Smart payment routing is the practice of choosing a payment provider per transaction, in real time, based on which one is most likely to approve it at the lowest cost. Static routing — one PSP per country, set once and forgotten — leaves a measurable share of revenue on the table every single day.
How the decision logic works
- Transaction attributes — amount, currency, country, card BIN, issuer, payment method, device and channel.
- PSP performance data — rolling approval rates, latency and error rates per provider, segmented by BIN and geography.
- Business rules — volume commitments, cost per provider, licensing constraints and provider caps.
- Risk signals — velocity, fraud score and 3DS outcome, used to steer or step up rather than blanket-decline.
Every attempt feeds back into the performance model, so routing decisions improve as traffic accumulates instead of decaying.
The five routing rule types
1. Country-based
Route to acquirers with local presence in the player's market. Local acquiring typically lifts approval rates sharply versus cross-border processing and reduces interchange.
2. Performance-based
Weight traffic toward the provider with the highest live approval rate for that BIN and market, and shift away automatically when performance drops.
3. Cost-based
Where two providers perform comparably, route to the cheaper one — while respecting minimum volume commitments.
4. Fallback / cascading
On a soft decline or a technical error, retry with the next provider in the chain within the same session. Hard declines are never retried.
5. Risk-based
Send higher-risk transactions through 3DS or to providers with stronger fraud tooling, and keep low-risk traffic frictionless.
Real-world impact
| LEVER | TYPICAL RESULT |
|---|---|
| Approval rate | +5–15% |
| Cascading recovery | +3–8% of attempted volume |
| Processing cost | −15–30% |
| Outage impact | Contained within seconds |
| Checkout latency | Unchanged — decisions run in milliseconds |
Common mistakes
- Retrying hard declines, which burns issuer trust and can trigger fraud flags.
- Optimizing on cost alone and losing more in declines than the saved fees.
- Setting rules once and never revisiting them as issuer behaviour changes.
- Judging providers on global averages instead of BIN- and country-level data.
- No cap on retry attempts, producing duplicate authorizations and player complaints.
- Routing without unified reporting, so no one can prove what actually improved.
Does cascading create double charges?
Not when it is implemented correctly. Retries only fire on failed attempts with no successful authorization, and attempt limits are enforced per transaction.
How much data do I need before routing helps?
Country and BIN-level rules deliver value immediately; performance-based weighting sharpens over the first weeks of live traffic.
Can my payments team change rules without developers?
Yes — rules live in the control panel and take effect without a deployment.
See it running on your own traffic
Schedule a 30-minute call with our payments experts and we will map your current stack against what orchestration would change.
BOOK A DEMO