Static routing, almost always. It captures the obvious gains, sending each currency to the matching processor and each local method to the processor that supports it, and it is easy to audit. Add adaptive logic only once you have data showing where it would help.
TL;DR. Every merchant with more than one payment connection is already using a routing method, whether or not anyone chose it. This guide compares static, rule-based, weighted, least-cost, geographic, performance-based and smart payment routing on approval rates, cost, resilience and the effort each takes to run, then gives a decision framework and the data any of them needs. PayAdmit provides the rules engine and the normalised reporting that makes those comparisons possible.
On This Page
- Introduction: Payment Routing Methods
- How a Routing Method Selects a Payment Path
- Static / Priority Routing
- Rule-Based Routing
- Weighted and Load-Balanced Routing
- Least-Cost Routing
- Geographic and Local-Acquiring Routing
- Performance-Based Routing
- Smart / Intelligent Routing
- Routing with Cascading and Failover
- How to Choose the Right Routing Method
- Data and Metrics Required for Routing
- Implementation Risks and Examples of Routing Rules
- Frequently Asked Questions
Introduction: Payment Routing Methods
A payment routing method is the logic a merchant uses to decide which payment provider handles each transaction. Every merchant with more than one payment connection is using one, whether or not anyone chose it deliberately. Sending every payment to the same processor is itself a payment routing method, just an unexamined one.
The methods differ in what they read, how often they change their answer, and how much operational attention they need. Some are fixed mappings a person writes once. Some evaluate payment attributes at runtime. Some respond to how each provider has been performing in the last hour. All are legitimate, and the right choice depends on payment volume, on how many payment providers are connected, and on what the merchant is actually optimising.
If the concept itself is new, what payment routing is covers the definition first. This guide takes each payment routing method in turn, explains the mechanics, and compares them on what matters commercially: approval rates, payment cost, resilience and the effort each takes to run. It closes with a decision framework and the data any of them needs to work.
Definition A payment routing method is the decision logic that selects which payment provider or acquirer receives a given transaction, ranging from a fixed mapping to a feedback loop driven by live approval rates.
How a Routing Method Selects a Payment Path
Every payment routing method runs the same three-stage process. Only the ranking step differs.
Normalise
stage one
The transaction is reduced to attributes: country, currency, amount, payment type, card BIN, merchant and any custom fields attached. Identical across every payment routing method.
Filter
stage two
Payment providers that cannot handle the transaction are removed for the wrong currency, an unsupported type or no coverage in that country. Also identical, and the reason coverage data has to be accurate.
Rank
stage three
The remaining payment providers are ordered. This is the only stage where routing approaches actually differ, and where all of the approval-rate and cost gains are won or lost.
Execute
and then recover
The transaction goes to the winner. If it fails recoverably, cascading presents it to the next provider, a separate mechanism sitting underneath every payment routing method.
Static / Priority Routing
Static payment routing is a fixed mapping written once: all euro payments to processor A, all card payments in one country to processor B. High-risk verticals lean on it heavily, as iGaming online payments shows. The ranking is a list somebody typed, and it does not change until somebody edits it.
Its strengths are underrated for a merchant starting out. Static logic is predictable, auditable and trivial to reason about during an incident. Nobody has to explain why a transaction went where it went. For a business with two payment processors and modest volume it captures most of the available gain at almost no operational cost.
The weakness is that it does not react. A payment provider that degrades at nine in the morning keeps receiving the same share of transactions until a person notices. In practice most merchants run static payment routing as the base layer and add reactive logic only where the data justifies it.
Priority routing. A static variant where payment providers are ranked in a fixed order and each transaction goes to the highest-ranked one currently available. It behaves like static routing under normal conditions and like simple failover when a processor is unreachable.
Rule-Based Routing
Rule-based payment routing evaluates transaction attributes at runtime and picks a payment provider accordingly. A person still writes the rules, but instead of one global mapping there is a set of conditions, each naming a segment and a destination.
The attributes used are already available before authorisation: issuer country, currency, card BIN, payment amount, payment type and merchant identity. A typical set has a dozen entries and covers most transaction volume with five of them.
Amount-based conditions are the most common first step. High-value payments go to the provider with the strongest approval rates on large tickets, low-value ones to the cheapest path, and the business accepts slightly higher processing cost on the transactions where an approval is worth most. BIN-level conditions come next, once there is enough evidence per issuer.
The failure mode for a merchant is sprawl. A set that grows to sixty entries becomes impossible to reason about, and conflicts produce payment routing decisions nobody intended. Keeping it small, ordered and documented matters more than making it comprehensive.
Weighted and Load-Balanced Routing
Weighted payment routing splits volume between processors by a fixed ratio: seventy per cent to one, thirty to another. It is not an optimisation approach, it is a policy instrument.
Businesses use it for three reasons. To keep a secondary payment provider warm so the connection is proven rather than theoretical. To meet a contractual volume commitment. And to gather comparable approval-rate data on identical transaction traffic, the only clean way to compare two providers fairly.
The point Weighted payment routing is how a business buys optionality, and the systems that execute it are compared in payment routing platforms. The few basis points it costs to keep a second provider live are insurance for a merchant against the day the primary one changes its pricing, its risk appetite or its availability.
Least-Cost Routing
Least-cost payment routing ranks candidates by cost and sends the transaction to the cheapest one that can handle it. The idea is obvious; execution is where businesses go wrong.
01 Cost is not one number
Interchange, scheme fees, provider margin, cross-border charges and currency conversion all move independently by country, card type and transaction amount. A provider cheaper on domestic debit can be dearer on cross-border credit.
02 Approval rates change the arithmetic
A payment provider twenty basis points cheaper that approves three per cent fewer payments costs the business more per completed transaction. Least-cost logic ignoring approval performance reliably makes the total worse while appearing to save money.
03 Measure cost per approved transaction
This single metric merges price and performance and is the only fair basis for ranking payment providers. Any least-cost setup ranking on headline rates is optimising the wrong thing.
04 Set a quality floor
The practical arrangement is a minimum approval rate below which a payment provider is excluded regardless of price, with cost choosing among those that clear it.
Geographic and Local-Acquiring Routing
Why domestic processing outperforms cross-border, and what it takes to use it.
Aspect
Approval rates
Interchange
Currency
Setup effort
When it pays
What changes
Domestic scores better with issuers
Domestic rates are usually lower
No conversion when settling locally
A local acquiring relationship
Enough volume in one country
Practical note
Consistently, across markets
Varies by scheme and region
Removes an FX cost line
Weeks to months per market
Model before committing
Approval rates
What changes
Domestic scores better with issuers
Practical note
Consistently, across markets
Interchange
What changes
Domestic rates are usually lower
Practical note
Varies by scheme and region
Currency
What changes
No conversion when settling locally
Practical note
Removes an FX cost line
Setup effort
What changes
A local acquiring relationship
Practical note
Weeks to months per market
When it pays
What changes
Enough volume in one country
Practical note
Model before committing
Performance-Based Routing
Performance-based payment routing ranks processors by how they have actually performed on comparable transactions over a recent window. It is the first approach where the system, not a person, changes the answer.
01 Define the segment
Performance means nothing in aggregate. Measure per country, per payment type and ideally per card brand, or the logic reacts to a mix shift rather than a real change in processing quality.
02 Choose the window
Too short and payment routing chases noise. Too long and it reacts slowly to a genuine outage. A rolling window in hours, with a minimum transaction count before the data is trusted, is the usual compromise.
03 Set floors and ceilings
No payment provider should fall to zero share on one bad hour, and none should take everything. Bounds keep the system stable and every connection warm enough to measure.
04 Keep the override
A person needs to be able to freeze the logic during an incident. Automated ranking is helpful right up to the moment it is confidently wrong.
Smart / Intelligent Routing
Smart payment routing is performance-based logic with more inputs and a model rather than a threshold behind it. Vendors use the label loosely, so ask what a given implementation actually does:
- Does it read more attributes than a written set could practically handle, or is it a rebranded rules engine?
- Does it learn per segment, or does it optimise a single global objective that hides regional differences?
- Can a person see why a transaction was routed where it was, after the fact?
- How much transaction volume does it need before its decisions beat a well-written static configuration?
- What happens during a cold start, in a market with no history?
- Can specific segments be excluded and handled by explicit rules instead?
The honest position is that smart payment routing outperforms a good written set at high volume and underperforms it at low volume, because a model with insufficient data is worse than a person who understands the business. Explainability matters too: payment routing nobody can explain is payment routing nobody can defend when a provider asks why its share collapsed.
Routing with Cascading and Failover
What sits underneath every routing method, and how the two recovery mechanisms differ.
Cascading
A failed payment is presented to the next-ranked provider. Recovers payments a single connection would have lost outright for the merchant.
Soft declines only
Cap the retry count
Track recovery rate separately
Failover
A payment provider is unreachable or timing out, so traffic moves wholesale. This is availability, not optimisation, and it should be automatic.
Trigger on latency and errors
Test it deliberately
Alert when it fires
The interaction
Cascading assumes the first choice was sensible. If payment routing is poor, cascading spends attempts recovering decisions that should never have been made.
Fix routing before tuning retries
Measure both together
Never cascade hard declines
How to Choose the Right Routing Method
A short decision framework, keyed to what the business is actually trying to achieve.
If the objective is approval rates
Start with geographic and BIN-level conditions, then add performance-based ranking on segments carrying enough transaction volume to be meaningful. The largest gains usually sit here.
If the objective is cost
Use least-cost logic with an approval-rate floor, and measure everything on cost per approved transaction rather than headline rates.
If the objective is resilience
Weighted payment routing to keep a second provider warm, plus automatic failover on latency and error thresholds. Resilience is bought before it is needed, never during.
If the objective is control
Static and rule-based payment routing, deliberately. Predictability has real value in a regulated business, and an auditable mapping beats a model nobody can explain.
The key conclusion Most businesses do not need to pick one payment routing method. They need static as a base, written conditions for the segments they understand, performance ranking where the evidence supports it, and cascading underneath all of it.
Data and Metrics Required for Routing
Every payment routing method above depends on data a business may not hold in comparable form. This is the work preceding any routing project.
01 Normalised decline codes
Payment providers report declines differently. Without a common taxonomy, no comparison between processors is valid and cascading cannot tell soft from hard.
02 Segment-level approval rates
Per country, per payment type, per card brand, per processor. Aggregate rates hide exactly the differences payment routing exists to exploit.
03 True cost per transaction
Including scheme fees, cross-border charges and conversion, not just the provider's headline margin.
04 Latency and availability
Response times degrade before decline rates move, which makes latency the earliest usable signal for both failover and performance ranking.
05 Cascade outcomes
Which retries succeed, on which processor, after which decline code. The feedback loop telling you whether cascading works or just adds attempts.
Implementation Risks and Examples of Routing Rules
Three risks account for most payment routing problems in production. Conflicts, where two conditions match the same transaction and the outcome depends on ordering nobody documented. Stale logic, where the processor mix changed and the configuration did not. And over-optimisation, where a business chases fractional approval-rate gains in segments too small for the numbers to mean anything.
The guardrails are unglamorous and effective: version the payment routing configuration, require a second reviewer, keep a manual override that works without a release, and set a review cadence rather than waiting for a problem.
As for the payment routing rules themselves, the useful ones are usually simple. Route transactions in a currency to the processor that settles it natively. Route payments in a market where the merchant holds local acquiring to that acquirer. Route above a value threshold to the processor with the best large-ticket approval rates. Exclude any processor whose decline rate in the last hour crosses a threshold. Four payment routing rules written from observation will outperform forty written from assumption.
If you are earlier in this topic, what payment routing is covers the definition and mechanics. For the systems that execute these methods, see payment routing platforms, and for how they behave under high-risk traffic, iGaming online payments.
Frequently Asked Questions
Which payment routing method should a business start with?
Can several routing methods run at once?
Yes, and mature setups usually do. Static rules cover the bulk of a merchant's payments, performance-based logic handles the segments with enough data to be meaningful, and cascading sits underneath all of it as recovery.
How much transaction volume is needed for smart routing?
Enough per segment for approval rates to be statistically meaningful, not enough in aggregate. A merchant processing heavily in one country may have plenty of data there and none at all in a market it just entered.
Does least-cost routing actually save money?
Only when measured on cost per approved transaction. A cheaper processor with lower approval rates costs a merchant more per completed payment, and least-cost logic that ignores approval performance reliably makes the total worse.
What is the difference between weighted and performance-based routing?
Weighted routing splits transaction volume by a fixed ratio you set. Performance-based routing changes the split itself according to recent results. One is a policy, the other is a feedback loop.
How often should payment routing rules be reviewed?
Quarterly at minimum, and immediately after any change to a merchant's processor mix. Routing logic decays because the conditions it was written for keep moving.
Can routing be configured without engineering involvement?
On a platform with a proper rules engine, yes, and that is much of the point of moving routing out of application code. Verify it during evaluation rather than assuming it.
Does routing help if a business uses one processor?
Barely. With one connection a merchant has nothing to choose between, though cascading between payment methods can still recover some declined payments. The real gains start with the second processor.
WHICH ROUTING METHOD FITS YOUR TRAFFIC?
Send us your processor mix and decline data and our team will tell you which routing approach is worth configuring first.
SIMILAR ARTICLES
The definition, the mechanics and the difference between routing, orchestration and cascading.
What a routing platform supplies, and how to compare one against a processor's built-in logic.
How routing and cascading behave under high-risk gaming traffic, end to end.