Reducing Payment Processing Costs with Smart Routing

Processing fees directly impact profitability at scale. If you’re handling 10 million transactions per month with an average transaction value of $100, a 2% effective processing rate results in $2M in monthly fees. Intelligent payment routing can reduce those costs by 30–40%, unlocking $600K–$800K in annual savings, without introducing latency or reliability trade-offs in high-throughput Erlang/Elixir production environments that demand fault tolerance and sub-100ms performance.

What fees are involved in payment processing?

Payment processing fees aren’t a single line item but a stack of distinct cost layers. Some are fixed by card networks, others are negotiated with providers, and a few can be actively optimized through routing, data quality, and local acquiring.

Payment costs accumulate through multiple layers. Understanding each component reveals optimization opportunities.

Gateway fees

Payment gateways transmit transaction data between merchants and acquiring banks. Most PSPs bundle gateway functionality into their pricing, though the underlying costs vary significantly. Stripe charges 2.9% + $0.30 for US cards, while Adyen's blended rate might be 1.8% + $0.15 for similar transactions. These differences compound at volume, on 100K transactions at $100 average, that's $140K monthly in gateway cost variance alone.

Acquirer fees

The acquirer markup represents what your PSP charges for providing acquiring services. This typically ranges from 0.5-2.0% depending on your volume, business type, and negotiated terms. Unlike interchange and scheme fees (which are mostly fixed), acquirer margins are highly negotiable, especially when you can demonstrate multi-PSP capabilities and credible threat to shift volume.

Scheme fees or network fees

Visa and Mastercard assess fees monthly, averaging 0.13-0.15% of transaction volume. These pass through from acquirer to merchant with minimal markup opportunity. American Express charges higher scheme fees (0.15-0.30%), which partially explains why Amex acceptance costs more. Scheme fees stay relatively constant regardless of provider, making them less useful for routing optimization.

Interchange fees

Interchange represents the largest cost component, flowing from the acquiring bank to the issuing bank. In the US, interchange ranges from 1.15% (regulated debit) to 3.30% (premium rewards cards). The EU caps interchange at 0.2% for debit and 0.3% for credit through regulation. Card type, merchant category code, transaction characteristics (card-present vs card-not-present), and authentication method all influence the specific interchange rate applied.

Interchange optimization focuses on qualifying for the lowest applicable rate through proper data handling. Level 2 and Level 3 data processing, passing additional transaction details like customer code, tax amount, and line-item data, can reduce interchange by 0.20-0.50% on commercial card transactions. For B2B merchants processing $10M monthly, that's $20K-$50K in monthly savings from data optimization alone.

Currency conversion fees

Cross-border transactions trigger forex markups when payment currency differs from settlement currency. Most PSPs charge 1-3% above mid-market rates for currency conversion. A €100 transaction processed through a US-based acquirer might incur 2.5% forex markup (€2.50), in addition to all other fees. Local acquiring eliminates this cost entirely by matching acquirer currency to transaction currency.

Refund fees

Processors charge for returning funds to customers, typically €0.25-€2.00 per refund depending on provider and transaction type. At scale, refund fees compound, 10K monthly refunds at €0.50 each costs €60K annually. Some providers like Stripe return the percentage-based processing fee but retain the fixed per-transaction fee. Others retain all fees. Routing high-refund-rate transactions to providers with favorable refund terms reduces this cost.

Chargeback fees

Dispute processing triggers fixed fees regardless of outcome. Stripe charges $15 per chargeback. Other processors range $20-$100. Beyond direct fees, excessive chargebacks (>1% of volume) trigger card scheme penalties and potential processing restrictions. Fraud prevention integration with routing, sending high-risk transactions to PSPs with stronger fraud tools, prevents chargebacks from occurring.

What factors influence payment processing fees?

Multiple transaction characteristics determine your effective rate:

Card network and type: Visa debit averages 1.29% + $0.05; Visa credit ranges 1.51–2.30%. Premium cards (rewards, corporate) carry higher interchange. Routing transactions by card BIN to PSPs with favorable terms for specific card types can reduce costs.

Geography and cross-border status: Domestic transactions cost significantly less than cross-border. An EU-to-EU transaction might cost 1.8% total through a local acquirer. Routing the same transaction through a non-EU acquirer adds ~2% forex markup, bringing total cost to 3.8%.

Merchant category code (MCC): High-risk categories (gambling, crypto) face higher rates, while regulated utilities and government entities enjoy preferential interchange. Correct MCC assignment impacts base costs before routing optimization begins.

Transaction size: Percentage-based fees hurt more on large transactions; fixed fees hurt more on small ones. A 2.9% + $0.30 fee structure costs 3.2% on a $100 transaction, but 5.9% on a $10 transaction. Routing small-value transactions to providers with lower fixed fees improves economics.

Authentication and data quality: 3D Secure authentication can qualify transactions for lower interchange and shift fraud liability. Passing complete AVS and Level 2/3 data reduces costs further. Network tokenization, replacing PANs with tokens, can boost authorization rates 2–4% while qualifying for preferential interchange.

Intelligent payment routing vs traditional static routing

Traditional single-PSP integration routes all transactions identically. Every payment flows through one provider at whatever rate that provider offers for each transaction type. When the PSP experiences downtime, payment processing stops. When the PSP's authorization rates decline, your revenue suffers. When they raise rates, you have limited negotiating leverage.

Intelligent routing connects to multiple PSPs through a unified orchestration layer. Each transaction gets evaluated in real-time, typically within 20-50ms, against criteria including total cost (fees + forex), historical authorization performance for similar transactions, geographic coverage and local acquiring capability, and current system health (response times, error rates). The routing engine selects the optimal provider for that specific transaction.

The architectural difference is fundamental. Static routing is a direct integration requiring no decision logic. Intelligent routing needs a distributed orchestration engine handling concurrent PSP fee evaluation through parallel queries, sub-100ms routing decisions under load, automatic failover when providers timeout or return errors, real-time circuit breaking when PSPs degrade, and metrics collection feeding continuous optimization.

For Erlang/Elixir systems, this maps directly to BEAM VM strengths: lightweight processes evaluate each PSP concurrently, message-passing avoids shared-state contention, supervision trees provide fault isolation, and hot code reloading enables routing rule updates without downtime.

How global smart payment routing works

Smart routing operates as a real-time decision engine layered between your application and PSP APIs.

Transaction ingestion: Payment request arrives with transaction details: amount, currency, card BIN (first 6 digits identifying issuer), customer location, merchant category.

Concurrent PSP evaluation: The system queries fee schedules and authorization rate history for available PSPs simultaneously. In Erlang/Elixir, lightweight processes spawn for each PSP evaluation:

Scoring and selection: Each PSP receives a score balancing cost optimization (lowest total fees including forex) and authorization probability (highest historical success rate for this transaction type). The algorithm might weight cost 60% and authorization rate 40%, adjustable per business priorities.

Routing decision: Transaction routes to highest-scoring PSP. The entire decision, from request arrival to PSP selection, completes in 20-50ms, adding minimal latency to checkout.

Cascading and failover: If the primary PSP returns a soft decline (timeout, technical error, temporary issue), the circuit breaker immediately routes to the secondary PSP. This differs from hard declines (insufficient funds, invalid card) which terminate without retry. Cascading recovers 10-15% of soft-declined transactions that would otherwise be lost.

Weighted distribution: Rather than pure optimization, weighted routing allocates transactions by percentage, 40% to PSP A, 30% to PSP B, 20% to PSP C, 10% to PSP D. This balances cost optimization with processor relationship management, enables A/B testing of new PSPs at controlled percentages (5% to new provider while monitoring performance), and provides resilience against single-provider over-reliance.

Metrics collection and learning: Every routing decision generates telemetry: PSP selected, total cost, authorization outcome (approved/declined/error), latency, decline reason codes. This data feeds back into scoring algorithms. If Stripe's authorization rate for French Visa cards drops from 94% to 89%, the system automatically shifts those transactions to Adyen. If Checkout.com's EUR processing costs decrease, routing adapts to capture savings.

Supervision trees ensure routing continues during partial failures. A crashed Stripe connector process restarts automatically while Adyen and Braintree routing paths remain operational, fault isolation that traditional monolithic architectures struggle to achieve.

When does it make sense to use a global intelligent payment routing solution?

Volume determines ROI. The build/maintain costs of orchestration (internal development or vendor fees) need to be justified by cost savings and revenue recovery.

Strong business case (>5M transactions annually):

• Multi-region operations with significant cross-border volume where local acquiring offers 40-60% cost reduction on international transactions

• Current effective rate >2.5% with optimization potential through multi-PSP strategy

• Authorization rates <95% with soft declines recoverable through cascading

• In-house Erlang/Elixir capability or partnership with specialists like Crafting Software who understand BEAM VM architecture

At 10M transactions monthly ($25M+ monthly volume), orchestration ROI appears within weeks. A 30% reduction on $7.5M annual fees ($2.25M saved) against ~$450K annual orchestration cost delivers 400% ROI in year one.

Marginal cases (1-5M transactions annually):

• Single-region operations with primarily domestic transactions see smaller savings from routing

• Already-negotiated aggressive rates (<2% effective) leave less optimization headroom

• Limited engineering resources require external partnership, adding vendor costs

Poor fits (<1M transactions annually):

• Build/maintain costs ($400K-$800K annually) exceed potential savings

• Payment mix heavily weighted to single method (95%+ ACH/bank transfer) limits multi-PSP benefits

• No engineering capacity and budget constraints prevent external partnership

8 ways to lower your payment processing costs

1. Work with multiple payment service providers (PSPs)

Multi-PSP integration enables immediate cost arbitrage while creating negotiation leverage and geographic optimization opportunities.

Different PSPs excel in different contexts. Adyen typically offers better rates for European transactions through extensive local acquiring. Stripe dominates North American processing with competitive domestic rates. Checkout.com provides strong UK coverage. Rather than accepting one PSP's rates across all transaction types, route each transaction to the PSP offering optimal economics for those specific characteristics.

Real pricing variance example: For Canadian Mastercard transactions, Adyen charges C$0.13 + ~2.0% while Stripe charges C$0.30 + 2.9%. On 100K transactions at $100 average value, Adyen costs C$213K monthly; Stripe costs C$319K—a C$106K monthly difference ($1.27M annually) from routing decisions alone.

Volume commitments work differently with multi-PSP strategies. Instead of promising 100% of volume to one provider for marginal discounts, commit percentages across providers: 40% to PSP A at rate X, 30% to PSP B at rate Y, 30% to PSP C at rate Z. This maintains competitive pressure and flexibility while securing volume discounts from each provider.

Negotiation leverage improves dramatically. Approaching renewals with live integrations and actual performance data from alternative PSPs transforms conversations: "PSP B processes 30% of our volume at 15% lower effective rates for EU transactions. Match their pricing or we shift 60% of European volume by quarter-end." Single-PSP merchants lack this leverage, threats to switch ring hollow without proven alternatives.

Implementation requires 2-3 weeks per PSP integration. Start with two providers, prove value through A/B testing, then expand to 4-5 for comprehensive geographic coverage.

2. Introduce local alternative payment methods (APMs)

Card networks aren't always the lowest-cost option. Alternative payment methods often provide better economics while improving authorization rates in specific markets.

Geographic payment preferences vary dramatically. In Indonesia, DOKU Wallet costs 2.50% while GoPay charges 4.50% (both via Adyen), knowing local APM economics enables optimization. In Germany, girocard and SEPA direct debit offer lower processing costs than international card schemes. Brazil's PIX provides instant settlement at minimal cost. Poland's BLIK, Netherlands' iDEAL, and Nordic region's Swish all deliver better customer economics than cards.

APMs shift fraud liability. Authentication burden falls on the payment method provider rather than the merchant. This eliminates chargeback risk and reduces fraud prevention costs. A transaction processed through iDEAL carries no chargeback exposure, the bank guarantees payment.

Conversion impact justifies APM integration beyond cost savings. Offering preferred local methods boosts approval rates 15-30% in specific markets. German customers seeing SEPA as the default option alongside cards experience familiar, trusted payment flows, reducing cart abandonment.

3. Enable local acquiring and on-us processing

Cross-border processing triggers forex markups (1-3%) that disappear with local acquiring. When issuing bank and acquiring bank exist in the same country, transaction costs drop significantly.

BIN-based routing identifies the card issuer's country from the first six digits. A card starting with 4532 11 indicates a German issuer. Routing this transaction to a PSP with German acquiring capability eliminates cross-border fees.

Cost comparison for a €100 German-to-German transaction:

• Cross-border route: 0.3% interchange + 2.0% acquirer + 2.0% forex = 4.3% = €4.30

• Local acquiring: 0.2% interchange + 1.5% acquirer = 1.7% = €1.70

• Savings: €2.60 per transaction (153% cost difference)

At 100K such transactions monthly, local acquiring saves €260K annually from routing intelligence alone.

On-us processing, when issuer and acquirer are the same institution, reduces costs further. Interchange fees drop significantly or disappear entirely for on-us transactions. Some European markets see effective rates below 1% for on-us processing compared to 2-3% for typical card transactions.

4. Improve your fraud prevention set-up

Chargebacks trigger €15-€100 fees regardless of outcome. At 0.5% chargeback rate on 1M transactions annually, that's €75K-€500K in direct fees. Add lost merchandise and time spent managing disputes, and fraud costs multiply.

Layered prevention reduces chargebacks while lowering processing costs:

Pre-authorization screening: ML models evaluate transactions before authorization, blocking high-confidence fraud before any fees incur. Real-time velocity checks (transactions from the same device/IP in a short timeframe) catch automated attacks. Device fingerprinting identifies previously-flagged devices.

Risk-based 3D Secure: Rather than forcing authentication on all transactions (increasing cart abandonment), trigger 3DS only for transactions scoring above risk thresholds. Low-risk transactions proceed with zero friction. Medium-risk transactions use biometric step-up authentication. High-risk transactions require a full 3DS challenge.

Routing integration: Send high-risk transactions to PSPs with stronger fraud tools, even if marginally more expensive. A transaction flagged 60% fraud probability might cost 2.3% through PSP A with basic fraud detection or 2.5% through PSP B with advanced screening. The extra 0.2% processing cost prevents a €50 chargeback, delivering 100x ROI on the routing decision.

Post-transaction monitoring through Ethoca and Verifi enables proactive refunds on confirmed fraud, preventing chargeback fees entirely.

5. Negotiate with your payment service providers

High-volume merchants (>5M transactions annually) secure meaningful rate reductions through informed negotiation.

Timing matters: Negotiate during contract renewal periods or after 12 months of processing at volume. PSPs invest significant effort in customer acquisition, retaining existing volume is cheaper than replacing it. Use this leverage.

Come prepared: Document your current effective rates by transaction type, geography, and card network. Bring competitive quotes from integrated alternatives showing specific rate differences. Demonstrate routing capability, you can shift volume immediately, not theoretically.

What's negotiable: Gateway fees and acquirer markup are fully negotiable. Refund and chargeback fees can often be reduced or waived at volume. Some PSPs absorb portions of interchange and scheme fees at enterprise scale. Minimum monthly fees, compliance costs, and statement fees are negotiable.

Volume commitments: Guaranteeing minimum monthly transaction counts or values justifies rate reductions. A commitment to process 2M transactions monthly might secure 10-20% rate decreases. Structure commitments with escape clauses, performance minimums, uptime guarantees, rate protection against unilateral increases.

Multi-PSP leverage: "We currently route 60% through you, 40% through competitor B. They've offered 0.25% better rates on EU transactions. Match their pricing or we shift to a 40/60 split by next quarter." Bring actual data on competitor performance and economics.

6. Verify that your MCC is correct

Merchant category codes determine interchange rate structures. Incorrect classification costs 20-40 basis points on every transaction.

Utilities, educational institutions, charities, and government entities receive preferential interchange rates. High-risk categories (gambling, adult content, crypto) pay premium rates. Misclassification in either direction causes problems, underpaying triggers scheme penalties; overpaying wastes margin.

Verify your MCC with each acquiring bank. If classified incorrectly, request reclassification with supporting documentation (business registration, product descriptions, transaction samples). Processing $10M monthly, a 25 basis point correction saves $300K annually.

7. Delayed capture & void authorization

Authorization reserves funds immediately; capture moves funds days later. Fees accrue at capture, not authorization.

Use case: Businesses with high refund rates (pre-orders, custom products, extended delivery times). Authorize at purchase, delay capture until shipping. If the customer cancels before shipping, void the authorization instead of issuing a refund.

Cost impact: Voiding costs only scheme fees (partial). Refunding costs full scheme fees plus refund processing fees. On 10% refund rate at 100K monthly transactions, voiding saves €20K-€40K annually versus refunding.

8. Regularly review your transaction history

Payment analytics reveal optimization opportunities invisible in aggregated reports.

Segment by geography: Identify high-cost cross-border routes where local acquiring would reduce costs. Discover regions where authorization rates lag, indicating need for additional PSP coverage.

Segment by payment method: Find expensive methods with viable lower-cost alternatives. Spot methods with declining authorization rates requiring investigation.

Segment by PSP performance: Track authorization rates by PSP, transaction type, and time period. Detect degradation early, if Stripe's UK authorization rate drops from 95% to 91%, investigate immediately and potentially shift volume.

Segment by decline codes: Distinguish soft declines (retry candidates) from hard declines (terminal failures). Measure how much revenue cascading recovers. Identify systematic issues causing preventable declines.

Monthly review cadence: Export PSP reports, analyze effective rates by segment, identify top 3 cost drivers, implement routing adjustments, measure impact. Consolidating 30+ PSP reports into unified analytics reduces this from days to hours.

Lower your payment processing fees with Crafting Software

Payment routing

Crafting Software builds custom Erlang/Elixir orchestration engines for enterprises processing at scale. Our implementations deliver:

Sub-50ms routing decisions through concurrent PSP evaluation using BEAM VM lightweight processes. Each transaction spawns isolated evaluation processes that query fees, authorization histories, and circuit breaker states in parallel, completing within strict latency budgets.

Zero-downtime rule updates via hot code reloading. When fraud patterns shift or PSP pricing changes, deploy new routing logic in seconds without restarting services or draining traffic. This responsiveness matters during active attacks or competitive rate opportunities.

Automatic failover through supervision trees and circuit breakers. When PSPs degrade, circuit breakers detect the issue within seconds and shift traffic automatically. Supervision hierarchies ensure individual PSP connector crashes don't cascade, failed workers restart while sibling processes continue routing.

Multi-layer caching (ETS + Redis + PostgreSQL) delivering 95% of fee lookups from in-memory stores. L1 cache (ETS) hits complete in <1ms. L2 cache (Redis) handles 2-5ms. Only cache misses hit PostgreSQL. This architecture keeps routing latency minimal even at millions of daily transactions.

Weighted distribution and AI recommendations combining static rules, percentage-based allocation, and machine learning suggestions. The platform continuously analyzes which PSPs perform best for specific transaction patterns and recommends optimizations to payment operations teams.

We've architected systems processing 3.2M daily transactions with 99.95% uptime, delivering 30-40% cost reductions through intelligent routing while maintaining fault tolerance and sub-500ms end-to-end latency.

Payment analytics

Unified analytics consolidate reports across 20+ PSPs into real-time dashboards:

Cost optimization: Compare actual costs against theoretical optimal routing. Quantify savings opportunities by transaction segment. Track effective rates by PSP, region, card type, and payment method.

Authorization rate monitoring: Measure approval rates by PSP, geography, and transaction characteristics. Detect degradation before significant revenue impact. Identify which PSPs excel for specific transaction types.

Geographic analysis: Break down costs and performance by customer country. Guide local acquiring investments and APM priorities based on actual transaction economics.

Real-time alerting: Get notified when PSP costs exceed thresholds, authorization rates drop below targets, or circuit breakers activate. Respond in minutes instead of discovering issues in monthly reconciliation.

Fraud prevention

PCI Level 1 compliant implementations integrate fraud detection throughout transaction flows:

Pre-authorization screening: ML models evaluate risk before routing decisions. High-risk transactions route to PSPs with advanced fraud tools. Low-risk transactions optimize purely for cost.

Network tokenization: Replace card PANs with network tokens, reducing fraud exposure while improving authorization rates 2-4%. Tokens qualify for preferential interchange rates in many cases.

3D Secure integration: Handle Strong Customer Authentication seamlessly. Maintain authentication across cascading, when payments retry through backup PSPs, preserve the original 3DS authentication to avoid multiple customer challenges.

Post-payment monitoring: Integrate with Ethoca and Verifi for confirmed fraud cases. Trigger proactive refunds avoiding €15-€100 chargeback fees entirely.

Team of payment experts

Crafting Software's infrastructure team specializes in Erlang/Elixir payment systems handling millions of daily transactions. We architect solutions that leverage BEAM VM advantages—fault tolerance, lightweight concurrency, hot code reloading—for production payment environments.

Erlang/Elixir expertise: We build payment solutions specifically exploiting supervision trees for fault isolation, GenServer patterns for stateful routing engines, ETS/Redis caching for sub-millisecond lookups, and OTP principles for distributed resilience. This isn't generic development—it's payment infrastructure designed for the unique demands of financial transaction processing.

Regulatory depth: PSD2, PCI DSS, and regional compliance requirements are architectural concerns from day one, not post-deployment fixes. Our implementations include audit logging, consent management, and authentication tracking designed for regulatory review.

Production optimization: Beyond initial deployment, we provide ongoing routing refinement based on production data. As PSP performance shifts and transaction patterns evolve, we analyze metrics and recommend routing adjustments delivering continuous cost reduction.

If your enterprise processes millions of transactions and needs Erlang/Elixir expertise to build fault-tolerant orchestration reducing costs while maintaining <500ms latency, contact Crafting Software for payment infrastructure solutions proven at scale.

Conclusion

Smart routing transforms payment processing from a fixed cost into an optimization lever. The economics are compelling: enterprises processing 10M+ transactions monthly see 30-40% cost reductions within weeks of implementation, translating to $600K-$800K in annual savings while simultaneously improving authorization rates through intelligent failover.

The architectural requirements—sub-100ms routing decisions, fault-tolerant failover, zero-downtime rule updates—align precisely with Erlang/Elixir's BEAM VM capabilities. Lightweight processes enable concurrent PSP evaluation. Supervision trees provide automatic recovery when connectors fail. Hot code reloading deploys routing adjustments during active business without service interruption.

For enterprises at scale, the decision isn't whether to implement intelligent routing—it's whether to build internally or partner with specialists like Crafting Software who understand both payment infrastructure and distributed systems architecture. The ROI timeline measured in weeks, not quarters, makes this one of the highest-impact optimizations available to payment operations teams.

The payment landscape continues evolving. PSP pricing fluctuates. Authorization rates shift. New processors enter markets with competitive economics. Smart routing isn't a one-time implementation—it's continuous optimization infrastructure that adapts as conditions change, protecting margins while maintaining reliability.

Smart Routing Payment FAQ

1. At what transaction volume does intelligent routing justify the implementation cost?

The breakeven point typically occurs around 5M transactions annually. Below 1M transactions, vendor platform fees ($200K-$300K annually) or build costs ($400K-$800K year one) exceed realistic savings. Between 1-5M transactions, ROI depends heavily on your current effective rate and geographic mix. Above 10M transactions, the business case becomes compelling—30% reduction on $7.5M annual fees ($2.25M saved) against ~$450K annual orchestration cost delivers 400% ROI in year one. The key variables are current effective rate (higher rates = more optimization opportunity), cross-border volume percentage (local acquiring offers 40-60% savings), and authorization rate gaps (cascading recovers 10-15% of soft declines).

2. Can we implement smart routing without disrupting our current payment infrastructure?

Yes. Implement in shadow mode first—route a copy of each transaction through the orchestration layer while your production flow continues unchanged. This validates routing decisions without risk. After 2-4 weeks of shadow testing, begin gradual rollout: 5% of traffic through orchestration in week 1, 25% in week 2, 50% in week 3, 100% in week 4. Monitor authorization rates, latency impact, and cost metrics at each stage. Rollback criteria should be clear—if authorization rates drop >2% or P95 latency increases >100ms, automatically revert to single-PSP flow. For Erlang/Elixir systems, feature flags control routing percentage without code deployment. The entire migration typically completes in 6-8 weeks with zero production incidents when properly executed.

3. How do we maintain PCI compliance when integrating multiple PSPs through an orchestration layer?

Your orchestration engine must never store raw card data—only tokenized credentials. Implement immediate tokenization at the client side using PSP-provided SDKs (Stripe.js, Adyen Drop-in) before data reaches your servers. Your orchestration layer handles only tokens, never PANs, reducing PCI scope to SAQ-A (simplest compliance level). Each PSP integration uses their provided tokens rather than sharing tokens across providers. For your orchestration infrastructure: encrypt all data at rest (AES-256), use TLS 1.3 for data in transit, implement role-based access controls limiting who can view transaction data, maintain detailed audit logs of all routing decisions and PSP interactions, and conduct annual penetration testing. Most third-party orchestration platforms (Primer, Payrails) are PCI Level 1 certified, transferring compliance burden to them. For custom-built solutions, work with PCI DSS experts like Crafting Software who architect compliant orchestration from day one.

4. What happens when our primary PSP experiences an outage—how quickly does failover actually work?

Circuit breakers detect PSP degradation within seconds and shift traffic automatically. The typical flow: orchestration layer sends payment request to primary PSP with 5-second timeout, PSP fails to respond or returns 5xx error, circuit breaker increments failure count for that PSP, after 5 consecutive failures within 60 seconds, circuit transitions from CLOSED to OPEN state, all subsequent transactions immediately route to secondary PSP without attempting primary, circuit remains OPEN for 30 seconds (cooldown period), circuit transitions to HALF_OPEN and allows one test transaction through, if test succeeds, circuit returns to CLOSED; if test fails, circuit returns to OPEN for another 30-second cycle. The critical metric: time from first failure to failover completion is typically 15-30 seconds. During this window, 5 transactions may fail before routing shifts. In a production system processing 100 transactions/second, that's 1,500-3,000 affected transactions during failover—versus complete outage for however long the PSP remains down (often hours).

5. How do we handle routing for transactions that require specific PSP features like recurring billing or fraud tools?

Implement context-aware routing rules that consider transaction metadata beyond just cost. Tag transactions with attributes: subscription_recurring: true, fraud_risk_score: 0.75, requires_3ds: true, customer_vip: true. Routing logic evaluates these tags: recurring transactions route to PSPs with robust subscription management (Stripe for US, GoCardless for EU), high-risk transactions route to PSPs with advanced fraud tools even if marginally more expensive (preventing €50 chargeback justifies 0.3% higher processing cost), 3DS-required transactions route to PSPs with seamless authentication flows, VIP customers route to PSPs with highest authorization rates regardless of cost (optimizing for customer experience over margin). The routing engine applies business rules first (hard constraints like "all subscriptions use Stripe"), then optimizes within remaining options for cost and authorization rate. This context-aware approach balances multiple objectives—cost, fraud prevention, customer experience, feature requirements—rather than optimizing purely for economics.