Outsource processing, not control
A PSP can handle acquiring, routing, risk tooling, settlement files, and reporting — but operators should stay clear on which parts of the stack remain portable, negotiable, and independently measurable.
TOPIC BRIEFING
Your PSP stack is not just a vendor choice. It decides who controls authorisation, settlement, tokens, redundancy, data portability, and market expansion.
Payment infrastructure is the operating layer behind acceptance. PSP selection, acquiring setup, MoR vs direct processing, token strategy, orchestration, settlement flows, and contract control decide how much flexibility an operator has when volume, markets, and risk increase.
Stack map
Each layer below has its own provider, its own contract, and its own operator implication. Compounding effects: the choice you make at one layer shapes the leverage you have at every layer downstream.
Checkout
Where payment intent starts — UX, conversion, and the first data captured.
Gateway
API, encryption, routing interface — the integration surface you maintain.
PSP / Acquirer
Authorisation, acquiring, settlement, and chargeback exposure sit here.
Orchestration
Redundancy, routing logic, and vendor abstraction across multiple PSPs.
Token Vault
Portability, recurring-payment resilience, and the deepest source of lock-in.
Settlement
Cash timing, reserves, and reconciliation complexity across providers and markets.
Reporting & Reconciliation
The source of truth across PSPs, markets, and currencies — and the audit trail.
The operator thesis
A PSP can handle acquiring, routing, risk tooling, settlement files, and reporting — but operators should stay clear on which parts of the stack remain portable, negotiable, and independently measurable.
Single-PSP setups reduce operational complexity, but they also concentrate outage exposure, pricing leverage, routing options, token dependency, and renewal risk in one vendor relationship.
Token strategy, data access, reconciliation structure, failover design, and contract exit clauses determine whether switching PSPs is a controlled migration or a painful rebuild.
Start here
Decide your operating model before evaluating providers. These three articles in order.
Understanding the real cost of your current PSP before adding complexity or moving.
For software companies evaluating Paddle, Polar, or Lemon Squeezy — and when to migrate off.
Briefings, grouped by decision
The foundational choices that shape everything downstream — operating model, vendor selection, contract terms, and token strategy.
MoR vs PSP: who owns tax remittance and chargeback liability, and when direct PSP beats paying the MoR premium. Decision framework for digital and SaaS sellers.
Stripe publishes rates, Adyen gates IC++, Checkout.com says contact sales. Compare landed effective cost per dollar — auth-rate gaps dwarf headline fee deltas.
PayPal's 440M-user network converts better in some demographics; Stripe Billing handles the full subscription lifecycle. How SaaS operators should choose.
The PSP contract clauses most operators sign without reading: the terms that shift liability, lock up cash, limit exit options, and cost more than expected.
Visa VTS and Mastercard MDES tokens are portable and auto-update; PSP tokens are processor-locked. Auth-rate and portability trade-offs for multi-PSP operators.
Raw PAN, vault token, PSP token, network token, device token — five owners, five migration paths. Token Requestor control decides what actually moves.
Domain verification, DPAN vs cryptogram, PSP-mediated vs direct decryption, and where wallet liability shift actually applies — the acceptance layer.
Token provisioning, spend controls, issuer-side disputes, funding, and portfolio KPIs — the operational layer that starts once a card programme is live.
FinCEN registration, state MTLs, and the agent-of-payee exemption — how the money-transmission trigger changes across PayFac, MoR, wallet, and payout models.
BIN sponsorship, issuer-processors, and program economics from the issuing seat: who is liable, how interchange works, and the terms that decide your exit.
RBI consolidated India's payment aggregator rules into one regime in September 2025 — three categories, net worth, escrow, and who needs to apply.
Advanced infrastructure for operators who have outgrown single-PSP simplicity — when redundancy and routing logic actually pay back.
Multi-acquirer routing is a tactic; orchestration is the infrastructure enabling it. When to use both, what routing strategies work, and the hidden complexity.
The Merchant of Record path end-to-end: the comparison, the tax mechanics, the exit thresholds, and the migration playbook to direct PSP.
Paddle, Lemon Squeezy, Polar & FastSpring compared by tax footprint, coverage, and fit — how to choose the Merchant of Record matched to your seller type.
MoR providers absorb VAT, GST, and US sales tax collection and remittance across dozens of jurisdictions. But income tax, corporate filings, transfer.
The rails and primitives behind recurring payments, embedded finance, and account-to-account billing on infrastructure that bypasses cards.
VRP: standing-consent bank pulls, no interchange, no expiry. Commercial VRP went live 2 June 2026 via UKPI — but subscriptions wait for Wave 2.
MIT flagging avoids SCA on recurring charges. Most operators assume it works automatically. Four implementation gaps driving European decline spikes.
Embedded finance is a $7 trillion opportunity in the US alone by 2026. But the infrastructure requirements — BIN sponsorship, ledger architecture, KYB.
Choosing a payout rail by market: FedNow, RTP, SEPA Instant, Faster Payments, PayNow, UPI, PIX, local ACH, and card push compared on speed, cost, reach.
How to fund payouts across markets without liquidity gaps: prefunding vs JIT, buffer sizing, cut-offs, safeguarding separation, and forecasting.
Other briefings in this topic
RPAA registration only makes a PSP eligible to apply. What Payments Canada membership, RTR participation approval, settlement and testing actually require.
The network rules give you no say in the rail. Only member banks submit. Whether your own bank will act on a routing request is in its terms, not theirs.
Merchant of record is a commercial label, not a scheme role. What the card rules classify by conduct, who carries dispute liability, and what contracts say.
Acquirer and PSP agreements set contractual deadlines to object to a statement — 30 to 60 days, on different triggers, with different consequences.
A payment link is a separate object from the payment and the order. Link state, expiry and revocation across Stripe, Adyen, Checkout.com and GOV.UK Pay.
Visa's account verification is a zero-amount check that a card is valid and not lost or stolen. What 00 vs 85 means, the stored-credential mandate, and limits.
What settles on weekends and holidays, per rail: FedACH banking days, FedNow cycle days, RTP prefunding, TIPS vs T2, card clearing, and PSP payouts.
What Request for Payment is on RTP and FedNow, legitimate uses, the message lifecycle, warranty obligations, and what operators should build.
iDEAL is being co-branded, then retired, in favour of Wero. What's live today, what's only announced, the stated end-2027 timeline, and what to plan for.
The SDD Core rulebook lets a creditor amend mandates without re-signing. Which changes need an amendment, which need nothing, and how the data travels.
South Africa's authenticated debit system is genuinely different from ACH, SEPA and Bacs — and the clean pitch for it is wrong in three ways.
SPAA lets banks charge for premium API access instead of giving it away under PSD2. Nearly three years in, the participant register shows brokers but no banks.
Amex publishes its change cadence and notice period. Visa and Mastercard largely do not. How operators actually find out what changes and when.
SRC is a specification, Click to Pay is a brand built on it, and neither is tokenisation or 3DS. What an operator actually integrates.
Account Updater doesn't guarantee stored cards keep working. Issuer participation is optional and coverage is partial — most operators over-assume it.
Online, every transaction authorises live. In-store, a terminal can approve with no connection at all — what changes for fraud, PCI scope, reconciliation
Mexico splits payment licensing into IFPE, ITF, and a separate acquirer/aggregator track. CNBV/Banxico capital, timelines, and IVA rules explained.
Wallets are accepted everywhere as a payment method. Here's the other side: how they're funded, who holds the float, and what safeguarding actually requires.
There's no single 'PSP licence' in Brazil — the category you need depends on the service you perform. Credenciador, subcredenciadora, IP, arranjo, explained.
Integration is a one-time project. Running hundreds of live terminals — provisioning, firmware, keys, tamper events, RMA — is a permanent operating discipline.
Local acquiring vs cross-border acquiring, defined — plus when each is required or optional in Indonesia, Brazil, the US, Korea, and Singapore.
An operator runbook for refund controls: the chargeback double-pay trap, refund-vs-void-vs-chargeback decisions, refund fraud, and refund reconciliation.
Day-2 operations runbook for marketplaces: split-payment funds flows, seller payouts and holds, negative balances, reserves, split refunds, and disputes.
An approved authorization is a hold, not settled money. The operator guide to the card lifecycle: authorization, capture, clearing, settlement — and the gaps.
The payment stack is a map of who you contract with, who holds your money, who carries liability, and who you call when a payment breaks — not a glossary.
When a payout fails, localize the failed leg, never blind-retry, and use idempotency to avoid paying twice. An operator runbook for outbound-money failures.
When a PSP or acquirer degrades, every second is failed payments. A runbook to detect the outage, fail over cleanly, and recover without double charges.
Operator reference comparing PSP, PayFac, acquirer, aggregator/marketplace, and MoR models: who holds the merchant account, onboards, owns risk, and settles.
Field-level reference for Stripe and Adyen settlement files: the universal object model, balance_transaction vs pspReference, and which IDs to persist.
Operator runbook for PSP reconciliation breaks: three-way match, per-PSP ID chains, 10 settlement-mismatch types with causes and fixes, plus escalation steps.
You have a PSP shortlist. How to evaluate finalists: score across 8–12 dimensions, run proof-of-claim tests, and disqualify before the contract.
How operators manage recurring billing: MIT credential setup, network tokens, direct debit mandates, retry logic, and involuntary churn recovery.
Direct debit payments explained for operators: mandates, ACH, SEPA SDD, Bacs, eGIRO, returns, disputes, timing, and when to use each rail.
Recurring payment rails compared for operators: card mandates, direct debit, A2A/VRP, and wallet billing — mechanics, risks, and when to use each.
A multi-axis decision matrix for choosing a PSP class by volume, operating model, and geography — before comparing individual providers.
Pure-play card acquirers face structural margin compression: interchange caps expanding, network fees rising, real-time rails eating volume. Why Stripe.
From topic to market
Reference
A payment gateway is the technical layer that transmits payment data between your checkout and the acquiring bank — it handles encryption, routing, and the API interface. A PSP (Payment Service Provider) bundles the gateway with acquiring, often adding fraud tools, reporting, and local payment method support in a single contract. An acquirer (or acquiring bank) is the licensed financial institution that holds the merchant account, settles funds, and takes on chargeback liability. Most merchants deal with a PSP that packages all three functions — Stripe, Adyen, and Checkout.com are PSPs with their own acquiring licences in most markets they operate in.
A Merchant of Record (MoR) makes sense when tax and compliance complexity exceeds your team's capacity to manage it. If you sell digital goods internationally, an MoR (Paddle, Polar, FastSpring) becomes the seller of record and takes on registering, collecting and remitting indirect tax in the jurisdictions it supports — the coverage is provider-specific, so check the list rather than assuming it is global. The trade-off is cost, since an MoR take rate sits above direct PSP processing, and control, since the MoR owns the customer relationship for payment purposes. There is no revenue threshold at which the model stops working: the trigger is structural, not a GMV milestone, and it arrives when the MoR's constraints cost more than the compliance infrastructure that would replace it.
The five highest-risk clauses: (1) Rolling reserve — how much of your funds the PSP holds and for how long; (2) Termination rights — whether the PSP can terminate with minimal notice and freeze funds; (3) Data portability — whether you can export your tokenised card vault if you switch PSPs; (4) Fee change notice period — how much notice you get before rate increases; (5) Chargeback liability — what thresholds trigger account review or termination. Most PSP contracts are heavily weighted toward the PSP's risk management needs, not merchant flexibility. The contracts for Stripe, Adyen, and Checkout.com all have specific red flags documented in the PSP contract red flags briefing.
Payment orchestration is a middleware layer (Spreedly, Primer, Gr4vy) that sits between your checkout and multiple PSPs, routing transactions intelligently — by geography, payment method, transaction value, or real-time success rates. It makes sense when you have: (1) multi-PSP redundancy needs (if one PSP goes down, failover automatically); (2) market-specific acquiring requirements (needing local acquirers in different countries for better auth rates); or (3) complex routing logic that your primary PSP can't express. For single-PSP merchants below roughly $10M in processing volume, orchestration adds integration overhead without proportional benefit. At scale, it is a meaningful authorisation rate and cost lever.
A PSP vault token (Stripe's tok_xxx, Adyen's recurring reference) is specific to that PSP — it represents your customer's card as stored in that PSP's system. If you switch PSPs, the token is useless and you have to re-vault all cards (expecting 10–25% customer drop-off during migration). A network token (Visa Token Service, Mastercard MDES) is issued by the card scheme and works across any PSP that supports network tokenisation. Network tokens also self-update when cards are reissued, eliminating subscription churn from card expiry. The practical implication: operators with large stored-card subscriber bases should prioritise PSPs that automatically provision network tokens, as it reduces both migration lock-in and recurring billing failure rates.
PSP selection, acquiring structure, gateway architecture, and routing logic are not commodity decisions — they directly determine authorisation rates, processing costs, vendor leverage, and how quickly an operator can adapt when a market shifts. The stack architecture you build in year one is often the one you’re working around in year three.
The decisions in this space have compounding consequences. A PSP contract with a punitive rolling reserve provision can tie up six figures of working capital. A vault tokenisation model that stores PSP-specific tokens rather than network tokens creates a de facto lock-in that costs 15–25% customer churn to escape. A single-acquirer architecture without routing logic leaves 2–5% of authorisation rate improvement unrealised. These are not theoretical risks — they are operational costs that most operators only discover after the contract is signed.
The briefings in this topic cover the full stack: how to evaluate and choose a PSP, how to structure multi-acquirer routing for resilience and rate improvement, how to read a PSP contract before you sign it, and when the complexity of direct acquiring outweighs the benefits of staying with a payment facilitator or merchant of record. The framing question throughout is the same: at every layer of the stack, who controls the next decision — you, or your PSP?