The real question isn’t whether distribution costs money. It’s whether the channel delivers access, revenue and value the airline could not generate more efficiently itself.
In September 2015 Lufthansa Group began charging €16 on every ticket issued through a global distribution system. It did not leave the GDSs. It told the UK competition regulator that it still sold through Amadeus, Sabre and Travelport, and a recent trade report puts the surcharge at roughly €18.50 on Amadeus and €24 on Travelport, though that report is undated and should be checked. Lufthansa’s chief commercial officer, Jens Bischof, said at the time that the move was not about a fee but about changing the distribution landscape.
That is the puzzle this piece tries to answer. An airline that wanted to eliminate distribution cost would leave the channel. Lufthansa priced it, kept it, and used the price to steer volume. American Airlines tried something harsher in 2023, then reversed itself in public. Ryanair, the airline most associated with going direct, has agreements with all three GDSs and signed with Booking.com’s owner in 2025.
The usual reading is that airlines hate GDS fees and love direct channels. The evidence is messier. Airlines appear to pay for access they cannot get any other way and to resist paying for access they believe they could own. That is a different thing from wanting lower costs, and it explains more of what they actually do.
Distribution cost is not one number
The first problem is that nobody is comparing like with like. Amadeus reported €3,119.2m of Air Distribution revenue on 484.5m bookings in 2025, which works out at about €6.44 per booking, up from roughly €6.25 in 2024. That is our arithmetic, and the revenue line includes some non-booking income. Amadeus says revenue per booking grew 5.0% at constant currency, driven by renewals, new contracts and inflation. Sabre reported an average booking fee of $6.04 in the second quarter of 2025, though that blends air and hotel.
The unit matters. Amadeus’s booking terms count a passenger on two connecting flights as two air bookings, so the GDS charges per segment while Lufthansa’s surcharge is per ticket. A return trip with four segments could carry about €26 in booking fees at the Amadeus average. A €16 to €24 per-ticket surcharge is therefore not obviously out of line with what a GDS collects on a multi-segment itinerary. The average also hides a second layer, because part of what airlines pay is recycled to agencies; Sabre says its contracts are priced so that booking fees over the contract life exceed the incentives it pays agencies.
Then there is the other side of the comparison. Lufthansa claimed in 2015 that GDS costs were several times those of its own website. Amadeus replied that Lufthansa’s figure for direct cost, €2, left out the cost of attracting online traffic, which it put at €15 to €20 per ticket. An agency-funded study by Infrata, published around 2017, estimated a network carrier’s direct cost at €2.56 per segment as airlines state it and €12.56 once acquisition, customer service, technology, payments and search are added back, close to its €14.21 estimate for a GDS booking. Infrata was paid by the agency side, and Amadeus has an obvious interest too. No neutral, audited per-channel cost exists in the public record, and anyone who quotes one is quoting a party to the dispute.
The test is whether the booking is incremental
If cost per booking cannot settle the argument, the question becomes what each booking is worth. The sharpest test is incrementality: would this customer have bought from the airline anyway?
Where the answer is yes, airlines push hard. United reported distribution expense falling 22%, from $626m to $487m, and its CFO attributed it to customers choosing the direct channel. (Our source does not state the period, so check the earnings release before quoting.) United pairs status points and a co-brand card with direct booking, so the customer is already United’s. Paying an intermediary to deliver that customer is a pure leak.
Where the answer is no, the arithmetic changes. In April 2023 American withdrew more than 40% of its fares from GDSs and left full content on NDC channels only. A year later chief executive Robert Isom told investors the airline had moved faster than it should have and had not executed well, and he linked a weaker revenue outlook to softer close-in bookings that he believed were partly due to the change. The commercial chief who drove the strategy left. American was trying to remove a cost from a channel that supplied corporate demand it could not yet serve another way.
Ryanair tells the same story from the opposite end. In 2014 Michael O’Leary said agent commission and GDS fees had once cost about 22% of revenue until the internet drove it to zero, then explained the return to GDSs by a need to be visible to companies. Ryanair left Amadeus in December 2016 when the sides could not agree financial terms, yet its 2025 annual report still lists Amadeus, Travelport and Sabre. Corporate customers are worth chasing, but only at the right price.
The difference is between cost per booking and economic value per booking. Delta’s president has estimated corporate travel at 30% to 40% of premium revenue and has said some high-end leisure yields more than corporate. A $10 distribution cost on a high-yield, high-attachment customer is cheap. The same cost on a customer who would have bought direct is not.
What a GDS sells that is hard to replace
If GDS distribution were only an unnecessary cost, the largest airlines, which have the most power to leave, would have left. None of the ones examined here has.
What they are buying is access to buyers who shop across carriers: travel agencies, corporate travel managers and the tools they use. A traveller who goes to an airline’s website has already chosen the airline. A corporate booker comparing six carriers on one screen has not, and the airline pays to be on that screen. Emirates, a carrier with a huge global footprint and no home market to lean on, signed with Amadeus in 2022 so that agencies could reach its content without a surcharge, while also running its own NDC direct connect. Air France-KLM cut its corporate surcharge from €24 to €3 for six months in 2025 to ease agencies across, and even then reported NDC at only 8% of corporate tickets in 2024, up from 4%.
Lufthansa’s own record shows the dependence. The group told the UK regulator in 2019 or 2020 that direct sales had reached 51% against 49% via the GDS, up from 30% direct when the strategy began, and a Skift interview later put direct at about 75%. Those are impressive shifts, and they took roughly five years, an NDC build and a public apology from Lufthansa’s CEO for deficiencies in its direct booking portals. Direct share here includes direct-connect and NDC channels, so the definition is wider than airline.com.
Amadeus makes a sharper claim: that airlines use their own websites as a bargaining chip in GDS negotiations. That is a GDS assertion, but it fits the surcharge pattern. A per-ticket surcharge set at or above what a GDS collects works as a signal to agencies and a lever with the GDS at once. It is hard to read it as a plan to exit.
NDC moves the toll more than it removes it
The usual framing asks whether NDC is cheaper than the GDS. The airlines’ own statements point elsewhere. On its third-quarter 2015 earnings call, Lufthansa’s chief executive called the surcharge’s effect on the bottom line broadly neutral and said the bigger prize was the freedom to roll out branded fares, grow ancillary revenue and later adopt a new revenue management system. Lufthansa put the ancillary target at €300m a year from 2018. What NDC solves is a merchandising problem: the old messaging standard could not carry the bundles and personalised offers airlines wanted to sell through agents. Cost, in that reading, is a lever and a justification, not the goal.
The costs of getting there are real and mostly unpublished. We found no airline that has disclosed what implementing and running NDC costs. Travel Weekly’s coverage of the Infrata study noted that Lufthansa’s GDS costs rose in 2016 despite the surcharge revenue, and presumed the group was also paying for its direct-connect build. That is a reporter’s inference, not a disclosed figure.
The larger point is who collects. Amadeus has said it aims to be the undisputed aggregator of NDC bookings, and it reports that the weight of NDC among airlines producing NDC bookings on its platform is in the teens, with its revenue per booking still rising 5%. Sabre reports 42 live NDC connections and says NDC was 4% of its air distribution bookings in 2025. Those numbers describe a transition in which the GDS keeps a position in the middle, now as an aggregator rather than an EDIFACT pipe. Meanwhile Amadeus said in 2024 that direct connections between one very large OTA and several carriers had weighed on its North American local bookings.
Airlines do gain control, data and the ability to sell differently. But the toll does not disappear. It moves: to aggregators, to NDC technology providers, to OTAs with leverage, and to the payment networks discussed below. Whether modern retailing changes who captures the value or only who sends the invoice is the question the industry still cannot answer with public numbers.
A channel’s fee is only part of what it costs once payments are counted. IATA, citing an Edgar Dunn study of 2023 data, estimates that airlines pay about $22bn a year to process roughly $1 trillion of payments: $18bn in fees and $4bn in operating costs. IATA turns that into about $4 per passenger, against industry net profit of roughly $7.70 per passenger in 2025. It also claims that one in five card transactions is not successfully executed by airlines and that payment acceptance now costs more than distribution. Those are IATA’s figures and IATA sells payment products, so they are an industry estimate with a commercial interest, not an audit.
The analytical point survives the caveat. A direct sale puts card fees, fraud and chargebacks on the airline. An agency sale puts part of that on the agent and on the settlement system between them. A channel that looks cheap on booking fees can look expensive once the airline takes the payment risk, and a channel that looks costly can buy cheaper settlement. Comparing channels without payments is comparing half of each.
Why African airlines cannot copy the playbook
The logic applies in Africa with sharper edges, and the public evidence is thinner than it should be. No African airline we found publishes its channel mix or its cost per booking. What exists is a handful of data points that point the same way.
An African Airlines Association dialogue in July 2025 reported that over 70% of ticket sales in some African markets still go through traditional travel agencies and that distribution can reach up to 15% of airline expenditure. The speakers came from two airline technology vendors, Hitit and DataWings, so treat the figures as industry claims. They also cited more than 600 million mobile money users on the continent, held back by diverse local payment methods, high conversion fees and blocked funds.
That matters for the strategy Lufthansa ran. A surcharge steers agency volume towards a direct channel, which only pays off if a large, well-served direct channel exists. Where agencies dominate, the substitutable share is small, and the direct channel itself carries higher payment friction. The incrementality test cuts differently: the intermediary is more often the only way to reach the customer.
Settlement can outweigh the fee. IATA says Nigerian blocked funds peaked at $850m in June 2023 and were later largely cleared. In 2022 Emirates’s blocked balance there was reported to be growing by about $10m a month, and IATA pointed to Emirates’s withdrawal from the market as proof of the connectivity risk. Globally, IATA counted $1.2bn blocked at the end of October 2025, 93% of it in Africa and the Middle East. Agencies and BSP do not cure a central-bank shortage: Qantas stopped ticketing through agents in Zimbabwe over outstanding BSP funds (the report is undated), and IATA has a resolution allowing a BSP’s billing currency to change when remittances are blocked. But BSP does give airlines a guaranteed monthly payment date and a way to suspend a defaulting carrier or agent, as with the suspension of Fly540. Kenyan agents have contested IATA’s plan to standardise remittance periods from mid-2026, arguing it ignores local conditions.
The carriers are also modernising. Kenya Airways chose Sabre in September 2026 to replace its core reservations platform and hired Branchspace for its digital commerce layer, aiming to grow direct sales and ancillary revenue; the airline’s claim that it will be the first African carrier to adopt modern retailing is its own. Ethiopian selected SabreMosaic in November 2025, with NDC content for agency partners. These airlines are building a retail capability while agencies stay essential. In June 2026 Kora joined IATA’s payment gateway, which the announcement says lets airlines take African cards, mobile money and bank transfers through one integration. Whether that changes the direct-versus-agency arithmetic is the question worth asking African commercial chiefs on the record.
When each channel is rational
The evidence does not support the idea that airlines are eliminating distribution costs, and it complicates the idea that they are simply becoming more selective. The pattern that recurs, with exceptions, is this: airlines don’t necessarily resist distribution costs. They resist paying for distribution when they believe they can own the customer, the capability or the access themselves, and they accept the cost when they cannot.
A GDS is rational where the airline needs corporate and agency reach it cannot build, as with a carrier outside its home market or one courting business travellers. Direct is rational where the customer is already the airline’s, through loyalty, a co-brand card or a strong home brand. NDC is rational where the airline has a merchandising problem that the old standard cannot carry and the technical capability to service what it sells. Each can fail when an airline misjudges how replaceable a distribution channel really is, as American’s experience illustrates, or underprices the cost of the alternative.
On the public numbers, nobody can yet show that modern retailing has changed who captures the value. The open question is whether it ever does, or only moves the toll.
