Eric Dumas on why the real challenge in airline distribution is only beginning and what the industry is still getting wrong.
The debate about whether airline distribution is really changing is over. Eric Dumas, CEO of TPConnects Technologies, is not interested in relitigating it.
“We’re well beyond the ‘if’ question,” he says. “The interesting question now is whether you’re shaping this transition or just reacting to it. The gap between those two groups is widening rapidly.”
Dumas sits at the operational center of distribution, connecting airlines, agencies, and TMCs across NDC, GDS, and LCC channels. His perspective is shaped less by industry narratives and more by what is actually happening in production environments.
The shift, in his view, became irreversible the moment airlines started pulling content and making traditional GDS distribution economically unviable for certain fares. “That wasn’t a negotiating tactic,” he says. “It was a forcing function. What we’re seeing now is the aftermath, and it’s revealing who was actually prepared and who was hoping this would all blow over.”
Some carriers responded by rebuilding their commercial models from the ground up, with some now operating above 60 percent NDC adoption. Others, including some quite large carriers, continue to treat distribution modernization as a controlled IT workstream, slowed by procurement cycles, internal processes, and legacy contractual constraints. Revenue teams understand the urgency. Getting commercial teams to move at the same pace is a different problem.
The Problems Still Being Avoided
If the shift is already underway, the question is why the industry still struggles to scale.
Dumas points to issues that are widely acknowledged but rarely addressed directly.
Procurement velocity remains a bottleneck. Distribution transformation requires rapid iteration, yet airlines are still operating within RFP-driven processes designed for stability rather than speed.
There is also the compute-to-order problem. The industry continues to generate large volumes of search requests that never convert, driving infrastructure costs at a scale that has yet to be fully accounted for. “Airlines can’t sustainably offer unlimited shopping API calls for free,” Dumas says. “Agencies can’t absorb the infrastructure cost of excessive search volumes. Someone needs to build smarter caching and economic models that align cost with value.”
Servicing is the third fault line. While booking flows have improved considerably, post-booking modification remains operationally complex for high-volume agencies and TMCs. Dumas is more optimistic here than the industry’s general mood. “Very significant progress has been made over the last 24 months,” he says. Driven by airlines looking to extend NDC into TMCs and high-profile agencies, automation ratios are expected to improve sharply. The progress is real. But this is where true maturity will ultimately be defined.
Not Everyone Can Be Amazon
As airlines push toward retailing and direct customer engagement, the role of intermediaries is not disappearing, but it is changing fundamentally.
Dumas draws a distinction that often gets lost in the retailing conversation. There are effectively two types of airlines: those that own their technology stack, covering PSS, pricing, and revenue management, and the majority that depend on third-party systems. The first group controls its own destiny, but at significant cost. The second group’s strategic freedom is directly constrained by whoever their technology partner is. “The bigger the partner,” Dumas says, “the less freedom, flexibility, or agility you may have for your strategic evolution toward retailing.”
This is why the aggregator layer has become more strategically significant, not less. Airlines can push differentiated content and capabilities to sellers without demanding content parity, something the GDS model historically required. As one concrete example, TPConnects has integrated automated group booking capability directly into its NDC portals, allowing agencies to submit and manage group requests without manual processes or separate systems. NDC does not address group bookings as a standard. But it is a significant part of B2B airline business in many geographies, and nobody is building it because the industry’s attention remains fixed on getting basic NDC working.
“Direct channels are important,” Dumas says. “But they are not one hundred percent of distribution.” The airlines thinking clearly about this recognize that owning the customer relationship does not mean building the entire distribution ecosystem themselves. The orchestration layer, aggregating supply, normalizing content, managing connectivity across GDS, NDC, and LCC APIs, is infrastructure work. “The intermediaries who position themselves as that infrastructure layer will be fine. The ones still trying to control distribution flow and customer data? They’re solving yesterday’s problem.”
What Agencies Actually Need to Survive This
For travel agencies, the transition introduces a level of operational complexity that most are not yet equipped to handle. Dumas is specific about what is required, and the list is longer than the industry conversation typically acknowledges.
The starting point is unified servicing. The content problem is largely solved. The servicing problem is not. “If you’re serious about travel, you can’t have your team toggling between five different systems to modify an NDC booking,” he says. Agencies are telling TPConnects directly that they secured the NDC content but cannot service it efficiently at scale. What is needed are platforms that allow PNR imports and full booking management within a single workflow. That is the baseline going forward, not a differentiator.
The second requirement is One Order readiness, and this is where technology partner selection becomes a strategic decision rather than a procurement one. TPConnects was the first provider to achieve NDC 24.4 certification, the version that serves as the bridge to OOSD. Agencies whose partners are still building on older NDC versions are accumulating technical debt quietly. “If your technology partner is still building on older NDC versions, you’re accumulating technical debt that will make the One Order transition considerably more difficult,” Dumas says. “The agencies asking about certification levels today are the ones who understand they don’t want to be migrating platforms again in 18 months.”
Third is intelligent revenue optimization, and this is where Dumas sees the most significant capability gap. Agencies need markup flexibility that goes well beyond flat percentage markups. What is required is context-aware pricing that understands route economics, competitive positioning, and customer segment. Critically, that flexibility needs to extend across the full range of ancillaries that airlines are now actively merchandising: seats, baggage, meals, lounge access, priority boarding. The agency that cannot markup and manage ancillaries dynamically is leaving revenue on the table in every transaction.
Fourth, for any agency operating at meaningful volume, LCC aggregation on a single platform is no longer optional. Sending agents to six different airline websites for every itinerary is not a workflow, it is a cost. Group bookings need to follow the same logic: automated, integrated into standard operations, not handled through disconnected manual processes that sit outside the main system.
The underlying issue connecting all four is one of strategic positioning. “The uncomfortable truth,” Dumas says, “is that agencies need to choose.” Deep expertise for complex travel, or highly automated efficiency for transactional volume. The operational infrastructure requirements are real in either direction. The problem is that most agencies remain underinvested in addressing them, and the middle ground that has sustained many of them is narrowing fast.
Africa Is Not a Future Roadmap Item
Much of the industry’s technology has been built around assumptions that do not hold globally: reliable connectivity, standardized payment systems, scalable local infrastructure. In emerging markets, those assumptions break down.
The evidence is not abstract. Online agencies in Central Europe are declining solutions because the APIs are too data-heavy, unable to store six months of transaction data within their infrastructure constraints. “That’s not an edge case,” Dumas says. “That’s a design assumption mismatch.” If the problem surfaces in Central Europe, it is acute across much of Africa, Southeast Asia, and Latin America.
“If your technology requires infrastructure your customer doesn’t have,” he says, “you haven’t actually built for their market.”
Yet these markets represent the industry’s fastest growth. African airline markets are expanding more rapidly than Europe or North America, with similar trajectories in Southeast Asia and Latin America. The opportunity is not theoretical. “We see those markets as an opportunity, not as emerging priorities for future roadmaps. Build specifically for their needs, deploy quickly, and establish position while others are still conducting feasibility studies.”
The companies that commit now will build positions that are difficult to displace later.
“If your technology requires infrastructure your customer doesn’t have, you haven’t actually built for their market.”
The Decision That Will Define 2030
NDC may be the current focus, but it is not the end state.
“The fundamental shift is One Order, modern settlement, event-driven servicing,” Dumas says. “That’s what actually changes the economics and enables proper retailing.” The decisions being made now, in how companies approach the architecture transition, will determine who leads in 2030.
The core question is whether organizations build disposable bridge solutions they will need to replace, or invest in scalable architecture designed for the next decade, using translation layers to manage the migration from legacy systems. “Don’t build throwaway solutions you’ll need to replace in 18 months. Build the modern architecture you’ll need in 2030.” Most vendors will not give this advice because the disposable approach is easier to sell and faster to deploy. But the architectural implications are significant, and airlines making these decisions now need to understand them clearly.
A less-discussed enabler of this transition is Model Context Protocol, a standardized layer that allows AI agents to interact consistently with the fragmented NDC landscape, which currently spans multiple schema versions and makes integration slow and expensive. MCP functions as a normalizing adapter that exposes airline capabilities in a consistent, AI-ready format, eliminating the need for custom engineering per airline-version combination. Airlines and aggregators that build their own MCP layers will control how AI models orchestrate offers, maintain architectural flexibility, and avoid vendor lock-in while enabling the AI copilot, chatbot, and corporate booking tool integrations that are already reshaping how distribution is discussed at the executive level.
The other defining factor will be who solves the compute-to-order economics. Not the most visible technology challenge, but potentially the most consequential. “That’s not sexy technology,” Dumas says. “But it’s essential.”
The broader theme running through all of it is timing. “The winners will be the ones who committed before it was comfortable. You can have excellent strategy, but if you execute it three years after your competitor, you’ve already ceded the advantage.”
What frustrates him is watching industry committees slow-walk standards development while every participant protects their current position. “We don’t need another consultation period. We need organizations willing to deploy imperfect solutions and iterate in production. That’s how innovation actually happens, not in working groups, in live operation with real customers.”
“Waiting for consensus,” Dumas says, “has never created advantage.”
This feature is part of the TDN Executive Conversations series.
By TDN Editorial Team
