Travel margins are thin; cash friction makes them thinner. A pan-EU booking aggregator came to me with the usual symptoms: chargebacks on card-only checkout, angry customers waiting weeks for refunds on cancelled legs, and suppliers complaining about slow remittance. We rebuilt payments around local collections, automated split settlement, and a refund SLA the support team could say out loud.
The reality of travel cash flows
Every cart is a mini-funds-flow puzzle: airline seats, hotel nights, ancillaries, the platform fee. One provider cancels and the whole booking becomes a refund, re-issue, or partial credit. If you collect cross-border on a single EU MID and pray, you’ll hemorrhage on scheme fees, disputes, and reconciliation.
The design goals (from a CFO’s notebook)
- Collect locally, in the shopper’s currency, with acquirers who know the corridor.
- Split settlement on rails, not in spreadsheets: your fee now, supplier share by rule.
- Refund within 24 hours for bank-transfer methods and “as fast as schemes allow” for cards—communicated up front.
- Minimize false declines with multi-acquirer routing and method diversity (cards, bank transfers, wallets).
- Make reconciliation a button, not a project.
Checkout that adapts to the booking
The page priced in the shopper’s currency and prioritized high-approval methods per corridor: domestic card routing in the EU/UK, local bank transfers where adoption is strong, and selective wallets for mobile flows. We didn’t drown users in options; we nudged the cheapest-to-clear, fastest-to-refund methods first.
Split settlement without spreadsheets
Each booking produced a rules object: platform fee, supplier shares, taxes, refundable components, and release triggers. On capture, funds landed into a virtual booking ledger. The engine:
- Released the platform fee immediately to the operations account.
- Queued supplier remittances on defined triggers: ticketing confirmed, stay consumed, or day-count post-check-out.
- Held refundable portions in a dedicated liability bucket so that cancellations didn’t require treasury gymnastics.
This is how you stop refunds from nuking your week.
Refunds that customers feel—fast
For bank transfer methods (local EU rails, UK FPS), refunds posted within hours during banking windows; cut-offs were codified so agents could quote value dates. For cards, we routed to acquirers with solid travel risk models and used 3DS only when risk scores demanded it, not by blanket policy. Refunds piggy-backed on original authorizations to shorten timelines.
Disputes: win fewer fights by starting fewer wars
We reduced chargebacks by:
- Transparent checkout descriptors and itinerary-specific references.
- Proactive messaging on schedule changes with one-click refund or rebook, before a dispute felt “necessary.”
- Clean evidence packs when a dispute landed—PNR, traveler acceptance, usage logs, refund trail—pulled automatically from the booking ledger.
Dispute win rate rose, but more importantly dispute count fell.
Supplier trust is a cash advantage
Suppliers saw predictable value dates and clean remittance files that mapped booking ranges to amounts. Exceptions (no-shows, partial refunds) were visible with reasons and supporting events. When suppliers know when they get paid—and why something changed—they negotiate less and collaborate more.
Treasury: less drama, more math
Collections in EUR/GBP and other currencies swept to working accounts by thresholds. We avoided FX on every micro-flow; we netted where we could, converted on rules (month-end, payroll, supplier runs), and tracked execution against a benchmark. Refund pools were pre-funded so agents never saw “insufficient balance for refund.”
What moved the needle
- Approval rate: up 8–12 pp on key corridors after local acquiring and multi-acquirer routing.
- Refund SLA: <24h for bank transfers, card refunds trending to the fastest scheme windows.
- Chargebacks: down 40–60% after descriptor cleanup, proactive comms, and data-driven 3DS.
- Reconciliation: automated close on >95% of bookings; finance focused on true anomalies, not hand-matching.
Numbers vary by mix, but the direction is repeatable.
Red flags to manage, not fear
- Wallet sprawl adds support debt if you can’t refund back to the original method; enable selectively.
- Supplier chargebacks (when they cancel late) will exist; keep the ledger’s evidence immaculate.
- Holiday cut-offs matter; publish and enforce them so agents stop guessing.
The short build that stuck
Eight weeks was enough: two acquirers, one bank-transfer provider, a booking-ledger service, and a payout engine. The hero wasn’t the vendor list—it was operational clarity: who gets paid, when, on which rail, with what evidence.