Most ancillary programmes have run out of new things to sell. Seats, bags, priority boarding, insurance and car hire are instrumented, tested and sitting close to whatever ceiling the traffic allows. The next percentage point on any of them is expensive to buy.
The visa line is the shelf space nobody has touched, largely because travel brands treat border documents as an administrative nuisance and never as a revenue line. It behaves differently from everything else on that shelf, and the differences are worth setting out properly before anyone models a number from them.
Where the ancillary shelf stands
Ancillary revenue is everything a travel brand sells alongside the trip itself: the seat assignment, the extra bag, the insurance policy, the lounge pass, the car waiting at the other end. It has absorbed most of the past decade's commercial attention for an obvious reason, which is that it lands closer to the bottom line than the fare does.
Two features run through nearly all of it. The first is a cost of goods. A seat assignment consumes inventory that had another use. A checked bag carries weight, handling and a lost-luggage tail. An insurance policy has an underwriter behind it and a claims ratio in front of it. The second is an operational path for when the product goes wrong, staffed by real people at an airport or in a contact centre. Both sit between the gross figure on the quarterly slide and the contribution finance eventually books.
There is also a ceiling on attention. A booking flow can carry only so many upsells before each new one starts taking conversion from the ones already there. This is why mature ancillary teams spend their time on merchandising and placement instead of new products, and why a genuinely additive line is worth more than its face value suggests: it does not have to be paid for out of an existing attach rate.
Why the visa moment behaves differently
An authorisation is a condition of travel. A traveler with a Vietnamese itinerary and a British passport is going to deal with the e-visa whether or not anyone sells it to them, because the alternative is being turned away at the gate. Nothing else on the ancillary shelf works that way. Nobody has to buy the lounge pass.
The demand therefore exists before any merchandising happens. The commercial question is only where it gets satisfied. That is a distribution problem, not a demand problem, and today the distribution belongs to a search engine. The traveler finishes the booking, types the destination and the word visa into Google, and arrives at whichever site bought the ad. A meaningful share of those sites are lookalikes charging three or four times the government fee for an authorisation, and doing very little for the difference.
The bill for that arrives at the travel brand regardless. Travelers who overpay complain to the company whose name they recognise, which is the one that sold the flight. Travelers whose document never arrives ask for a refund from the same place. When the document is missing entirely, the cost is denied boarding: the rebooking, the carrier fine that follows in many jurisdictions, and a public review that outlives both.
So the line already exists in your book. It is currently being monetised by someone else, and the service failures are being attributed to you. That combination is unusual enough that it is worth treating the visa moment as a category on its own rather than as another item to slot into the upsell carousel.
Four commercial models, compared honestly
There are four commercial shapes available in this market. They differ on one axis, which is who legally sells to the traveler. Price control and refund exposure both follow from that single answer.
On revenue share, the traveler buys from SimpleVisa. We are merchant of record, the service fee is $39 per application, and the partner takes a share of it on volume terms. There is no cost of goods on the partner side and no operational path either: government submission, traveler support, refunds and chargebacks all sit on our licence, and no passport data reaches the partner's systems. Starting costs nothing. The trade is that our name appears at checkout, because card networks require the merchant of record to be named, and the traveler price is held at the $39 list fee.
Wholesale runs through Desk, built for agencies and agency networks selling at a counter or over the phone. The agency buys applications at $29, falling to $24 between 100 and 500 a month and $19 above that, then sets whatever retail price its market supports. Margin is the spread, and on a counter that sells advice alongside documents the spread can be wide. The agency is merchant of record, so it keeps the customer relationship and the refund conversation that comes attached to it. Seats are free and unlimited, so putting the whole counter on it costs the agency nothing.
The Platform licence is a full white label on the partner's own payment gateway and under the partner's own brand, with our name absent from the flow. It runs at a $5,000 monthly minimum drawn down against $12 per application. The partner keeps the whole traveler price above that $12 and carries the fixed floor in months when the volume does not show up. This suits groups with existing checkout infrastructure and a reason to own the payment relationship end to end.
The fourth shape is the affiliate programme, which most consumer visa services offer. The traveler clicks out to the provider's site and the referrer receives a slice of the service fee. It is the least work of the four and the least ownership: no control over the price shown or the catalogue behind it, no data about who bought what, and a traveler who has left your site at the moment they were most engaged. For a book where the visa moment is genuinely rare, that can still be the sensible answer.
One line does not move across any of these. Consular fees are collected at cost and remitted to the government at cost. They never enter anyone's margin, and any model that treats them as a revenue line should be read carefully. Which of the four fits your business is mostly a merchant-of-record question, and that question is unpacked at length in merchant of record, explained for travel teams.
Partners who have run the revenue-share model for years tend to describe it in flat terms. Stefan Born, Director Ancillaries at Etraveli:
"We like their professional team and easy to integrate API. After several years of partnership, we are very pleased with the service and revenue generated."
What actually drives the number
The arithmetic is short. Applications are attach rate multiplied by bookings, the service-fee pool is applications multiplied by $39, and the partner's share of that pool is set by volume terms. It is worked line by line, on a book of 60,000 bookings, in what a 4% attach rate looks like on 60,000 bookings. Your own version of the sum is on the revenue calculator on our home page, and every rate underneath it is published on the pricing page.
Before any of that, the ceiling is set by the book itself. A visa requirement is a function of the passport and the route, transit stops included, so a European short-haul book has a thin eligible slice and a long-haul book out of a mixed-nationality market has a thick one. The same attach target is a different amount of work in each case. Look at what your top twenty route and nationality pairs actually require before committing to a figure.
After that, the size of the number is decided almost entirely by the booking flow around it. Four inputs do most of the work:
A first integration is usually one placement on the confirmation page, and it lands in low single digits against total bookings. That figure is a statement about placement rather than about the product, which is the most useful thing to understand before setting a target for year one.
What it costs to launch
On revenue share, nothing. Sandbox keys arrive with the account, with no sales call and no procurement cycle in the way, and the traveler-facing surface is either a hosted checkout on your own subdomain or the drop-in Elements components, which are free on every model including the Platform licence. A single placement is an afternoon of front-end work, and it produces a real attach number on your own traffic inside a quarter, which is cheaper than modelling one.
Coverage is the other launch question, and it has two numbers. Requirement data covers 190+ destinations, which is what the checker and the API answer from. More than 80 of those are processed end to end, meaning application, government submission and status back. The second number decides what you can put in a checkout today, and it is the one to ask any provider for.
GET /v1/requirements takes a nationality and a full itinerary, transit stops included, and returns the documents, government fees and processing times. Partners sending applications through us query it as often as they like at no charge, because the applications fund the coverage team that keeps it current.
The integration itself is deliberately small. The partner passes a route and an email address. Travelers upload their own documents directly to us, which is why the visa line rarely turns into a security review: there is no new class of personal data arriving in the partner's stack, and nothing new to store, encrypt or delete on request.
Questions to ask any provider
Most of the risk in this category is commercial before it is technical, and it surfaces in five questions. They are answerable on a first call, and the answers are more predictive of how the partnership goes than any demo.
Who is the merchant of record? There is one answer per transaction. It decides whose name appears on the traveler's statement, who issues refunds out of whose balance, and whose compliance perimeter the card data sits inside. Vagueness here is itself an answer.
Who absorbs a refused application? The consular fee has already gone to the government by then, and the traveler will still ask for their money back. Get it in writing: whose balance covers that, and under what policy.
Is the consular fee marked up? Ask for the government's published fee alongside the price the traveler sees. Any gap between them is a markup, and your customers will eventually find it. Providers who separate their service fee from the government charge can show you both without preparation.
Can you start without procurement? Whether a test key requires a signed contract tells you how the rest of the relationship will run. A provider that gates a sandbox behind a demo will gate other things behind other meetings.
What happens on a multi-stop itinerary? Hand over a real route with a transit point that carries its own requirement, and a passport that triggers it. This single test separates a tool that informs from one that fulfils, and it is where most of them have nothing to say.
Two more questions are worth asking inside your own business. Whether nationality is captured before the payment page decides how much of this revenue is addressable at all, and whether anyone currently owns border documents decides who will run the line once it exists. In most travel companies the honest answer to the second is nobody, which is precisely why the moment is still leaking to a search engine and the revenue is still on the table.