Airline Revenue Management Fundamentals: From EMSR to Dynamic Offers

The earlier posts in this series have focused on the supply side of the airline business: how carriers design their networks, choose their fleets, structure their financing, and plan their operations. This article turns to the demand side, or more precisely, to the discipline that determines how airlines actually convert supply into revenue. Revenue management (RM), sometimes called yield management, is the systematic practice of controlling which seats are sold to which customers at which prices, with the goal of maximizing revenue per available seat kilometre (RASK) across the entire network.

For a moderately sized regional carrier, revenue management is one of the two or three most consequential commercial disciplines, alongside network planning and fleet strategy. A one-percentage-point improvement in RASK can translate to tens of millions of dollars in annual profit for a mid-sized operator, without any change in fleet, network, or operational cost base. Airlines that treat revenue management as a strategic capability outperform peers that treat it as an administrative function, and this gap has widened substantially as artificial intelligence, machine learning, and the ongoing NDC transition have expanded what revenue management systems can actually do.

This article walks through the framework: what revenue management is, how the classical model works, how the industry is transitioning to a modern offer-and-order world, and what the practical implications are for regional carriers in Asia-Pacific.

1. The problem revenue management solves

The core problem is this. An airline has a fixed inventory of seats on each future flight. Once the flight departs, unsold seats become worthless (a perishable inventory problem). Different customers have different willingness to pay for the same seat, driven by trip purpose (business vs leisure), booking timing (early vs late), flexibility requirements (refundable vs restricted), and network context (point-to-point vs connecting). If the airline sells all seats at the lowest price, it forgoes revenue from customers who would have paid more; if it holds seats too aggressively at high prices, it forgoes revenue from customers who would have paid less but ends up flying empty seats.

Revenue management is the systematic solution to this optimization problem. The traditional formulation, which has framed the discipline since the 1970s, breaks the problem into three components:

Demand forecasting predicts, for each future flight, how many customers will want to travel and what they will be willing to pay. Historical booking patterns, seasonal cycles, competitive factors, and event-driven demand shifts all feed the forecast.

Inventory control determines how many seats to make available at each fare level. Rather than releasing all seats at all prices, the airline “protects” seats at higher fare levels for customers expected to book later, using algorithms like Expected Marginal Seat Revenue (EMSR) to calculate the optimal allocation.

Pricing determines what fare levels to file with the distribution channels (traditionally through GDSs, increasingly through NDC-enabled direct channels). Fare structures typically include multiple fare classes with different restrictions, refundability, and change flexibility, allowing the airline to segment customers by willingness to pay.

The classical framework was developed initially for US domestic operations after deregulation in the late 1970s, and it has been progressively refined over four decades. The core scientific foundations, sometimes described as the “axioms” of revenue management, have proven remarkably resilient: forecast demand, optimize inventory allocation, adjust pricing based on marginal revenue analysis. What has changed dramatically is the sophistication with which these principles are executed.

2. Expected Marginal Seat Revenue (EMSR): the foundational algorithm

The most influential algorithm in classical revenue management is Expected Marginal Seat Revenue, developed in various forms since the early 1980s and refined into the EMSR-b variant that remains widely used today. The concept is intuitive: for each seat on a flight, the airline calculates the expected marginal revenue from selling that seat at a higher fare (accepting the risk that no high-fare customer arrives) versus selling it at a lower fare (accepting the certainty of a lower fare but ensuring the seat is not empty).

The mathematical formulation, in simplified form, says: protect a seat at fare level i if the probability of selling it at fare i or higher, multiplied by fare i, exceeds fare i-1 (the next lower fare level). Iterating this analysis across all fare levels produces a set of “booking limits” or “protection levels” that specify how many seats to make available at each price point at each point in the booking curve.

EMSR-b assumes independent demand across fare classes (a simplification that is not strictly accurate in practice, since some customers will “buy down” to lower fares if higher fares are available) and pointed booking demand distributions. Its refinements over the years have addressed these limitations while preserving the core computational structure.

For network operations rather than single flight legs, the extension is O&D (Origin-Destination) optimization, which considers the value of each seat not just for the current flight leg but for the entire network revenue generated by the connecting itineraries that seat supports. A seat on Singapore-Bangkok might be valued differently based on whether the customer is flying only to Bangkok versus connecting through Bangkok to Chiang Mai versus connecting through Bangkok to Hanoi. Network optimization requires substantially more computational complexity and richer data than single-leg optimization, and it is the standard approach at all major full-service carriers.

Bid pricing is the practical mechanism through which network revenue management is executed. For each future flight, the RM system calculates a “bid price” representing the minimum revenue at which a seat should be sold. The bid price is dynamically updated as bookings develop and as forecast conditions change. The reservation system compares each fare request against the current bid price, accepting fares above the bid price and rejecting those below.

3. The classical RM organization

In a full-service carrier of moderate size, the revenue management organization typically operates as follows:

The RM system, which for major carriers is either Amadeus, Sabre, PROS, or increasingly a hybrid combination, runs continuous forecasting and optimization cycles for every future flight in the network. Overnight batch runs update the forecast and produce new booking limits; real-time systems adjust availability as bookings arrive.

Revenue management analysts, typically organized by market or route family, monitor the system outputs, override recommendations where local knowledge suggests better decisions, and manage exceptional situations (sudden demand shifts, competitive actions, operational disruptions). At major carriers, this team may number 30 to 100 analysts covering the entire network.

Pricing analysts, sometimes organized separately from RM analysts, monitor competitive fare filings and adjust the airline’s own fare structures. Pricing changes flow through to the GDS systems (or NDC channels) and affect what the RM system has available to sell.

Ancillary revenue teams manage the pricing and distribution of unbundled services (baggage, seat selection, meals, priority boarding), which have grown from 9.1% of total airline revenue in 2016 to 15.7% in 2025 as unbundling has spread across the industry.

For LCCs, the organizational model is different. Traditional LCC revenue management focuses on simpler fare structures (fewer classes, less network optimization), with more emphasis on dynamic pricing based on load factor curves. Ancillary revenue is proportionally more important, often approaching or exceeding 30% of total revenue at major LCCs like Ryanair and Wizz Air.

For regional carriers in Southeast Asia, the model typically sits between full-service and LCC extremes. A mid-sized regional flag carrier might operate a full-service RM system for its widebody long-haul routes while using simpler dynamic pricing on regional narrowbody sectors.

4. Dynamic pricing and the transition to modern retailing

The classical RM framework, based on discrete fare classes and pre-filed fares, has come under sustained pressure over the past decade. The pressure comes from three directions: the emergence of e-commerce-style dynamic pricing in adjacent industries, the availability of much richer real-time data, and the strategic push by IATA and airline groups toward “modern airline retailing” enabled by the New Distribution Capability (NDC) standard.

The transition can be summarized as a move from three separate activities (inventory control, pricing, distribution) to a single integrated capability: dynamic offer construction. Rather than filing static fares and separately controlling inventory, the airline uses real-time computation to construct a specific offer for each customer at each shopping request, bundling the flight, the ancillaries, and the pricing into a single response tailored to the request context.

Adoption of dynamic pricing has grown steadily. As of 2025, approximately 260 carriers worldwide, roughly 80% of IATA member airlines, apply some form of dynamic pricing, a 20% increase from just two years earlier. However, the sophistication varies dramatically. Basic implementations use simple rules-based systems that adjust prices according to seat availability or booking timelines. Advanced implementations leverage a much wider range of variables, including competitor pricing, customer context, and machine-learning-driven demand forecasts.

The reality behind the marketing rhetoric is that only about one-quarter of all air tickets sold in the marketplace in 2024 were dynamically generated in any meaningful sense. The vast majority of flight prices are still set through predetermined static price points, typically within the conventional 26 booking classes. IATA has publicly acknowledged that the industry is largely still relying on legacy frameworks from past decades to price today’s flights.

The gap between the potential of modern retailing and current industry practice is substantial. McKinsey estimates that airlines could unlock up to USD 45 billion in additional value over five years through widespread adoption of modern retailing. IATA cites similar figures. For individual airlines, MIT research puts the revenue uplift from dynamic offer capabilities at 3%, while BCG estimates up to 10% for airlines effectively using real-time behavioural data. Airlines that have implemented full NDC-enabled dynamic pricing capabilities report 2.5% to 5.0% incremental revenue gains above traditional GDS-based distribution, according to Amadeus and Sabre case studies published in 2024 and 2025.

5. NDC, Offer Management, and Order Management

Understanding modern airline retailing requires distinguishing three related but distinct concepts:

The New Distribution Capability (NDC) is an IATA-developed XML-based data exchange standard, launched in 2015, that modernizes how airlines communicate offers to third-party distribution channels (travel agents, OTAs, corporate booking tools, aggregators). Before NDC, airlines filed static fare data into GDSs, which then distributed those fares essentially unchanged. NDC allows airlines to build richer, dynamic, more personalized offers and distribute them through indirect channels without the constraints of static GDS formats. Over 70 airlines are now certified through the IATA Airline Retailing Maturity (ARM) index, indicating meaningful NDC implementation. The most recent generation of NDC schemas (24.1) was released in 2024, with continuing enhancement.

Offer Management is the capability, separate from NDC, that determines what specific offers to construct in response to each shopping request. NDC tells the channel what products are available; the offer management system determines which products to construct, for whom, at what price, in real time. Many airlines have invested in NDC connectivity while leaving offer construction logic largely untouched, meaning they can transmit richer content through the new channel while still relying on pre-filed fares and manually configured bundles. The gap between NDC connectivity and true dynamic offer management is where much of the incremental revenue potential currently sits.

Order Management is the complementary system that captures the customer’s accepted offer, manages fulfilment across partners, handles servicing events, and consolidates everything into a single order record. IATA’s ONE Order standard replaces the fragmented Passenger Name Records (PNR), e-tickets, and Electronic Miscellaneous Documents (EMD) with a unified order that carries through the customer’s full journey.

The vendor landscape has consolidated around two dominant integrated platforms. Amadeus Nevio, launched in 2023, provides an integrated offer and order architecture. Finnair was its launch customer, and British Airways, Air France-KLM, and Saudia have since selected it. In 2025, the Lufthansa Group’s nine airlines announced an AI-native retailing partnership built on Nevio. SabreMosaic takes a comparable modular approach with ten independently deployable product suites. Sabre signed new SabreMosaic agreements with Aeromexico, Alaska Airlines, Avelo, and GOL Linhas Aereas for AI-powered offer management products in early 2025.

A 2025 Accelya report found that 72% of airlines recognise the strategic importance of Offers and Orders transformation, but only 27% have begun serious transformation programmes. Most of the remaining 45% are stuck on scoping decisions, concerned about vendor lock-in, or waiting for standards to mature before committing to specific implementation paths.

6. AI and machine learning in revenue management

The introduction of artificial intelligence and machine learning has substantially expanded what revenue management systems can do. Modern RM platforms integrate ML in several ways:

Demand forecasting has moved from time-series statistical models to ML-based models that incorporate a much wider range of features (competitor pricing, weather, local events, macroeconomic indicators, search data, social media signals). Amadeus, PROS, and Sabre all publish research demonstrating measurable forecast improvements from ML-based approaches.

Willingness-to-pay estimation, historically inferred from discrete fare-class response, is now increasingly modeled through direct estimation of customer utility functions using ML models. This enables continuous pricing (moving beyond discrete fare buckets) and personalized pricing (offering different prices to different customer segments in real time).

Choice modeling, which estimates how customers select among available offers rather than treating demand for each offer as independent, is a substantial ML application. Sophisticated choice models account for competitive substitution effects that classical RM models handle only crudely.

Reinforcement learning has been applied to bid pricing and dynamic offer construction, learning policies from operational experience rather than optimizing against static forecast models. Amadeus and academic researchers have published foundational work on RL applications in RM (Bondoux et al., 2020, in the Journal of Revenue and Pricing Management).

Generative AI is the newest frontier, with applications ranging from customer segmentation to offer construction to natural language interaction with customer service. Amadeus has publicly discussed generative AI applications in demand forecasting (Nanty et al., 2024-2025). New entrants like Fetcherr have built businesses on generative AI applied specifically to airline demand prediction and pricing.

The airline RM market is projected to grow from $6.8 billion in 2025 to $14.2 billion by 2034, a compound annual growth rate of 8.5%, driven primarily by adoption of AI, machine learning, and cloud-based platforms across the industry.

7. Ancillary revenue

Ancillary revenue, the sale of services beyond the base fare, has become one of the most important revenue streams in commercial aviation. Baggage fees, seat selection charges, priority boarding, meal purchases, extra legroom, lounge access, and increasingly bundled or unbundled fare families all contribute to ancillary revenue.

The industry-wide numbers illustrate the shift. Ancillary revenue grew from 9.1% of total airline revenue in 2016 to 15.7% in 2025. For major LCCs, ancillary revenue often exceeds 30% of total revenue and in some cases approaches 40%. Ryanair, Wizz Air, and other European ultra-LCCs derive substantial proportions of their profitability from ancillary sales rather than base fares.

The revenue management implications are significant. Traditional RM systems optimized base fare revenue, largely treating ancillaries as separate transactions. Modern integrated offer management systems consider the full offer (base fare plus ancillaries plus fare rules) as a single optimization problem. The customer who books a bundle including seat selection, baggage, and priority boarding may generate substantially more revenue than a customer buying the base fare alone, and the offer management system should account for this in constructing offers.

For regional carriers in Asia-Pacific, ancillary revenue development remains an area with substantial upside. Many mid-sized regional operators have not yet developed sophisticated ancillary strategies, and the incremental revenue available from doing so can be substantial. AirAsia and VietJet have built their commercial models around aggressive ancillary sales, while more traditional flag carriers have been slower to commit to unbundling.

8. The Asian regional context

Several considerations distinguish airline revenue management in the Asia-Pacific region from the North American and European markets:

Cash bookings and offline distribution remain more prominent in Asia than in the more digitized US and European markets. This affects both the pace of NDC adoption and the practical implementation of dynamic pricing, since offline sales channels do not support real-time offer construction as effectively as online direct channels.

Currency volatility and cross-border payment friction complicate multi-market revenue management. Airlines operating multi-country networks (Singapore Airlines, Malaysia Airlines, Thai Airways, Vietnam Airlines) must manage revenue in multiple currencies with varying levels of hedging and exchange rate risk.

Chinese and Indian domestic markets have distinctive dynamics. TravelSky in China provides GDS-equivalent services with different technical standards from Amadeus and Sabre, and Chinese carrier revenue management operates within a domestic ecosystem that is not fully integrated with global standards. India’s rapid growth has created similar market-specific dynamics, with substantial LCC market share and specific corporate travel patterns.

The rise of regional platforms and Chinese e-commerce channels for travel distribution has created alternative distribution paths that global GDS-based systems do not fully address. Some Chinese airlines have developed direct partnerships with WeChat, Alipay, Meituan, and other domestic platforms that serve as significant distribution channels.

For most Asian regional carriers, the practical revenue management priority in 2025-2026 is upgrading from legacy systems to modern platforms that support ML-based forecasting and NDC-enabled offers. The gap between what advanced systems can deliver and what most regional carriers have currently deployed is substantial, and closing that gap is one of the highest-return investments available.

9. Three rules for the revenue management team

For airlines evaluating or upgrading their revenue management capabilities, three working rules:

Rule 1: Start with data quality before advanced algorithms. Modern ML-based RM systems require substantially better data than classical RM systems. Booking history, cancellation patterns, fare mix, ancillary sales, competitive positioning, and search behaviour all need to be captured, cleaned, and made available to the analytics infrastructure. Airlines that invest in advanced algorithms while running on poor data typically achieve disappointing results.

Rule 2: Treat NDC adoption and offer management as separate initiatives with sequential timing. NDC connectivity, while operationally complex, is a defined technical initiative. Offer management is a substantially larger organizational transformation requiring new team capabilities, new commercial processes, and new performance measurement frameworks. Airlines that treat these as a single project typically fail on both; treating them as sequential related initiatives allows the technical foundation to precede the organizational transformation.

Rule 3: Measure revenue management performance against clear counterfactuals. Improvements in RASK can come from many sources (fleet upgrades, network changes, market conditions, competitive dynamics), and attributing gains specifically to revenue management requires disciplined measurement. Best-practice teams maintain clear baselines, define specific KPIs (spillage, spoilage, revenue per shopping session, incremental ancillary attach rates), and measure performance against those metrics rather than against absolute revenue outcomes.

Closing

Revenue management is not a static discipline. The classical framework that shaped the industry for four decades is being progressively augmented, and in some cases replaced, by dynamic offer construction, AI-based forecasting, and NDC-enabled distribution. Airlines that treat this transition strategically, investing in the data infrastructure, the vendor platforms, and the organizational capabilities required to compete, will operate with real advantages over airlines that treat revenue management as a routine administrative function.

For Asian regional carriers, the opportunity is particularly significant. The gap between current capabilities and industry frontier is often substantial, and closing that gap can deliver meaningful improvements in RASK without any change in fleet, network, or operational base. Revenue management is one of the highest-leverage strategic disciplines available to commercial aviation leaders, and it deserves attention proportional to its impact.

---

This article describes general airline revenue management practice as it has evolved through 2026. Specific vendor capabilities, implementation approaches, and operational metrics vary by airline and by market. Market data, revenue statistics, and vendor references cited reflect publicly reported information as of 2025-2026 and are subject to change. Practitioners should consult their revenue management vendors, IATA published resources, and current industry research for detailed operational guidance.