Most explanations of NDC start in the wrong place. They start with the standard itself, with Offer and Order, with XML versus EDIFACT, and they leave the reader with a technical distinction rather than an understanding of why any of it needed to happen. To actually understand NDC, you have to start decades earlier, before GDS existed as a shared industry system at all, when airlines controlled their own distribution outright and gave that control away one system at a time.
Airlines built their own systems first
The Global Distribution Systems that now sit at the center of travel commerce did not begin as neutral infrastructure. They began as internal tools. American Airlines built Sabre in the 1960s to manage its own reservations. United built Apollo for the same reason. These were proprietary, airline-owned platforms, built to solve one airline’s operational problem, with no intention of becoming shared industry rails.
That changed as the systems proved useful beyond their original scope. Travel agents wanted access to real-time availability across multiple carriers rather than calling each airline separately, and the economics of extending an existing system to outside agents were far better than building distribution from scratch. Airline-owned reservation systems began opening to agencies, and eventually to each other’s inventory. Sabre spun out from American. Apollo evolved into what would become part of Travelport’s lineage. Amadeus was different from the start, built directly as a shared platform by a consortium of European carriers, Air France, Lufthansa, Iberia and SAS, precisely because they saw the American carriers building a distribution advantage they did not want to be locked out of.
By the late 1970s and into the 1980s, what had been individual airline tools had become genuine Global Distribution Systems, multi-airline, agency-facing, and increasingly indispensable to how tickets were sold.
The plumbing that made it possible
None of this consolidation works without a shared data standard, and that standard was EDIFACT, a message-based format for exchanging structured data that predates the airline industry’s use of it by years. Airlines needed a common language to file schedules, fares and availability into systems used by every other airline and every connected agency, and EDIFACT provided the rigid, standardized messaging that made that interoperability possible.
Fares specifically flowed through a related piece of infrastructure, ATPCO, the Airline Tariff Publishing Company, a cooperative utility owned by the airlines themselves. Airlines filed fares into ATPCO once, and ATPCO distributed that data out to every subscribing GDS multiple times a day. It replaced what would otherwise have been a impossibly fragmented process of every airline negotiating fare loads individually with every distribution channel.
This combination, EDIFACT for messaging and ATPCO for fares, is what let GDS scale into the position they eventually held. It is also, structurally, a system built for static, filed, rule-based content. It could carry a fare and a flight number. It was never designed to carry a personalized bundle or a dynamic offer, and that limitation sat quietly in the architecture for forty years before anyone built a standard to address it.
The toll booth
Once GDS held the primary channel through which travel agencies and corporate booking tools accessed airline inventory, they were positioned to charge for it, and they did, through a per-segment booking fee charged to the airline for every leg booked through the system. A single connecting itinerary could generate multiple billable segments from one passenger, and full-service carriers running large connecting networks generated enormous segment volumes as a structural feature of their business model.
At global scale, the numbers involved are difficult to overstate. Industry passenger data puts total worldwide air passengers carried, a figure that functions as a rough proxy for total flight segments, at somewhere in the range of five billion a year. Even accounting for the large share of that volume that never touched a GDS at all, the segments that did generate a recurring, high-margin fee stream measured in billions of dollars annually, collected for connecting an airline’s inventory to the point of sale rather than for any operational role in flying the aircraft.
For airlines, this was a cost with almost no negotiating leverage attached to it. As long as GDS controlled the only efficient channel to reach agencies and corporate travel managers at scale, airlines paid what amounted to a toll on their own product, regardless of fare value, regardless of route economics, simply because the GDS sat in the middle of the transaction.
The exception that came first
It is worth noting that airlines did not wait for a formal standard to try to escape this. Low-cost carriers, Southwest in the United States, Ryanair and easyJet in Europe, built their own separate reservation systems specifically to avoid GDS distribution entirely, years before NDC existed. Their entire commercial model depended on selling direct, without a per-segment toll attached to every booking.
This matters because it shows the instinct to reclaim direct distribution predates NDC by decades. What LCCs did with proprietary technology built from the ground up, full-service carriers eventually needed a standardized way to do without abandoning the agency and GDS channel altogether, since unlike the LCCs, they could not simply walk away from the corporate and agency booking relationships GDS had spent decades building.
The reclaiming
That standardized way arrived in 2012, when IATA introduced NDC, New Distribution Capability, as Resolution 787. NDC gave airlines an XML and API based structure for publishing their own offers directly, including branded fares, ancillaries and dynamic pricing, content that the EDIFACT and ATPCO architecture had never been built to carry.
The framing that matters here is not new technology for its own sake. It is airlines using modern infrastructure to take back a level of direct control over their own distribution and pricing that the earliest generation of airline-owned reservation systems, Sabre and Apollo in their original form, had once given them by default. NDC did not invent direct distribution. It gave full-service carriers, the group that had the most to lose to the toll booth model and the least ability to simply opt out of it the way the LCCs had, a path back toward the arrangement airlines held before GDS consolidated the market around itself.
Has NDC just built a new toll booth
There is a growing argument in the industry that NDC has not actually removed the toll booth, it has simply relocated it. NDC itself, the standard IATA maintains, charges nothing and sits in no transaction. But the practical experience of using NDC at scale has produced its own middle layer. Most agencies cannot realistically integrate with dozens of airlines’ individual NDC APIs one at a time, so aggregators, companies like AirGateway, Verteil and TPConnects, have stepped in to normalize that fragmentation into a single connection, charging their own fee for doing so. GDS have not stood aside either. Amadeus, Sabre and Travelport have all built their own NDC aggregation capability, absorbing NDC content into the same platforms that once ran purely on EDIFACT, and in many cases still charging for the privilege.
The result is a more accurate way to state the original claim. NDC was built to break GDS’s monopoly on the toll booth, not to eliminate toll booths as a category. What has actually happened is that the single, dominant collector has been replaced by a more contested field of collectors, aggregators and GDS themselves competing to be the layer airlines and agencies pay to pass through. The technical standard changed. The underlying economic function, someone sitting between the airline and the point of sale and taking a cut for the connection, has proven far harder to eliminate than the format it was built on top of.
Why this history is the point
The industry still largely talks about NDC as a technology upgrade, a newer format replacing an older one. That framing misses what is actually at stake. GDS were never simply infrastructure providers charging for a service rendered. They became, over several decades, the toll collectors on a road that airlines themselves had originally built and then handed over. The per-segment fee model that generated billions in annual revenue was not a reflection of the value GDS added to any single transaction. It was a reflection of how much leverage came from sitting in the one place a booking had to pass through.
NDC is best understood as a correction to that leverage imbalance, arriving forty years after the leverage was first lost, using a standard built for the internet era to do what proprietary CRS technology once did for American and United in the 1960s. But correcting a leverage imbalance is not the same as eliminating the conditions that created it, and the rise of NDC aggregators alongside GDS’s own NDC absorption suggests the industry has not escaped the toll booth model so much as multiplied who gets to run one. Every current fight in distribution, GDS surcharges, aggregator economics, which airlines push NDC hardest and why, is downstream of this one structural fact. The question was never really about XML versus EDIFACT. It was always about who gets to own the toll booth, how many of them the market will end up supporting, and for how long the rest of the industry is willing to keep paying to pass through.
