AI dynamic routing

Last updated:July 31, 2026

Multi-acquiring was originally adopted for resilience, so that one acquirer having a bad day would not take payments down with it. AI Dynamic Routing is what turns that investment into revenue. Most merchants run multiple acquirers expecting better acceptance rates, broader geographic coverage, better resilience, and stronger commercial terms. AI Dynamic Routing turns multi-acquiring into a continuously learning authorization engine, one that authorizes more payments and recovers more revenue on every transaction.

See Payments orchestration for the full multi-acquiring-to-control story. This guide picks up at optimization and follows it through recovery.

The hidden revenue problem

Many payment failures are avoidable. The customer:

• Has funds
• Passed authentication
• Is legitimate

Yet the payment is still declined. Issuer decisions depend on context, including:

• Acquirer
• MID
• Geography
• Local vs. cross-border processing
• Historical performance
• Recent approval trends

Static routing leaves money behind

Traditional routing asks a fixed question: which acquirer should receive 30 percent of traffic? A percentage split applies the same logic to every transaction and never asks the question that actually decides approval: which acquirer gives this transaction the highest probability of clearing, right now?

Dynamic routing changes the question

Every authorization becomes an optimization opportunity. Instead of a fixed split, each eligible acquirer is scored in real time. Same three acquirers, one before-and-after:

How dynamic routing works

AI Dynamic Routing evaluates every eligible transaction for routing. It sits at the point where a transaction is about to be sent, and checks whether a better decision is available before it goes.

Most of the time the dispatch rule already picked well, and the AI checkpoint confirms it. When a higher probability path exists elsewhere, the transaction is sent there instead, through the merchant account tied to that acquirer.

Same transaction, four real decisions

Follow one transaction through the system: a domestically issued Visa card, eligible for three acquiring paths. Four dynamics decide whether it clears or falls through, and a fixed split has no way to see any of them.

Same card, three moments where a fixed split guesses wrong, plus a fourth dynamic, issuer partnerships, that a fixed split can't see either. For this transaction, local acquiring is the deciding factor: dynamic routing reads all of it and routes to the domestic local MID, the higher probability path.

Dynamic routing and recovery

AI Dynamic Routing selects the best first attempt. Most of the time that is the end of the story: the transaction is approved and revenue is captured with no further action. When the first attempt is still declined, Smart Retry picks up the same transaction and resubmits it on the healthiest fallback path.

The two capabilities divide the work: AI Dynamic Routing gets the first attempt right as often as possible, Smart Retry recovers what is left when it does not.

For recurring and subscription billing, an immediate retry is not always the right move. When the decline itself signals a timing problem, for example a Merchant Advice Code for insufficient or temporarily unavailable funds, the better recovery is to wait rather than resubmit right away. Dynamic routing hands that case to a recovery plan instead of Smart Retry.

The acquirer the model chose stays the acquirer used on every later attempt, whether the recovery happens immediately through Smart Retry or days later through a recovery plan. For subscription billing, that consistency matters as much as the recovery itself. See the MAC Scheduler guide for how recovery cadences are configured.

Explainable optimization

Every decision is explainable. Merchants can see why a given MID was selected, not just that it was: the model returns the reasoning and the ranking for every acquirer it considered, not only the one it picked. Local acquiring, MID history, transaction context and issuer partnerships, the four dynamics covered above, are exactly the kind of signals that resolve into the reasoning below.

What decides that ranking is transaction context, not a fixed score to interpret. The model keeps learning and adapting: it looks at signals across the full customer base it processes, not only this merchant's own transactions, continuously refining its read on which acquirer performs best in which context. Among the signals it weighs on every transaction:

Business impact

Back to this domestic Visa transaction. On a fixed split it clears at 88 percent. AI Dynamic Routing alone lifts that to 94 percent by recognizing the local acquiring advantage a static split can't see. Whatever is still declined does not stop there: Smart Retry and Retry Later give it further chances instead of letting the sale disappear. Multiplied across every transaction, acquirer and market, that is the shift AI Dynamic Routing makes for the merchant.

Beyond this one transaction, the same mechanism delivers, across the merchant's full portfolio:

Final positioning

Multi-acquiring gives you options. Orchestration gives you control. AI Dynamic Routing gives you optimization.

Across an immediate Smart Retry or a scheduled retry later, the model keeps the same acquirer it originally selected. For recurring and subscription billing, that stickiness is part of the value: the merchant is not just recovering a decline, it is recovering it without disturbing the acquirer relationship the billing history is built on.

AI Dynamic Routing is a continuously learning system, not a fixed rule set. As more acquirers, MIDs and alternative payment methods enter the routing decision, the model keeps doing the same job at greater scale: score every eligible path on every transaction and select the one most likely to succeed.

See also

• Payments orchestration guide
• Smart retry integration guide
• MAC scheduler guide