What Amadeus and Sabre Don’t Teach Us About Experiences

Having traced what the existing topology of a booking looks like in Essay 4, the natural next question is whether a different topology could emerge. Every serious conversation about infrastructure in the experiences industry eventually produces some version of the same sentence. “Who will be the Amadeus of experiences?” Or “this industry needs its own Sabre.” Or “we are still waiting for a GDS.” The sentence arrives with the weight of analogy, which is a powerful kind of weight, because it lets the speaker import a conclusion (a neutral infrastructure layer eventually emerges and extracts real value) without importing any of the structural work that conclusion depends on.

The analogy is worth taking seriously. Not to settle who will win a race no one can plausibly name the finish line of, but to do the aviation comparison honestly, and to notice how little of what made the airline clearinghouse layer possible is present in the market we actually operate in. What transfers is instructive. What does not transfer is more instructive still.

What a GDS actually was

The contemporary reader encounters “GDS” as a category of travel-technology company. The historical reality is narrower and more specific. SABRE, launched by American Airlines and IBM in the early 1960s, was built as the carrier’s internal reservation system before it was externalised to travel agents in the mid-1970s.^¹ Apollo, at United, and later Galileo, followed similar paths from the US side. Amadeus was assembled in 1987 as a European consortium between Air France, Lufthansa, Iberia and SAS, explicitly to put European carriers on equal footing with American ones that had already productised their reservation systems. Worldspan followed in 1990.^² The pattern is recognisable once named. The infrastructure emerged inside airlines, became external infrastructure over the course of a decade, and was then, through a combination of regulation and commercial pressure, divested into neutral ownership.

Three structural facts about this trajectory matter, and are usually glossed in casual analogy. The layer was not invented from outside the industry it eventually served; it was extruded from the carriers that needed it. Its neutrality was an institutional outcome, not a design choice. And it solved a specific pre-internet problem (aggregated airline inventory visible to travel agents) that was bounded by the commercial shape of aviation in that period.

What transfers

Three structural lessons from that period do carry over to any market trying to develop a comparable layer. They are worth stating plainly, because they are almost the only things in the GDS story that do.

Neutrality. A clearinghouse functions as a clearinghouse only to the degree that its operator is not also a distributor. Once the clearinghouse competes with its own buyers, its buyers route around it. This constraint is why the airline GDSs ultimately had to be divested from carrier ownership, under the regulatory rules that established neutrality requirements as a precondition of operating the layer at all.^³ The comparable-layer question in any other market is always, at root, a question about ownership structure.

Network effects. Supply and demand concentrate on one another, once concentration starts. The network effect is unglamorous and difficult to seed, but once a layer has enough sellers on one side that buyers find it efficient to start there, the process runs on its own.^⁴ The GDSs benefited from a period in which travel agencies were commercially indispensable and the airline commission structure rewarded aggregation on their side.

Transaction-fee economics. The unit economics of a winning infrastructure layer are transaction fees, not software licences. The layer that charges by the booking has a cost base and a revenue base that scale together. The layer that charges by the seat, the user, or the module eventually finds itself misaligned with the volume passing through it. This lesson is structural and survives the specific commission rates of aviation in the 1980s.

These three points transfer. Most of the rest does not.

What does not transfer

Four asymmetries stand between aviation in the GDS era and the experiences industry today. Each is structural enough that the analogy cannot be imported wholesale past it.

Inventory cardinality. An airline in 1980 sold a bounded vocabulary of products. A schedule, a route, a flight number, a published fare, a class of service. Tens of thousands of distinct units existed, but the schema describing them was small and well understood.^⁵ A neutral layer could be built on a finite taxonomy because the taxonomy was finite. An experience does not have a bounded vocabulary. A single operator may sell the same underlying product in three languages, from four meeting points, with two pick-up variants, in private and shared configurations, combined with transfers, combined with meals, with and without a kid seat, on seasonal hours, under weather-dependent operations, and with crew-dependent capacity. Across the industry, the permutations are effectively open. Standardisation in this market is not a taxonomy problem; it is a translation problem. Those are different disciplines, and the second is harder.

Buyer structure. The aviation GDSs formed in a world where demand aggregated through a small and commercially coherent set of intermediaries. Travel agencies operated on predictable commission economics, corporate travel departments routed through a handful of managed providers, and the airline commission cut produced strong incentives for consolidation on the agent side. Experiences has no equivalent buyer-side concentration. The demand side today is an uneasy mix of large online marketplaces, destination-marketing organisations, hotel concierges, ground handlers, tour resellers, and hundreds of thousands of direct bookings mediated by search.^⁶ None of these produce the buyer-side centre of gravity that made the airline clearinghouse economically coherent. A neutral layer in aviation had ten doors to knock on. A neutral layer in experiences has a continuous distribution of them.

Institutional apparatus. The aviation industry had, and still has, an institutional layer that predated the GDS and made many of its choices possible. IATA sets schedule standards, ticketing conventions, and settlement rules. National regulators and route authority govern who can fly where. Airline alliances produce shared codes and interline agreements.^⁷ Much of what appears in retrospect to be technical standardisation in aviation is actually institutional standardisation, enforced by bodies with real authority over market participants. Experiences has no equivalent. There is no IATA, no route authority, no universal product taxonomy, and no institution with the standing to impose one. A clearinghouse cannot be designed by assuming an institutional scaffold that does not exist.

Settlement infrastructure. The Billing and Settlement Plan on the international side, and the Airlines Reporting Corporation on the US side, clear money between agents and carriers under rules negotiated by the industry through IATA.^⁸ This is not a minor administrative detail. It is a neutral, industry-operated piece of financial plumbing without which the commercial logic of a neutral inventory layer would not hold together. Experiences settles peer-to-peer. Commission terms vary by channel. Payment timing varies by operator. Refund policies vary by product. Voucher redemption practices vary by destination. A neutral inventory layer is easier to imagine than a neutral settlement layer. Together they made the aviation clearinghouse load-bearing.

Each of these asymmetries on its own is enough to weaken the analogy. Together they change what kind of problem the experiences infrastructure question actually is.

The false confidence the analogy produces

The temptation of the analogy is that it smuggles in a conclusion. A neutral layer exists in adjacent travel categories; therefore, the reasoning runs, it must eventually exist here. The smuggled part is the preconditions. Aviation had finite inventory vocabulary, buyer-side concentration, an institutional standardiser, and a settlement plan before it had a clearinghouse. Experiences has none of these in comparable form. Any reasoning that imports the conclusion without importing the preconditions is doing less work than it appears to.

What the actual industry conversation reveals is that the analogy smuggles two opposing conclusions, not one. At the Arival Executive Summit in Valencia in 2026, a participant whose background was in the hotel sector described the GDS layer as exactly the kind of solution the experiences industry needs. A senior executive of a major online travel agent in the same conversation described the same layer as exactly what their organisation would resist, on the grounds that a non-fragmented supply side would dissolve the structural advantage their position depends on. The analogy was doing the work of two opposing arguments at once. Both speakers were right about the implications for their own position.

This matters because the industry’s infrastructure debate is mostly a debate about conclusions. Who will build it. Which company is positioned to win it. When it will consolidate. The more useful conversation begins from the preconditions and reasons forward.

What the analogy, taken seriously, actually suggests

If the structural conditions that produced the aviation clearinghouse do not exist in experiences, the interesting question is not who will be the Amadeus of this industry. It is under what conditions a neutral layer could emerge in a market with open-cardinality inventory, no buyer-side concentration, no institutional standardiser, and no settlement infrastructure.

Some of those conditions may be substitutable. A translation-based standardisation model may do some of the work a fixed taxonomy would have done in aviation. A specification with enough voluntary adoption across the category may produce a soft institutional apparatus over time, without a regulator ever convening. A settlement layer may be assembled commercially rather than institutionally, by participants who agree to route payments through a shared piece of plumbing because it is cheaper than the alternative.

None of these substitutions is automatic, and each of them carries its own set of structural trade-offs that the GDS history does not help us reason about. The industry is not owed the aviation shape. It will develop, if it develops at all, a shape of its own.

Close

The analogy is tempting because it offers a familiar outline for a market whose shape is still being drawn. Familiar outlines are worth something, but not more than the preconditions that drew them. The experiences industry needs infrastructure (Essays 3 and 4 made that case at the operational level), but it does not need a clone of the aviation clearinghouse. It needs a layer shaped to its own supply shape, buyer structure and institutional reality. The clearest partial answer to the infrastructure question that the experiences industry has produced so far is the channel manager, which began as a routing tool for a narrower problem and has been pushed, over two decades, to carry weight it was not built to carry. The next essay takes that layer on its own terms.