Account-to-account payments (A2A), the category most US shoppers know as pay-by-bank, move money directly from a buyer’s bank account to the merchant’s account without a card network in the middle. The pitch to retailers is blunt: cut interchange, settle faster, and stop renting rails from Visa and Mastercard. The reality at the checkout in 2026 is more nuanced, and the gap between the pitch and the receipt is exactly where merchants get burned.
This guide treats A2A the way a payments lead actually treats it: as one line in a cost-and-friction spreadsheet, weighed against cards, wallets, and the broader crypto digital wallets conversation that A2A is often lumped into. The same bank-rail plumbing (RTP, FedNow, open banking) sits underneath stablecoin settlement experiments and direct-debit checkouts alike, so the trade-offs rhyme even when the branding differs. We will get specific on fees, settlement timing, refund mechanics, and the handful of retail contexts where pay-by-bank already earns its place.
In short
- A2A wins on cost for high-ticket and recurring payments, where saving 150 to 250 basis points per transaction beats the convenience tax of cards.
- Settlement is fast but irreversible: instant rails like FedNow and RTP push real-time, yet that same finality makes refunds and chargebacks a manual operations problem, not an automatic one.
- Consumer habit is the real moat. Shoppers default to saved cards and one-tap wallets; A2A needs a concrete incentive (discount, loyalty boost, or the only accepted method) to change behavior.
- It is not crypto, even though it lives in the same digital-wallet bucket. A2A settles in fiat through bank rails, so regulatory and tax treatment differ from on-chain payments.
- Best fit in 2026: bill pay, B2B invoicing, marketplaces with thin margins, and account top-ups, not impulse retail at $40 baskets.
What account-to-account payment actually means in retail
An account-to-account payment is a bank transfer initiated at checkout, where the shopper authorizes a push (or a pull) from their bank account straight to the merchant. There is no card primary account number, no card network authorization, and no interchange fee paid to an issuing bank. In the US the rails are real-time systems like FedNow and The Clearing House RTP, plus open-banking connections through aggregators that read account balances and trigger payments with the shopper’s consent.
This matters because the cost structure is fundamentally different from a card swipe. A typical US card transaction carries interchange of roughly 1.5% to 2.5% plus a fixed assessment, and the merchant rarely controls the rate. An A2A transfer prices closer to a flat fee or a low single-digit basis-point cut, which is why the savings only become material above a certain ticket size. To understand why that delta exists, it helps to know how the card side is built, which the pillar guide on how card networks really work behind every retail checkout lays out in plain terms.
The catch is that A2A inherits the bank world’s habits, not the card world’s. Banks built rails for one-directional money movement, so a payment that completes is, by design, final. That single property reshapes everything downstream: refunds, disputes, fraud handling, and reconciliation all behave differently than a retailer trained on cards expects.
The fee math: where A2A saves and where it does not
Start with the answer: A2A saves real money on large and recurring payments, and saves almost nothing on small impulse baskets once you price in onboarding and support. The break-even point depends on your average ticket and your current effective card rate.
Consider a merchant paying a blended 2.3% effective card rate. A2A priced at a flat $0.30 plus 0.2% changes the comparison entirely as the ticket grows. The table below models the per-transaction cost at three common price points.
| Basket size | Card cost at 2.3% | A2A cost ($0.30 + 0.2%) | Saving per order |
|---|---|---|---|
| $25 | $0.58 | $0.35 | $0.23 |
| $120 | $2.76 | $0.54 | $2.22 |
| $600 | $13.80 | $1.50 | $12.30 |
| $2,400 | $55.20 | $5.10 | $50.10 |
At a $25 basket the saving is real but thin, and it can evaporate the first time a confused shopper opens a support ticket. At $600 and above the math is hard to argue with, which is why pay-by-bank lands first in furniture, electronics, travel, tuition, B2B invoicing, and subscription top-ups rather than at the grocery checkout.
There is a second cost line that the basket math hides: the cost of a declined or abandoned payment. Cards fail for expiry, insufficient funds, and issuer fraud blocks; A2A fails for timeout, unsupported banks, and authorization drop-off when a shopper bounces to a banking app and does not return. A 2% improvement in fee rate means nothing if A2A converts five points worse than a saved card. The serious way to read the economics is net revenue per presented checkout, not headline fee per completed transaction, and that is the number your pilot has to move.
Factor in the soft costs too. Onboarding a shopper to pay-by-bank for the first time carries a one-time friction tax: explaining the method, handling the “is this safe” hesitation, and absorbing a higher first-transaction support rate. Those costs amortize on recurring relationships and never amortize on one-time impulse buyers, which is the clearest signal of where A2A belongs in your mix.
The fee structures themselves vary by provider. Some price A2A as a flat per-transaction fee regardless of size, which favors high tickets even more aggressively than the blended model above. Others add a small percentage to cover open-banking aggregator costs. Always get the all-in number, including any monthly platform fee and per-refund charge, because a low headline rate with a per-refund fee can be worse than cards for a high-return category like apparel.
Recurring billing deserves its own note. When the same customer pays every month, you amortize the onboarding friction across many transactions, and the saved interchange compounds. This is the same logic that makes installment products attractive on the cost side, a dynamic explored in our comparison of Klarna, Afterpay, and Affirm for US merchants, where the fee story runs in the opposite direction (merchants pay more for conversion lift). A2A and buy-now-pay-later sit at two ends of the same spectrum: one trades convenience for cost, the other trades cost for convenience.
Settlement speed, finality, and the refund problem
Here is the honest trade-off: instant settlement is genuinely better for cash flow, and instant finality is genuinely worse for customer service. On FedNow or RTP, funds can land in the merchant account in seconds, twenty-four hours a day, with no multi-day card batch settlement and no rolling reserve. For a small retailer that lives on working capital, that is a meaningful upgrade.
The same property creates the operational headache. A card refund is a network primitive: you issue a credit and the rails handle reversal. An A2A refund is a brand-new outbound payment that you initiate, which means you need the customer’s correct account details, a process to verify them, and a control to prevent refund fraud. Disputes are worse, because there is no chargeback mechanism baked into bank rails the way there is on cards. If a shopper claims goods never arrived, you resolve it through your own policy and your own outbound transfer, not through an issuer’s dispute queue.
To deploy A2A without generating a support fire, work through this sequence:
- Capture and validate the return path first. Store a verified account reference at the time of payment so a refund does not require chasing the customer for bank details later.
- Write an explicit refund SLA. Decide whether refunds go out same-day or next business day and publish it, because shoppers expect card-speed reversals and will complain otherwise.
- Build a manual dispute playbook. Define who reviews “item not received” claims, what evidence triggers an automatic refund, and where the fraud threshold sits.
- Reconcile daily against bank statements. Real-time credits arrive outside normal batch windows, so a daily match prevents silent mismatches between your order system and your bank.
- Set a clear failure fallback. When an A2A authorization fails or times out, route the shopper to a card or wallet instantly rather than abandoning the cart.
Done well, this turns finality from a liability into an advantage: fewer reversible-payment fraud vectors, no chargeback fees, and faster access to cash.
Why A2A gets confused with crypto digital wallets, and why it is not
A2A frequently shows up in the same category page as on-chain payments, which muddies the strategy. Both promise to disintermediate the card networks, both live in a “digital wallet” mental bucket, and both get pitched as the future of low-fee checkout. The distinction that matters for a retailer is settlement asset and regulation.
Account-to-account settles in fiat through regulated bank rails, so the money never leaves the conventional banking and tax system. A crypto payment settles in a digital asset (or a stablecoin pegged to fiat) on a blockchain, which introduces volatility, custody, and a separate accounting and tax posture. The practical effect: A2A reconciliation looks like normal banking, while crypto reconciliation requires price-at-time-of-sale capture and a custody decision. We separate the real adoption from the noise in crypto payments in retail: real adoption versus hype, and the short version is that stablecoin-settled checkouts are closing the operational gap with A2A faster than volatile-asset checkouts ever did.
For most US retailers in 2026, the cleanest framing is this: pay-by-bank is a cost-reduction play on rails you already trust, while crypto is a treasury and customer-segment play that carries extra operational weight. Treating them as one project is how teams stall both.
A decision framework for adding A2A in 2026
The short answer to “should we add pay-by-bank” is: only where the saved interchange clears the cost of changing shopper behavior. That is a math question first and a technology question second, so run the numbers before you scope an integration.
Qualify by ticket and frequency. Pull your last ninety days of orders and segment by basket size and repeat rate. If a meaningful share of revenue sits above roughly $200 per order, or arrives as recurring billing, A2A has a real target. If your distribution clusters under $50 with low repeat purchase, the interchange you would save is too thin to justify the support and onboarding overhead, and you are better off optimizing card and wallet acceptance instead.
Score the operational lift honestly. The integration itself is rarely the hard part; the refund path, the dispute playbook, and daily reconciliation are. A retailer that already runs disciplined finance operations absorbs A2A easily, while a lean team without a reconciliation habit will feel the finality of bank rails as constant friction. The same scrutiny you apply to any new vendor at checkout applies here, and you should weigh the integration against the acceptance you already run rather than in isolation.
Pilot before you default. Turn A2A on for one high-ticket category or one recurring plan, attach a concrete incentive, and measure adoption, refund rate, and net fee saving over a full billing cycle. Do not promote it to the default method until the refund and dispute volume proves manageable. A controlled pilot also surfaces the edge cases (failed authorizations, partial refunds, mismatched account names) that a slide deck never mentions.
The retailers winning with pay-by-bank in 2026 are not the ones who flipped a switch. They are the ones who picked a narrow lane, priced the incentive, and built the operations to handle finality before a single customer hit “pay.”
Common mistakes retailers make with pay-by-bank
Most A2A failures are not technology failures. They are merchandising and operations failures dressed up as payment problems.
- Offering A2A with no incentive. Shoppers will not abandon a saved card to type bank credentials for nothing. A small discount or loyalty bonus is the price of behavior change.
- Treating refunds as an afterthought. Launching without a verified return path and a published refund SLA guarantees a wave of “where is my money” tickets.
- Pushing A2A on tiny baskets. The fee saving on a $25 order does not cover one support contact; reserve pay-by-bank for high-ticket and recurring flows.
- Ignoring the fraud profile shift. A2A removes chargeback fraud but adds first-party refund fraud and account-takeover risk, which need different controls.
- Hiding it at the bottom of checkout. If A2A is the cheap option you actually want used, it has to be visible and pre-incentivized, not buried under the default card field.
- Conflating it with crypto in tooling and reporting. Different settlement assets need different reconciliation; one shared dashboard hides problems in both.
FAQ
Is pay-by-bank the same as account-to-account payment?
Yes, in practice. Pay-by-bank is the consumer-facing name for an account-to-account payment, where money moves directly between bank accounts at checkout instead of through a card network. Some providers reserve “A2A” for the broader category including B2B transfers, while “pay-by-bank” usually means the retail checkout experience built on open banking or instant rails like FedNow and RTP. For a retailer choosing a method, treat them as the same product with the same cost and refund trade-offs.
How much can a retailer actually save with A2A versus cards?
The saving scales with ticket size. On a $25 basket the difference against a 2.3% card rate is roughly $0.20, which support costs can erase. On a $600 order the same comparison saves around $12, and at $2,400 it approaches $50 per transaction. The rule of thumb: A2A pays off on high-ticket, recurring, or B2B payments, and barely moves the needle on small impulse purchases. Run your own average ticket against your effective card rate before committing.
What happens when a customer wants a refund on an A2A payment?
Unlike cards, there is no automatic reversal. A refund is a fresh outbound transfer the merchant initiates, which requires a verified account reference and a defined refund SLA. Without those, you end up chasing customers for bank details after the fact. Build the return path at the moment of payment, publish how fast refunds go out, and create a manual playbook for disputes, because bank rails do not include a chargeback mechanism the way card networks do.
Is account-to-account payment a type of crypto?
No. A2A settles in fiat currency through regulated bank rails, so the funds stay inside the conventional banking system with normal accounting and tax treatment. Crypto payments settle in a digital asset or stablecoin on a blockchain, which adds custody, volatility, and separate reporting. They get grouped together because both bypass card networks and both sit in a “digital wallet” category, but the operational and regulatory profiles are different and should be managed as separate projects.
Will A2A replace card payments at retail checkout?
Not broadly, and not soon. For everyday in-store and impulse online purchases, saved cards and one-tap wallets remain faster and more familiar, and the fee saving is too small to justify behavior change. A2A is taking share in specific lanes first: bill pay, account top-ups, marketplaces with thin margins, B2B invoicing, and high-ticket categories. Expect a coexistence model through 2026 where pay-by-bank is one option among several rather than a wholesale replacement.
What rails power pay-by-bank in the United States?
The two instant rails are FedNow, run by the Federal Reserve, and RTP, run by The Clearing House, both of which move funds in seconds around the clock. Open-banking aggregators sit on top, connecting to a shopper’s bank to verify the account and authorize the payment with consent. Some implementations still fall back to ACH for lower cost when instant settlement is not required, trading speed for an even lower per-transaction fee.
What fraud risks change when you add A2A?
The fraud profile shifts rather than shrinks. You lose card chargeback fraud and the associated fees, which is a genuine win. You gain exposure to first-party refund fraud (customers claiming non-delivery to trigger an outbound refund) and to account-takeover attempts during authorization. The controls are different from card fraud tooling: focus on identity verification at authorization, refund-fraud thresholds, and a manual review queue for high-value or first-time A2A transactions.
What is next
Treat A2A as a targeted instrument, not a checkout overhaul: turn it on for your highest-ticket and recurring flows, attach a real incentive, and instrument refunds before you instrument savings. As instant rails mature and stablecoin settlement narrows the gap with bank transfers, the line between pay-by-bank and the broader digital-wallet stack will keep blurring, so build your reporting to tell them apart from day one, and keep the cost comparison grounded in how the underlying economics flow back to issuers and acquirers, as the pillar explainer on how card networks really work behind every retail checkout lays out. For the wider context on how these shifts ripple through merchants and vendors, our roundup of tools and vendors for department stores and chains in 2026 is a useful next read, and you can pressure-test the cost claims here against any provider quote using a neutral reference like the Federal Reserve payment systems resources.