Airline Distribution: 200,000 Searches for One Ticket

Eric Marketts

In the early online era, an airline saw somewhere between one hundred and two hundred searches for every ticket it sold. Today it is closer to ten thousand to twenty thousand. OAG's projection for an agent-driven world is two hundred thousand. Same one ticket at the end of it.

Steph Nell

Two hundred thousand looks for one sale. Somewhere a pricing engine is quietly filing a grievance .

Eric Marketts

And it has a case.

Steph Nell

The entire economics of distribution rests on an assumption that just stopped being true.

Eric Marketts

Welcome back to The V1 Airline Retailing Report. I'm Eric Marketts.

Steph Nell

And I'm Steph Nell. We're back to the regular weekly format after four straight Mondays of something different.

Eric Marketts

We are. For four weeks we ran a special season called State of the GDS. Part one, the foundation. Part two, Amadeus, renovating from strength. Part three, Sabre, selling a limb to survive. Part four, Travelport, the rebuild. If you haven't heard it, start at part one. It is not a news series. It is the foundation under every topic in airline retailing right now, and it was built to still make sense a year from now.

Steph Nell

And the season ended on one line. Same machine, three bets. This week those three bets showed up in the market at the same time, and you can finally see what they are actually aimed at.

Eric Marketts

Three narratives this week. Like most weeks, they are not a list. They are one argument.

Steph Nell

For anyone new — every Monday we take the developments that actually move airline and travel distribution and run each one through the bull case, the bear case, and Eric's critical take. Not the press release. The strategic read. Sources are in the episode details.

Eric Marketts

Narrative one. The GDS oligopoly stopped running the same playbook. Amadeus, Sabre, and Travelport are now placing three genuinely incompatible bets, and we will show you why none of them is actually a bet on the pipe.

Steph Nell

Narrative two is a follow-up, and I think it is the most important thing we do this week. In episode eight we said the look-to-book economics were breaking under AI agents, and we left a watch item. Eight weeks later we can grade it. Somebody answered. Not the somebody we expected, and not the way we hoped.

Eric Marketts

And narrative three. The order management system is quietly being named the new center of gravity in retailing, right as a wave of PSS contracts comes up for renewal. The industry is conceding something it spent thirty years denying.

Steph Nell

The through-line. All three are the same story at different altitudes. The pipe is commoditizing. The interface is dissolving. And value is migrating up into the offer layer — which turns out to be two jobs, not one. Building the offer, and owning the shelf it lands on. Almost every move this week is a company deciding which of those two it can actually win.

Eric Marketts

Watch the offer layer, not the booking share. Let's get into it.

Eric Marketts

Here are the facts, clean. Travel Distribution News laid out the state of the three global distribution systems, and the headline is that they have stopped being interchangeable. On the widely reported estimates, Amadeus holds roughly forty percent of global GDS share, Sabre roughly thirty-five, Travelport roughly a fifth — the same distant third we laid out in part four. In February, Amadeus acquired SkyLink, a conversational booking system already running tens of thousands of AI-driven bookings, and it is steering Altea customers toward its own offer and order tooling. In May, Sabre alleged publicly that Amadeus holds a dominant monopoly position in passenger service systems. Sabre is also running the consumer play — Mindtrip with PayPal — and says more than thirty partners are piloting its agentic APIs. And on July first, Travelport expanded TripServices to more than four hundred European agencies, put fifty million dollars behind it, and integrated Anthropic's Model Context Protocol. Its API-driven transaction share has gone from about forty-three percent of customer volume in 2022 to about sixty-three percent now.

Steph Nell

So three companies, three completely different answers to the same question. Give me the bull case, because divergence is not automatically bad.

Eric Marketts

It isn't, and this is a real bull case. For about a decade the honest description of this sector was "all three do roughly the same thing, slowly, and compete on rebate size." That is a terrible market for a buyer. What we have now is competition on architecture. Want an open pipe with clean APIs an agent can call? That is a Travelport conversation. Want an integrated stack where the offer engine, the order layer, and the distribution reach come from one vendor? That is an Amadeus conversation. Genuinely different products, genuinely different risk profiles, and you can finally choose on strategy instead of on price. Real architectural choice here is overdue by about fifteen years.

Steph Nell

And the bear case is that everybody downstream now pays for that choice .

Eric Marketts

Every bit of it. Travel management companies, online travel agencies, aggregators, corporate booking tools — they used to maintain three connections to platforms that were, functionally, the same shape. Now they maintain three connections to platforms that are diverging on purpose. That is not a one-time integration cost. It is a permanent, compounding maintenance burden, and it lands hardest on the smallest sellers, who can least afford three engineering tracks. The stated goal of every modernization program in this industry is to reduce the cost of intermediation. In the short run, this raises it.

Steph Nell

Nothing says modernization like tripling your integration backlog . Give me the critical take.

Eric Marketts

None of these three bets is about the pipe. Everyone in that room knows the pipe is a commodity. But I want to be precise here, because the easy version of this take is wrong, and the correction is the actual insight. They are not all fighting over the same thing. Above the pipe there are two different jobs, and only one of them is a fight all three are in.

Steph Nell

Separate them for me.

Eric Marketts

Job one is building the offer. The airline's own dynamic, personalized, machine-readable answer to a specific request. Amadeus does that with Nevio and Altea. Sabre does it with SabreMosaic — its own language is offer management that lets airlines define, manage, and retail their content. Job two is owning the shelf. Take offers built by many different airlines, normalize them into something comparable, enrich them, and decide what the buyer actually sees. Different job, different skill, and it only happens after the offer exists.

Steph Nell

Careful, because that is too clean and you know it. Sabre is not only building. Look at the actual product line. SabreMosaic is the airline platform, yes. But there is also SabreMosaic Travel Marketplace, and that is a seller-facing product — NDC content from more than forty carriers, so an agency or an online seller does one integration instead of airline by airline, with servicing on the back end. Normalizing many airlines' content into one comparable set that a seller can work in. That is a shelf, by your own definition, sold under Sabre's own name. Then add Mindtrip and PayPal on the consumer end. Sabre is in both columns. So is Amadeus. You have just drawn a map that two of the three refuse to sit still on.

Eric Marketts

That is fair, and I would rather say the honest thing than the tidy thing. Anybody with scale plays both. Amadeus plays both. Sabre plays both. This space is genuinely complicated — it is the "it's complicated" relationship status of enterprise software, and anyone handing you a clean three-box diagram is selling you something. So the useful question is not which column a company sits in. It is which column it is betting the company on.

Steph Nell

Then bet them.

Eric Marketts

Amadeus is not choosing, because it does not have to. It has the balance sheet to fund both, and its whole thesis is that owning construction and distribution together is the moat. Sabre is hedging — and hedging is a real strategy, not a failure. It is what you do when you are mid-rebuild, carrying debt, and cannot fund both at full strength. So it litigates to protect its position in construction and pushes hard on the shelf at the same time. Travel Marketplace, agentic APIs, thirty-plus partners, Mindtrip. Travelport is the only one that cannot hedge. No airline PSS, no construction business to protect, and a two-billion-dollar debt wall in 2028. One column available. All in.

Steph Nell

And Travelport having no construction business — weakness, or the point?

Eric Marketts

I think it is the point, and here is why . Both jobs sit above the pipe, but only one of them is what an AI agent actually talks to. An agent does not ask an airline to construct a bespoke offer out of nothing. It asks a question and expects a comparable set back. That is a shelf request. Which is why the market share numbers everybody still quotes — forty, thirty-five, roughly a fifth — are measuring the wrong thing twice over. They measure volume moved last year, and they measure it on the pipe. Travelport is dead last on that scoreboard and has sixty-three percent of its traffic on modern APIs, the highest disclosed number of the three. Smallest share of the thing that is commoditizing. Furthest along on the thing that is not. Having nothing to protect is not always a disadvantage.

Steph Nell

So if you take one thing out of this segment, take this. Stop asking who is the biggest GDS. Ask what each one is betting on, and whether it can afford the bet. Amadeus is betting it can own both jobs. Sabre is betting it can stay in both long enough to finish rebuilding. Travelport is betting the shelf is the job that matters, because it does not get a second bet.

Eric Marketts

And all three still have to answer a machine that does not get tired. Which is where our follow-up comes in, because eight weeks ago we asked who was going to govern that, and now we have an answer.

Eric Marketts

This is a callback, so let me set it up for anyone who missed it. Episode eight, back in June. We covered Skift's reporting on a traveler named Kiruba Shankar, who wanted one airline ticket, pointed an AI coding agent at the Etihad website, and pulled eight hundred eighty-one thousand, seventy-six fare options. Every date in a range, every stopover, every route combination. For one ticket. Our argument then was that this was a cost imposed on the airline without a commercial relationship and without permission.

Steph Nell

And we closed that segment with a specific watch item. I want to read it back, because this is what a follow-up is for. We said the signal that the industry had finally internalized the problem would be the first airline to publish an explicit AI traffic policy. Rate limits. API terms governing agent access. Or pricing tiers for machine-generated search.

Eric Marketts

Eight weeks. So how did we do.

Steph Nell

Partial credit, and not from where we were looking . Two things happened. First, OAG put a curve around what was a single anecdote — one hundred to two hundred searches per ticket in the early online era, ten to twenty thousand today, up to two hundred thousand in an agent-driven world. That reframes episode eight entirely. Kiruba Shankar was not a stunt. He was an early sample of the mean. Second, and this is the closest anyone has come to what we asked for, Air France-KLM is filtering what an agent gets back rather than letting every permutation spill. Not a published policy. But a carrier deciding, deliberately, what a machine is allowed to see.

Eric Marketts

And the answer we did not expect.

Steph Nell

Amadeus. And it did not build a policy. It built a workaround. Amadeus is betting on precomputed fares — answering agent-scale search volume by serving prepared answers instead of constructing offers lye'v.

Eric Marketts

Give me the bull case on that, because it is a serious piece of engineering and I do not want to be glib about it.

Steph Nell

It is serious, and there is a real bull case. Precomputation is how you survive the next twenty-four months. The volume is coming whether you like it or not, your pricing engine was sized for humans clicking, and serving a prepared answer is how you stay up. It also opens the better door. Once you accept that a machine will ask you two hundred thousand times, you have to decide who is asking, what they are entitled to see, and what it costs them. That is an authenticated, metered, contextual offer API, and it is the most durable control point in distribution because it sits upstream of every channel. Episode eight framed this as a cost problem. Right frame in June. Small frame now. This is a control problem.

Eric Marketts

And the bear case, which I suspect is the reason for the word retreat in the segment title.

Steph Nell

It is. A precomputed fare is, by definition, not a dynamic, personalized offer. You cannot precompute context. You cannot precompute who is asking, what they bought last time, what the aircraft looks like tonight, or what bundle would actually convert for this traveler. So the first significant industry answer to the agent era is a partial step back toward the cached-fare world that NDC and Offer and Order existed to escape. Twenty years of work to make the offer live and specific, and the response to machine-scale demand is to prepare the answers in advance. And notice the irony we flagged in episode eight has not gone anywhere. Richer content is heavier to serve. The carriers furthest along on modern retailing still have the most expensive search to run at machine scale. Precomputation relieves the symptom by undoing some of the thing they paid for. Twenty years spent making the offer live and personal, and the industry's answer to the future is a very sophisticated leftovers drawer. And nobody has published the policy. Eight weeks on, the watch item is still open.

Eric Marketts

So what is the take.

Steph Nell

The industry is optimizing the cost of the look instead of pricing it. That is the difference between episode eight and today. In June the question was who pays for the search. The answer the market is converging on is that the airline pays, and it pays by making the search cheaper to serve. Engineering answer to a commercial problem, and those have a shelf life. Two hundred thousand searches per sale does not get solved by caching. It gets solved by someone deciding that a machine asking a question is a commercial relationship with terms. Until a carrier does that — publishes the terms, authenticates the caller, prices the query — every efficiency gain just raises the ceiling on how much free looking the industry absorbs. And the airline that builds that governing layer is not solving a compute bill. It is taking back the thing thirty years of distribution history took away. So the watch item stands, and it just got more specific. First carrier to publish agent terms — authentication, rate limits, and a price for the query. That is still the signal. Nobody has hit it.

Eric Marketts

And an offer engine that has to answer a machine in real time, at volume, under a policy, is not a shopping problem. It is a core systems problem. Which is exactly where the industry just moved its center of gravity.

Eric Marketts

Daniel Friedli at Travel in Motion published the Modern Airline Retailing outlook the World Aviation Festival has been circulating, and one line in it is doing a lot of quiet work. He argues that vendors and airlines should treat the order management system, not the passenger service system, as the new "gravitational centre" of retailing. And he names the forcing function: a wave of expiring PSS contracts pushing carriers into decisions they have deferred for years. The vendor field has moved to match. Amadeus positions Nevio as an order management system. Navitaire has IATA ONE Order Capable status and unveiled Stratos, an offer-and-order-native successor to New Skies, cloud-native, with inventory, fares, schedules, and orders in real-time sync. Meanwhile the money keeps growing. IdeaWorks puts global ancillary revenue at one hundred forty-eight point four billion dollars in 2024, close to fifteen percent of worldwide airline revenue.

Steph Nell

So the bull case is that the order finally gets to be the thing everything else hangs off.

Eric Marketts

And it is a strong one. Everything painful about airline retailing traces back to expressing a modern commercial construct inside a system built around a fare class, a PNR, and an e-ticket. You cannot cleanly say bundle, dynamic price, partial change, or mixed-vendor purchase in that vocabulary. An order-native platform can. It holds inventory as flexible components and the customer's commitment as one object you can modify without a fare rule fight. That is what makes dynamic offers and real ancillary optimization possible instead of bolted on. And the timing is the actual news. PSS contracts are the single largest lock-in in airline IT, and expiry is the only moment a carrier has both leverage and budget. That window does not come around often.

Steph Nell

And the bear case is the sentence every airline CIO has nightmares about. Say it out loud at a conference and watch three of them reach for a drink.

Eric Marketts

"We're going to replace the PSS." It is the most dangerous sentence in airline IT and it has a body count. These programs run for years, not quarters, and the failure mode is not a clean loss — it is the middle, where the carrier runs a hybrid stack, part legacy PSS, part new OMS, with a reconciliation layer more complex and more fragile than either system alone. Airlines have lived in that middle for three, four, five years. And notice what is missing: nobody demos settlement. Nobody demos interline. The offer screen is where the vendor puts the money, because the offer screen sells the deal. The accounting is where the transition actually stalls.

Steph Nell

Critical take.

Eric Marketts

The industry just conceded, quietly, that the PSS was never the center of gravity. The order was. Important admission and almost nobody said it out loud. But here is what I would put in front of a board. Moving to an order-native platform is only worth the risk if the resulting system can answer a machine. If your new OMS still takes half a second to construct an offer, and has no way to authenticate an agent, meter it, or price differently for it, then you have spent four years and a very large number replacing your back office and you still cannot compete in the channel that is coming. Order-native is necessary. Not sufficient. The question for every vendor in that renewal cycle is not "can you handle orders." It is "what happens when two hundred thousand machine requests arrive for one sale, and how do I govern who gets an answer."

Steph Nell

And the honest answer from most of the field is going to be a roadmap slide with a lot of confident arrows on it.

Eric Marketts

Which is why the renewal window matters so much. You get one conversation with real leverage. Spend it on the right question.

Eric Marketts

Three narratives. One argument. Let me close it and Steph takes the last word.

Steph Nell

Go.

Eric Marketts

The three GDS bets are not bets on the pipe, and they are not the same bet. Two jobs sit above it, and the question is not who is in which column — Amadeus and Sabre are in both — it is which column each is betting the company on. Amadeus is betting it can own both. Sabre is hedging. Travelport cannot hedge, and went all in on the shelf. On our episode eight watch item, partial credit and a surprise: the first real answer to agent-scale search came from Amadeus, and it was precomputed fares. An engineering answer to a commercial problem. And the move to order-native platforms is the industry conceding the PSS was never the center of gravity. Same story, three altitudes. Intermediation is commoditizing, the interface is dissolving, and the value is collecting in the offer layer — building it, owning the shelf it lands on, and governing who is allowed to ask.

Steph Nell

And here is what I would say to someone newer to this industry. For thirty years, the power in this business sat with whoever controlled the pipe between the airline and the traveler. That is why the GDS mattered, why NDC happened, why our four-part season exists. What changed this month is that the pipe stopped being the scarce thing. Clean, structured access is cheap now. The scarce thing is the offer — a correct, current, governed, machine-readable answer to a specific question from a specific buyer — and the shelf it lands on. Airlines that can produce the offer own half of that. Airlines that cannot will be represented in someone else's answer by whoever can, with no vote on how.

Eric Marketts

So if you are making distribution decisions right now, change the metric you watch. Booking share by channel told you who won the last thirty years. It will not tell you who wins the next five. Watch instead for the thing we have now asked for twice: a carrier publishing real agent terms — authentication, rate limits, a price for the query. Eight weeks on, nobody has done it. That is still the single clearest signal that this industry has stopped treating machine demand as weather and started treating it as a customer.

Steph Nell

If this was useful, send it to someone making a PSS renewal decision or writing an AI strategy, because those are the same decision now and most organizations still treat them as two. And if you disagree — on the three bets, on whether precomputation is a retreat, on whether order-native is worth the risk — reach out. These conversations are better with more informed people in them. And if you missed the State of the GDS season, start there. It is the foundation under all of this.

Eric Marketts

And if you are the carrier that publishes those terms first, tell us. We will say your name on air, which is worth roughly nothing, but we will say it with feeling .

Steph Nell

High praise from a podcast.

Eric Marketts

We'll be back next Monday. I'm Eric Marketts.

Steph Nell

I'm Steph Nell. Thanks for listening.

Eric Marketts

Stay sharp out there.