Travel Payment Processing: Reserves, Future Delivery, and the Unflown Ticket Problem

The chargeback ratio is under half a percent. The refund rate is normal, the fraud numbers are clean, and the account has run two years without incident. Then a strong booking season arrives, forward bookings double, and the processor raises the rolling reserve and stretches funding by a week. Nothing went wrong. The reserve is not a punishment for past performance, it is collateral against trips that have not happened yet, and a good quarter makes that pile bigger. This is a guide to how travel gets underwritten, why supplier risk becomes your risk, and what a stack built for future delivery looks like.

You are underwritten on the gap, not the ratio

Most high-risk verticals are priced on dispute behavior. Travel is priced on time.

When a customer pays in March for a trip in October, the acquirer takes on seven months of contingent liability. If you fail before October, every open booking converts to a services-not-provided chargeback that the acquirer pays, with the reserve as the only asset it controls.

This is why two merchants with identical chargeback ratios get very different terms. A same-week hotel booking engine carries days of exposure, while a tour operator selling escorted trips a year out carries a year of it, against a book that grows every time marketing succeeds. Underwriters model your booking window, average ticket, supplier concentration, and balance sheet, then size the holdback to the undelivered part of the book.

So arguing about your chargeback ratio in an underwriting call does not move the reserve much. Producing a forward-booking ledger does.

The ledger the reserve conversation turns on

Most travel merchants can produce a revenue report but not an exposure report. Revenue says what you earned, exposure says what the bank is on the hook for if you stop trading tomorrow, and only the second sets a holdback.

The ledger worth building tracks total collected against undelivered travel, that figure by departure month, average days to departure, and supplier concentration inside it. Add the share already paid through to suppliers, because money you have forwarded is money the acquirer cannot claw back from you.

Report it monthly without being asked and the reserve stops being a category default and becomes a negotiated number on your file. Merchants who cannot answer "how much have you sold that has not travelled yet" get the conservative default.

The supplier failure that lands on your account

The second structural fact here is that fulfillment is usually not yours to control.

An agency or OTA sells inventory owned by airlines, hotels, and tour operators. When one of those suppliers stops operating, cardholders do not dispute against the supplier, they dispute against the name on their statement. An upstream collapse arrives at your MID as services-not-provided chargebacks on transactions that were legitimate when they settled, at the full ticket price rather than your margin. Supplier concentration is therefore an underwriting question: one carrier's inventory into one region is a single point of failure your bank co-signs.

Three mitigations are real, and none covers all of it. Diversify suppliers so no single failure takes out a meaningful slice of the book. Where the model allows, make the airline or hotel merchant of record for the travel and yourself merchant of record only for your fee, moving the contingent liability to the party that delivers. And keep cancellation and pass-through terms click-accepted at checkout, because in a supplier-failure representment your terms are most of the evidence you have.

Three tiers, and what each one costs you

	Aggregator	Domestic dedicated travel MID	Specialist or offshore acquiring
Boards long-window travel	Sometimes, until volume grows	Yes, with financials and accreditation	Yes, specialist appetite
Underwriting focus	None at signup	Booking window, suppliers, balance sheet	Varies, often exposure-capped
Reserve	None, then a 90 to 180 day hold	Rolling, sized to the unflown book	Often higher, sometimes upfront
Response to a volume spike	Automated hold	Reserve resize with notice	Negotiated, relationship-driven
Realistic role	Short-window volume only	Primary processing	Cross-border volume, redundancy

Aggregators are the specific trap here, because the failure mode is timed against you. Stripe, PayPal, Square, and Shopify Payments underwrite after volume arrives, and their risk models react to exactly the signals a healthy travel business generates: seasonal spikes, large tickets, and a growing forward book. The hold lands during peak season, precisely when you owe suppliers deposits, so a cash-flow problem and a processing problem arrive the same morning. That deferred review is what the high-risk underwriting checklist exists to get ahead of.

A dedicated MID with an acquirer that has real travel appetite trades convenience for predictability. Underwriting is slower and reserves are explicit, but a reserve you negotiated and can forecast is a manageable cost, while a surprise 120-day hold on peak volume is an extinction event. How those holdbacks are sized and released is in the rolling reserves explainer.

Specialist and multi-region acquiring earns its place on cross-border volume, at the cost of higher pricing and another compliance relationship.

Accreditation and coding

Travel agencies typically run under MCC 4722, and acquirers want to know whether you hold ARC or IATA accreditation and whether you are merchant of record for the travel or only for a service fee. Accreditation is not required, but an accredited agency has already been vetted by a party the bank recognizes, and without it the burden falls on your own financials and booking-window data.

Whatever the answer, describe the business accurately at boarding. Selling long-window packages under a description implying immediate delivery gets accounts terminated rather than repriced, and misrepresentation at onboarding is the reason most likely to put the principal on the MATCH list for five years.

The disputes are recognition and expectation gaps

Travel's dispute mix is unusual because most of it is not fraud. The largest cluster is cancellation conflict, where the customer cancels under the policy they believe applies and you refund under the supplier's stricter one. Then services not provided, after a schedule change or a failed supplier, and trip quality, where the delivered room or itinerary did not match the listing. The last is pure recognition failure: a booking made in February, read on a statement in March, under a brand the cardholder does not remember.

Descriptors therefore do more work here than in almost any other vertical, and one naming the destination or the trip, matched to confirmation and pre-trip emails from the same brand, removes a large share of "I do not recognize this" disputes before they start.

Your dispute clock is not the one you think it is

The detail that catches travel merchants is when the filing window opens. For most ecommerce a cardholder's dispute rights run from the transaction date, so a settled month is a closed month. For services not provided, the networks generally run the window from the date the service was expected rather than the date it was paid for, with a longer outer limit in some travel scenarios. Confirm current figures with your acquirer, because the networks revise these tables.

A March booking for an October trip is therefore still disputable well into the following year, so your exposure is the whole undelivered book plus a tail behind it. Merchants who count disputes against the month a transaction was charged conclude their rate is fine, while the ratio that matters, disputes counted against the month they post, drifts toward a monitoring program. Measure it the way the networks do, using the thresholds in the Visa VAMP and VDMP reference, and reserve against the tail rather than the quarter.

Prevention alerts are the other high-value control. Wiring Ethoca and Verifi RDR into the booking system rather than the finance inbox means an early dispute on an untraveled itinerary can trigger a cancellation and refund instead of a full-severity chargeback.

For representment, the winning file is fulfillment evidence: timestamped click-accepted cancellation terms, itinerary delivery confirmation, supplier check-in or boarding data, and any post-trip customer contact.

Architecture for delivery-horizon risk

Split volume across at least two acquiring relationships, and where you can, split by booking window rather than at random. Short-window bookings can carry ordinary terms, while long-window packages are the exposure that drives reserves, so separate facilities stop your highest-risk book from pricing all of your volume. Keeping that routing maintainable instead of a pile of conditional logic is what the payment orchestration explainer covers.

Then store card credentials above the acquirer layer, in a vault you control. Tokenization does double duty in travel: it keeps a repeat customer base portable when you change acquirers, and it lets you charge balances and incidentals months later without asking for the card again. An agency whose credentials sit inside the processor it wants to leave cannot move without losing the ability to collect a final balance.

Practical takeaway

Travel is not hard to bank because travel merchants dispute badly. It is hard to bank because you collect today for something delivered months from now, because a supplier's failure arrives as your chargebacks, and because growth increases the acquirer's exposure instead of reducing it. The merchant who lasts boards with an acquirer that already runs travel, describes the booking window honestly, negotiates the reserve against forward-booking data, splits short and long-window volume across two banks, and keeps credentials in a neutral vault so an acquirer change is an inconvenience rather than a rebuild. The merchant who fails runs peak season through an aggregator and finds out in July that the money for supplier deposits is on a 120-day hold. For a walkthrough of what a durable travel payment stack looks like at your volume and booking window, apply for an architecture review.