AI Doesn’t Recognize Your Org Chart

An AI system flags a market anomaly. Demand on a key leisure route has spiked past what the forecast supports. A competitor has pulled capacity. The search shows increased conversions.

Acting on it requires a fare change, an inventory adjustment, a check against the ancillary implications, and alignment with whatever the digital channel is running for that destination. In most commercial organizations, those are four separate conversations. Sometimes five. Each one is a handoff to a different team, on a different cadence, working from a different view of the same data.

By the time the coordination finishes, the window is closed.

The org chart is doing exactly what it was designed to do. That is the problem. It was designed for a different set of problems, and those problems are gone.

How the structure was drawn

The commercial org chart most airlines run today reflects a logic that made sense for its time. Pricing files fares. Revenue management controls inventory. Network planning designs the schedule. E-commerce owns the digital channel. Ancillary manages add-ons. Each function has its own leadership, its own tools, its own cadence, its own definition of success. They hand off across clearly drawn boundaries, and the data they depend on lives in separate systems that do not talk to each other.

That separation was rational. It was built when a price-controlled ticket was the only thing being sold. Then came variably priced inventory, ancillaries, fare families. Rather than rebuild, carriers bolted each new capability onto the original systems until the result resembled Frankenstein’s monster: functional, but assembled from parts that were never meant to operate as one body.

Convergence is not consolidation

The premise underneath that structure, that each function needs its own lane because the data is siloed, is eroding. Shared analytics platforms and the pressure toward offer and order management mean pricing, RM, and ancillary are no longer condemned to disconnected views of the same airline.

But converging and consolidating are different conditions, and most carriers are living in the gap between them. Some data has been unified. A great deal still sits in legacy source systems feeding separate warehouses, on separate schedules, in separate schemas. Some of it lives in a spreadsheet fed by a manual pull. The fact that commercial functions could consume the same data is not the same as having that data in a form cross-functional teams can use at the same time.

This is what the org chart question has to reconcile. Restructuring teams around data instead of output assumes something most airlines cannot yet claim: that the data supports it.

Before the org can change, the data has to change

For most carriers, the map looks like this. Booking data lives in the PSS. Revenue data lives in revenue accounting. Competitive intelligence lives in a subscription tool. Ancillary data lives in a merchandising system. Search data, where it is captured at all, lives somewhere else entirely. Different schemas, different refresh cadences, different definitions of the same business concepts.

The definitional problem alone is enough to stall a transformation. What the PSS records as revenue for a booking is not the figure revenue accounting reports, is not the figure the RMS uses for yield, is not the figure finance recognizes at settlement. Each is describing the same transaction. Each is correct inside its own frame. None can be compared directly without a layer that resolves the discrepancy. Put a cross-functional squad in front of a shared revenue target with that underneath them, and they will argue about the numbers before they ever reach the decision.

Building the data layer that makes cross-functional work possible takes three things, in order.

Consolidation. Bring the data into a single platform, typically a lakehouse or unified warehouse, on a common refresh cadence. This is an engineering problem: pipelines that extract from source systems, normalize the schemas, and load into a shared environment on a schedule that serves both operational decisions and financial reporting.

Definition. Every metric needs one designated source of truth, resolved before any tooling is built on top of it. What is the airline’s operating definition of revenue for commercial decisions? Gross or net, recognized at booking or at lift, inclusive or exclusive of taxes? What counts as a market for RM versus for network planning?

This part is not a technical exercise, and it is not comfortable. It means putting department heads in a room where they discover they have been defining the same words differently for years, and holding them there until it is settled. An org chart cannot be restructured around data that the organization defines differently depending on who is being asked.

Access. The consolidated, consistently defined data has to reach the people making the decision, in a usable form, at the moment they need it. Technical, because query environments and tooling have to serve commercial professionals, not just data engineers. Organizational, because deciding who sees which commercial data, at what granularity, touches pricing governance, competitive sensitivity, and accountability. That requires deliberate policy, not system defaults.

None of this shows up in a demo or a product launch. It is the foundation the restructuring stands on. Skip it and you do not move faster. You fail in production.

Offer and order makes the requirement acute

Offer and order treats pricing, inventory, and ancillary as a single transaction instead of sequential steps in a legacy booking flow. That does not just change the architecture. It changes what the organization has to be able to do.

Under the old model, a carrier could optimize the fare, hand off to inventory control, then hand off to e-commerce. Each handoff cost time, but the system was built around the delay, so the delay was survivable. Offer and order collapses the sequence. Pricing, RM, and ancillary now have to reason about the same transaction at the same moment. That is impossible when the data those functions depend on still refreshes on separate schedules in separate systems.

AI does not recognize the org chart

The AI systems commercial teams are deploying, for forecasting, dynamic pricing, inventory optimization, offer construction, do not operate inside departmental boundaries. They optimize across the whole commercial picture whether the humans overseeing them are coordinating or not.

That produces a specific failure mode. One system optimizes inventory on a route. Another optimizes ancillary pricing on the same route. A third runs the digital channel. If they are not informed by the same commercial intent, or if they are drawing on data layers that disagree with each other, their outputs work against each other. No single output is wrong by its own logic. The aggregate is incoherent, and a fragmented org has no structure that can catch it.

What the friction actually costs

Fragmentation shows up in three places.

Decision speed. The 48-hour window in the opening is not unusual. Markets move faster than approval chains built for sequential sign-off, and the cost of that latency compounds every time the window closes before the organization does.

Accountability. When an outcome requires coordinated action across four or five teams, good results get distributed and poor results get disputed. Without a clear owner for the integrated commercial outcome, the feedback loops that should improve decisions never close, and decision quality cannot be measured. The org chart is the structure producing that.

AI governance. This is the hard one. A commercial organization split into separate functions with separate tools and separate data cannot field a coherent AI strategy. Each function will advocate for its own AI investment, and each will be individually defensible. Together, with no unified data layer and no unified accountability, they build parallel infrastructure that amplifies the fragmentation instead of resolving it.

What restructuring around data might look like

Specialization does not disappear. Deep expertise in network economics and demand forecasting is still valuable. What has to change is the organizational model built to manage fragmented data and siloed systems, because it generates friction the commercial operation can no longer afford, and the AI investments already underway will make that friction more expensive, not less.

Melissa Skluzacek described a Revenue Architect who is capable of holding the whole commercial picture in mind. That role cannot function inside a structure that fragments the data it needs to see. The Revenue Architect is the operating unit. The cross-functional trading squad is the container. The unified data layer is the prerequisite.

A centers-of-excellence model gives it a workable shape. Functional depth lives in the center of excellence, the yield methodologist, the pricing specialist, the network economist, setting standards, governing methodology, developing talent. The cross-functional squad is the operating unit that deploys that expertise against specific commercial outcomes. The data layer is shared. The accountability is unified. The AI is governed at the level of the integrated commercial picture, not the individual function.

This is not a minor change. The data infrastructure has to be resolved for the organizational design to operate, and the definitional and governance questions that sit between the two have to be resolved alongside it. Neither can fully precede the other. They move together.

The technology changed. The problems changed.

The org chart most airline commercial organizations run today was drawn to solve real problems, and it solved them. The fragmentation it encoded was real. The specialization it created was necessary. The boundaries it drew matched the technology of its time.

That time is over. The data has converged. The decisions are integrated. AI is already optimizing across the whole picture. The organization is the last piece still built for the old shape.

The question for commercial leaders is not whether to restructure. It is whether to do the data and the organizational work together now, or to let the AI investments already in motion expose the missing foundation in production, where it costs the most.

Author’s note: telos builds decision infrastructure for airline commercial teams. The question of how to structure the data layer that makes cross-functional commercial work possible is one we work on directly with carriers.