over cards for recurring revenue is mandate permanence: bank mandates have no expiry dates and no reissued card numbers, and GoCardless claims a 97.3% first-time success rate with its bank-pull system while Success+ retries recover roughly 70% of failures - GoCardless. The company hit its first adjusted-EBITDA-positive quarter in FY25 on £160.9M revenue, launched an MCP server in February 2026 for agent-driven integration, and in June 2026 launched Recurring Pay by Bank in the UK as the first open-banking alternative to recurring card payments - GoCardless blog.

The honest limits: settlement takes about 4 business days standard (2 with ACH Pull), there is no card acceptance at all, support is email-only, and its 2.5/5 Trustpilot average is dominated by compliance-review holds. The pending Mollie acquisition (agreed December 2025, closing expected late 2026) should eventually fix the no-cards problem by combining the two platforms, but integration churn is a real near-term consideration. Best for: B2B and subscription businesses with recurring high-value payments, used either standalone or as the bank-debit rail next to a card processor.

Plan the cash-flow geometry before you switch an invoice book over, because bank debit trades speed for cost. A standard GoCardless ACH payment takes about 4 business days from charge to cleared funds, customers must first authorize a mandate (a two-minute online flow, but a real step in your onboarding), and failures carry a $5 fee. Instant Bank Pay, the open-banking sibling, confirms one-off payments instantly where it is available and costs the same plan rate, which neatly covers deposits and first invoices while the mandate handles the recurring tail - GoCardless support. Bank debit is the wrong tool where fulfillment is instant and tickets are small: consumer e-commerce needs the authorization guarantees of cards, and no restaurant is asking diners to sign a mandate. The fit is precise, and inside that fit it is unbeatable.

The practical pattern for most small businesses is not either-or but routing by payment size: cards for checkout-sized transactions where instant confirmation matters, bank debit for anything invoice-sized. Stripe, Helcim, and GoCardless all support this split natively, and the capped-fee math means the blended rate of a B2B business can land closer to 1% than 3% with one afternoon of setup work.

6. Commerce Platforms and the Enterprise Tier: Shopify, Adyen, Checkout.com

Some platforms sell payments as part of a bigger promise, and for two opposite kinds of businesses that bundle beats any standalone processor. At the small end, Shopify Payments folds processing into the storefront itself. At the large end, Adyen and Checkout.com sell negotiated, interchange-pass-through acquiring to businesses big enough to clear their minimums. Knowing which tier you are actually in prevents both overpaying and being rejected.

6.1 Shopify Payments: checkout as a conversion weapon

Shopify Payments is the strongest argument that payments should not be chosen in isolation. Online rates are competitive (Basic plan: 2.9% + 30¢ plus the $39/month subscription, falling to 2.5% + 30¢ on Advanced), but the real asset is the checkout itself: Shop Pay lifts conversion by up to 50% versus guest checkout and beats other accelerated wallets by at least 10%, per a study Shopify commissioned from a major consulting firm - Shopify. In 2026 Shopify also became the most agent-distributed commerce platform: Agentic Storefronts make merchants discoverable and purchasable in ChatGPT, Perplexity, and Microsoft Copilot by default, and the Google-co-developed Universal Commerce Protocol put Shopify merchants inside Google's Universal Cart announced at I/O 2026 - Shopify news. Stablecoins are native too: USDC checkout settles in local fiat with no FX fees across 34 countries.

Two costs hide in the bundle. First, gateway lock-in is priced in: using any third-party processor triggers an extra 0.6-2% Shopify fee on every order on top of that provider's fees - Shopify help. Second, the fine print moved recently: Shopify quietly reduced its US annual-billing discount in early June 2026 (Grow now $95/month paid yearly, Advanced $360, versus $79 and $299 the week before), a change we verified against successive captures of the official pricing page. Add documented payout reserves of up to 120 days for merchants whose dispute rates climb, and the platform demands the same risk hygiene as any aggregator. Best for: product businesses that live or die on checkout conversion and want AI-channel distribution without integration work. If you are still deciding how to build the storefront around it, our AI website builders market map covers that layer.

6.2 Adyen and Checkout.com: graduate schools, not starting points

Adyen publishes the cleanest enterprise pricing in the industry: interchange at cost + 0.60% + $0.13 per Visa or Mastercard transaction, no monthly or setup fees, but a monthly minimum invoice that varies by industry and effectively excludes small businesses - Adyen pricing. What the minimum buys at scale is real: a single acquirer-processor-gateway across online, in-app, and in-store (Unified Commerce grew 24% year over year to €196M net revenue in Q1 2026), processing for Uber, Spotify, and Microsoft, and a €750M loyalty acquisition (Talon.One) announced in April 2026. Onboarding is sales-led and takes weeks, which is the point: Adyen underwrites before, not after. Treat it as the platform you graduate into at roughly $5-10M+ annual volume or genuine multi-channel needs.

Checkout.com plays the same tier with a different geography: strongest in Europe, the UK, and MENA (first global PSP licensed by the UAE Central Bank), processing $300 billion in 2025, up 64%, with its first full-year EBITDA profitability - Checkout.com newsroom. It publishes no rates at all, and anecdotal reports suggest roughly $10M in annual volume to qualify. Its June 2026 launch of 24/7 stablecoin settlement for US merchants with Fireblocks (over $300M USDC settled in the pilot) is the most production-grade stablecoin payout offering of any PSP in this guide. Best for: established e-commerce and fintech businesses past $1M+ in card volume, especially those selling into Europe or MENA; small merchants should not start their journey here.

7. Agentic Commerce and Stablecoins: What Is Actually Real in June 2026

No payments topic generated more noise in the past nine months, so this section separates shipped reality from announcements. The structural force is real: when AI assistants mediate shopping, discovery and checkout decouple. The assistant decides which products surface; the payment platform decides whether a purchase can complete without the customer ever visiting your site. Every major processor spent 2025-2026 racing to wire themselves into that loop, and four protocols emerged: Stripe and OpenAI's Agentic Commerce Protocol (ACP), Google's AP2 (donated to the FIDO Alliance in April 2026), the Shopify and Google Universal Commerce Protocol, and the Linux Foundation's x402 for machine-to-machine micropayments - Linux Foundation.

The keynote below is the single best primary-source overview of where this is heading: Stripe's Sessions 2026 keynote from April 29, where the Agentic Commerce Suite expansion with Meta and Google, the Link agent wallet, and 286 other launches were announced in one sitting.

The cautionary tale matters as much as the hype. OpenAI's Instant Checkout, the in-ChatGPT purchase flow launched with Etsy in September 2025, was wound down by March 2026 after struggling with merchant onboarding, stale product data, and missing carts; OpenAI pivoted to product discovery that hands off to merchant-owned checkout - TechCrunch. Meanwhile the discovery side exploded: Salesforce measured $67 billion in AI-influenced Cyber Week sales (about 20% of orders), and Adobe found AI-referred traffic converting 42% better than regular traffic by March 2026, a complete reversal from a year earlier. The chart makes the swing concrete.

What this means practically for a small business choosing a platform today is reassuringly simple. Protocol risk has collapsed onto the platforms, not you: ACP, AP2, UCP, and x402 all moved to neutral governance in the first half of 2026, and Stripe, PayPal, Adyen, and both card networks sit inside all of them, so choosing a major processor buys exposure to every protocol. On Stripe, agentic enablement is roughly one line of configuration; on Shopify, agent discoverability is on by default; on PayPal, Agent Ready extends existing acceptance to agent-initiated transactions with no new contract. The thing actually worth doing this quarter is unglamorous: make sure your product catalog is structured and machine-readable, because agents cannot recommend what they cannot parse. The OpenAI side of this ecosystem moves particularly fast, and our founder's guide to Codex and OpenAI Sites tracks the adjacent tooling.

The card networks are building the layer most merchants will actually feel, and the good news is that it requires nothing from you. Visa Intelligent Commerce launched its Connect platform in April 2026, a single integration through the Visa Acceptance Platform covering agent authentication, tokenization, and spend controls, after completing hundreds of live agent-initiated transactions with pilot partners - The Paypers. Mastercard Agent Pay binds card credentials to specific registered agents through network tokenization, flows through existing processors, and this week extended into Agent Pay for Machines, an open protocol for machine-to-machine payments with sub-cent micropayments across cards, bank accounts, and stablecoins - Mastercard. Both networks target mainstream consumer availability by the 2026 holiday season, and merchants largely inherit the capability through their processor rather than integrating anything. A caution flag for the most futuristic corner: x402, the HTTP-native micropayment protocol, saw daily transactions fall roughly 92% from its December 2025 speculative peak, a reminder that machine-to-machine payment volume is infrastructure-in-progress, not a retail acceptance decision for this year - Blockeden.

Stablecoins crossed from speculation to a cost lever in the same window. Stripe accepts USDC at checkout with automatic USD settlement for a flat 1.5%, roughly half its card pricing - Stripe docs. Shopify runs USDC natively with local-fiat settlement and no FX fees. Checkout.com settles enterprise merchants in stablecoins around the clock. The regulatory picture favors merchants: the GENIUS Act (signed July 2025) regulates stablecoin issuers, not accepting merchants, does not take effect before 2027, and explicitly permits merchant discounts for stablecoin payment - Sidley. Further out sits machine-to-machine payment infrastructure: Stripe-incubated Tempo launched its mainnet in March 2026 with a protocol for autonomous agent payments, and Mastercard launched Agent Pay for Machines this week. For a deeper technical view, the session below covers the protocols underneath.

The first-principles read: payments platforms are becoming distribution platforms. For two decades the processor was the last mile of a sale your marketing created. When agents mediate discovery, the processor's protocol memberships and catalog syndication decide whether you appear in the conversation at all. That inverts the old buying logic: the cheapest processor that leaves you invisible to agent traffic may be the most expensive decision on this page within two years.

8. Account Freezes and Holds: The Risk Nobody Prices In

Every platform in this guide reserves the right to hold your money, and the only honest comparison is how often and how badly each one exercises it. The mechanism is structural, not malicious: aggregators approve merchants in minutes without underwriting, which means risk gets managed after the fact by algorithms watching for dispute spikes, sudden volume growth, and category risk. The cruel irony is that a successful promotion looks identical to fraud from the algorithm's seat, so the freeze most often lands at the exact moment a small business breaks through. Founder forums and communities (we cataloged the most active in our guide to founder communities) are dense with these stories, and the documented record backs them up.

The cases from the past 18 months span every major platform. Stripe: a $450K/month e-commerce brand frozen after a flash sale pushed chargebacks to 2.4%, triggering a 90-day rolling reserve, and the Flipcause collapse, where Stripe froze a $1.45M reserve as the donation platform spiraled into a Chapter 11 that left 3,200+ nonprofits as unsecured creditors for $29M - Oakland Voices. PayPal: a January 2025 class action alleging accounts are frozen without explanation and funds seized after the 180-day hold. Square: a petition by 1,600+ sellers over 20-30% rolling reserves with no appeal process - IT Pro. The MoRs concentrate the same risk one layer up, with Paddle and Lemon Squeezy both documented holding funds 180 days after account closures.

Why does this keep happening across every platform? Trace it down and the root cause is the business model, not any individual company's culture. Instant onboarding is only possible because underwriting was deferred, deferred underwriting means risk must be priced after money is already flowing, and post-hoc risk pricing can only act on statistical anomalies, of which the most common is a sales spike - Terms.law. The 90-to-180-day windows are not arbitrary either: they mirror the card networks' chargeback windows, so the processor is holding your funds for exactly as long as your customers can still claw payments back. Understanding this does two things. It tells you the freeze risk is structural and portable (switching aggregators does not remove it), and it tells you exactly which levers reduce it, because everything the algorithm reads (dispute ratio, growth shape, category, account age) is at least partially under your control.

If a hold does land, the record suggests process beats outrage. Respond to the review request with complete fulfillment evidence (tracking numbers, delivery confirmations, customer communications) rather than partial answers that restart the clock, get every commitment about release dates in writing, and escalate formally if dates slip, since documented cases show informal promises sliding for months - Terms.law. Most balances do release once the chargeback window closes; the businesses that fail in the interim are the ones that had no second rail and no cash buffer, which is the real lesson.

The defensive playbook is straightforward and almost nobody runs it until after their first hold. Keep your dispute rate well under 1% (clear descriptors, instant refunds, honest delivery estimates do most of the work). Warn your processor before planned volume spikes: a support ticket announcing your launch week genuinely changes how the anomaly reads. Sweep balances to your bank on the fastest payout schedule you can afford rather than letting funds pool. And once revenue is real, run a second processor warm: even a trickle of volume through a backup means a freeze is a bad week instead of a payroll crisis. None of this is paranoia; it is the operational equivalent of backups.

There is also a structural mitigation: pick the underwriting model that matches your risk profile. A business in a category processors flag (supplements, coaching, dropshipping, anything resembling them) will live better on a platform that underwrites upfront, even at the cost of weeks of onboarding, than on an aggregator that will approve it today and freeze it at the first chargeback cluster. The freeze sections in each profile above are not color commentary; for some businesses they are the whole decision.

9. The Decision Framework and Integration Playbook

All sixteen profiles compress into a decision tree most founders can walk in five minutes. Start from what you sell and where your customers are, not from brand familiarity: the structural fit (PSP vs MoR, cards vs bank debit, storefront vs API) determines 90% of the outcome, and the remaining 10% is taste. The tree below covers the common paths, and the worked scenarios after it add the numbers.

Run the math on three concrete scenarios. A US-only SaaS at $30K MRR: Stripe with Billing and Tax costs roughly 4.1% all-in (about $1,230/month) versus Paddle's 5% + 50¢ (about $1,600); DIY wins, and the gap widens with scale. The same SaaS selling 60% international: the Stripe stack drifts toward 5.5-6% once cross-border and FX surcharges land, while the MoR price stays flat and removes VAT registration in a dozen countries; the MoR wins on cost AND workload. A $1M/year agency invoicing 20 clients monthly: cards cost about $29,000 a year, GoCardless or capped ACH costs under $1,500; the rail choice matters more than any processor choice. These are the three archetypes; most real businesses are a blend, and the platforms genuinely do not mind you splitting volume across two of them.

Checkout quality is the other half of the decision, because the platform you choose ships the checkout your customers actually see. The data says checkout is where money quietly dies: the average documented cart abandonment rate is 70.22%, and the reasons are mostly fixable platform-level issues - Baymard Institute.

Integration itself has stopped being the hard part, and this is the quietest revolution in the whole stack. Every platform in this guide now offers a no-code path (payment links, hosted checkout, or plugins) that takes a first payment the same day the account clears review, and the coded paths have collapsed from weeks to hours: Polar and Dodo advertise framework adapters of roughly ten lines, Creem onboards via an AI agent reading a SKILL.md file, and Paddle, GoCardless, and Dodo all run MCP servers so AI coding assistants can drive their APIs directly. The state of AI-assisted building, which we covered in depth in Building Software With AI and the latest coding-model review, has made "integrate payments" a prompt rather than a sprint for most stacks. The endpoint of that trend is platforms where billing is simply part of the company from day one: Lovable ships Paddle built in, and AI company builders like Founden generate businesses with Stripe products, checkout, and subscription logic already wired and live. When the integration cost rounds to zero, the platform choice becomes purely about the economics and trust profiles in this guide, which is exactly how it should be.

A realistic first week looks like this, on almost any platform in the guide. Day one: sign up, submit KYC, and create a payment link for your flagship product; on Stripe, Square, or Creem you can often share it the same afternoon, while Paddle, Airwallex, and the bank-debit platforms make you wait out verification first. Days two and three: replace the link with a hosted checkout on your site (a plugin on Shopify, WooCommerce, or Wix; a few lines of script elsewhere) and turn on the wallets, since Apple Pay and Google Pay cost nothing extra everywhere and measurably cut abandonment. Week two: wire webhooks into whatever you use for fulfillment and accounting, enable the dunning or retry tooling your platform includes, and send yourself a refund to learn that flow before a customer forces you to. The deliberate omission from this list is anything custom: every hour spent rebuilding what the hosted tools already do is an hour of new PCI scope and zero conversion gain at this stage.

One last operational rule: whatever you choose, own your customer and billing data exports from day one. Stripe will export card data only to another PCI Level 1 processor (and Link credentials never leave), MoR migrations are provider-mediated and slow, and Shopify's exit fee is priced into every order. The time to understand a platform's exit door is before you walk in the entrance - Stripe docs.

10. Future Outlook: Where Payments Go From Here

The macro tailwind behind every platform in this guide is unambiguous: payments keep growing faster than the economy. Global payments revenue hit $2.5 trillion in 2025 on roughly $2 quadrillion in value flows, and McKinsey projects $3 trillion by 2029 - McKinsey. Digital wallets already carry 56% of global e-commerce value - Payment Expert, and cash fell to 46% of global transactions. The pie grows; the fight is over who slices it.

Three forces will reshape the list above by this time next year. First, consolidation is accelerating: Mollie absorbing GoCardless, Stripe digesting Lemon Squeezy into Managed Payments, Adyen buying loyalty with Talon.One. The independent mid-tier is thinning, which historically precedes price increases; the counterweight is the unusually aggressive 2024-vintage MoR cohort and capital still flowing into challengers, a funding environment we mapped in our US VC landscape review. Second, agent-mediated commerce will keep shifting value from checkout to catalog: the OpenAI checkout retreat showed in-chat purchasing was premature, but the conversion data shows agent-referred customers are already better customers, and every protocol that matters reached neutral governance this year, removing the excuse to wait. Third, stablecoin settlement at half the price of cards puts the first real ceiling on card economics in decades; merchants will not switch for ideology, but they will switch for 140 basis points.

Europe is running its own parallel plot worth tracking if your customers are there. The European Payments Initiative's Wero wallet is absorbing national schemes: iDEAL, the method behind the majority of Dutch e-commerce, began its co-branded transition in January 2026 with full migration running into 2027, and Mollie joined EPI as a Principal Member to roll Wero out across Germany, Belgium, and France - EPI. Account-to-account rails like Wero and the UK's new open-banking schemes are Europe's structural answer to card fees, the same economic pressure stablecoins express in the US. The other quiet trend is platforms preparing to be integrated by machines rather than people: GoCardless, Paddle, and Dodo all run MCP servers, Helcim publishes its documentation in an AI-agent-readable index, and Creem lets an agent onboard a store end to end. Within a year, "which platform does my AI coding assistant integrate best" will be a normal selection criterion, and the platforms above are positioning for exactly that question.

For founders, the structural conclusion of this entire review is almost boring: the payment layer has commoditized downward in difficulty and upward in strategic weight simultaneously. Getting live takes an afternoon on any platform here. Getting it right (the legal-seller question, the rail mix, the freeze hygiene, the agent readiness) compounds for years. The global founder population keeps growing, as we documented in the worldwide founder data guide, and every one of them will make this decision; the ones who make it deliberately keep two to four points of margin the others donate to defaults.

Conclusion

The 2026 field sorts cleanly once you ask the structural questions. Stripe (8.1) is the best default for the same reason it tops the table: it is the only platform where a payment link today can grow into MoR, stablecoins, and agentic checkout without replatforming. Helcim (7.4) is the cost champion for established North American businesses; Mollie (7.4) is the European equivalent. Shopify Payments (7.3) wins wherever checkout conversion and AI-channel distribution decide outcomes, and Square (7.1) still owns the counter. In the MoR lane, Paddle (6.4) is the institutional pick, Polar (6.4) the developer and AI-billing pick, and Creem (6.2) the price-led challenger for those who size the counterparty risk honestly. GoCardless (6.5) is not optional knowledge for B2B: capped bank-debit fees are the single largest payment saving most invoice businesses will ever find.

Decide in this order: legal seller first (PSP margin versus MoR liability transfer), rails second (cards for checkout, bank debit for invoices), distribution third (storefront and agent readiness), and freeze hygiene always (dispute rate, balance sweeping, a warm backup). Price last, because by the time the first four are answered, the price differences between your remaining candidates are usually smaller than one month of the conversion or compliance gains at stake.

Written by Yuma Heymans (@yumahey), founder of Founden, where AI-built companies ship with working Stripe billing from day one, and co-founder of HeroHunt.ai.

This guide reflects the payment platform landscape as of June 11, 2026. Every price was verified against official pricing pages during the week of publication, but fees, features, and policies in this industry change monthly: verify current details on the official pages before committing.