Airlines have practiced revenue management for decades, but artificial intelligence is changing the level at which pricing decisions can be made. Rather than relying primarily on predefined fare classes and broad demand forecasts, airlines can increasingly use machine-learning models to understand how travelers respond to different prices and construct more dynamic offers in real time.

Lufthansa Group provides one of the clearest examples of this transition. Working with travel-technology provider PROS, the airline group has progressively introduced AI-enabled revenue management, continuous pricing and request-specific pricing across its network. In 2022, Lufthansa selected PROS as the foundational technology for its Revenue Accelerator Program, bringing Revenue Management and Real-Time Dynamic Pricing together across its seven full-service airline networks. The objective was to improve forecasting, offer optimization and continuous pricing across the group.

The most important aspect of the Lufthansa case, however, is not simply that the airline introduced a new pricing algorithm. Lufthansa effectively restructured the revenue-management problem so that AI could be applied more effectively.

Traditional airline revenue management attempts to solve two difficult problems simultaneously.

The first is an inventory problem: what is the economic value of the remaining seats on a flight? Because an airline seat disappears as inventory the moment the aircraft departs, airlines must continuously estimate whether selling a seat today is preferable to protecting it for potentially higher-value demand later.

The second is a customer-pricing problem: how much is a passenger willing to pay for a particular itinerary?

Historically, these problems have been closely connected through fare classes, availability controls and origin-and-destination revenue-management systems. Lufthansa concluded that as airlines attempted to model customers at increasingly granular levels, combining both problems could become cumbersome.

Florian Martin, Lufthansa Group's Head of Commercial Offer Methods & Automation, explained that the passenger perspective involves variables such as customer preferences, purchase behavior and willingness to pay, while the airline side must deal with thousands of interconnected flights, capacity constraints and bookings taking place as much as a year in advance.

Lufthansa's approach was therefore to disentangle the two problems.

On the capacity side, Lufthansa can estimate the value of capacity using flight-level demand forecasts and bid prices. A bid price essentially represents the minimum economic value the airline assigns to the next available unit of capacity.

On the pricing side, Lufthansa can model the passenger independently and apply significantly more sophisticated machine-learning techniques to understand how travelers react to different prices.

That separation is important because it removes much of the airline network complexity from the customer-pricing model. According to Martin, once the passenger problem is separated from capacity optimization, Lufthansa can apply highly granular models incorporating passenger type, preferences, price sensitivity and willingness-to-pay effects.

The result is not one giant AI model controlling everything. Instead, Lufthansa describes an architecture in which specialized models are orchestrated together: one side estimates capacity value and demand, while another models the customer's response to price. The two are then brought together when the airline makes the final availability and pricing decision.

This architecture laid the foundation for Lufthansa's adoption of Request-Specific Pricing, or RSP.

Traditional airline pricing is heavily dependent on predefined booking classes. An airline might make several fare classes available for an itinerary, with the revenue-management system determining which classes remain open depending on expected demand and remaining capacity.

Request-Specific Pricing moves closer to an e-commerce model.

Rather than relying solely on a fixed fare structure, the system can analyze the context of an individual shopping request and adjust the price accordingly. Lufthansa began rolling out PROS Request-Specific Pricing in July 2022. According to PROS, the technology uses interest and preference information together with shopping context to adjust prices in real time, with the goal of producing more market- and customer-relevant offers.

This represents a fundamental change in the revenue-management logic.

Instead of asking only:

“Which fare class should still be available?”

the airline can increasingly ask:

“Given what we know about this shopping context, demand and the economic value of capacity, what price should we present?”

AI therefore allows Lufthansa to treat pricing much more like a digital-retail problem. Machine-learning models can concentrate on the relationship between customer behavior and price while traditional revenue-management tools continue to protect the economic value of limited aircraft capacity.

The Lufthansa case is particularly valuable because the company has disclosed an actual revenue result associated with the pricing transformation.

At PROS' Outperform 2024 conference, Martin explained that an earlier Lufthansa-developed predecessor to the Request-Specific Pricing system generated a 2.3% revenue uplift.

After Lufthansa implemented the newer PROS Request-Specific Pricing solution, the reported uplift increased to 5.2%. Importantly, Martin stated that this figure was net of the impact of continuous pricing, meaning additional revenue gains from continuous pricing were not included in the 5.2% result.

That result illustrates why AI can be economically significant in airline revenue management.

Airlines do not necessarily need AI to generate enormous changes in the price of individual tickets. The value comes from improving a very large number of pricing and availability decisions at the margin. Even relatively small improvements in revenue captured per booking can become material when repeated across millions of passenger transactions.

In Lufthansa's case, AI is therefore not functioning primarily as an automation tool. It is being deployed directly against one of the airline's most important commercial optimization problems: matching demand, willingness to pay and constrained capacity in a way that increases total revenue.

Lufthansa has subsequently expanded the approach beyond the base airfare.

In January 2025, the group announced an expansion of its use of Request-Specific Pricing alongside the adoption of Dynamic Ancillary Pricing. That technology applies AI to the pricing of ancillary products using factors including trip characteristics, day of the week, market conditions and demand.

This matters because airline revenue increasingly extends beyond the ticket itself. Baggage, seat selection and other ancillary products can all contribute to the economics of a passenger transaction.

The optimization problem is consequently moving from:

“What should this seat cost?”

toward:

“What combination of products and prices creates the most attractive offer while maximizing the economics of the transaction?”

By 2026, Lufthansa and PROS had extended their partnership across Revenue Management, Real-Time Dynamic Pricing, Continuous Pricing, Group Sales Optimization and new forecasting and optimization methodologies. PROS described the strategy as moving beyond traditional class-based fare structures toward more flexible, contextualized and attribute-based offers.

The strategic progression is clear:

Traditional revenue management: determine how much inventory to protect and which fare classes to make available.

Continuous pricing: introduce more granular price points rather than relying only on fixed fare-class boundaries.

Request-specific pricing: use AI to incorporate shopping context and customer response into the price presented for a specific request.

Dynamic offer optimization: extend the same logic beyond the airfare into ancillaries and potentially the entire travel offer.

Lufthansa demonstrates that the most powerful application of AI in pricing is not necessarily replacing an existing revenue-management system with a single predictive model.

The company instead redesigned the decision architecture.

Capacity optimization remains a specialized airline problem. Lufthansa still needs to forecast demand, manage interconnected routes and determine the economic value of scarce capacity. But by separating that problem from passenger willingness to pay, the company created a domain in which modern machine-learning methods could operate with significantly greater flexibility.

This is an important lesson for AI adoption more broadly. Companies may not generate the greatest value by asking AI to solve an entire operational system end-to-end. Greater value can come from decomposing a complex decision into separate problems and applying AI precisely where higher-dimensional prediction creates an economic advantage.

For Lufthansa, that advantage is visible in the revenue numbers. The move from an internally developed pricing approach producing a reported 2.3% uplift to Request-Specific Pricing producing 5.2% suggests that greater sophistication in willingness-to-pay modeling can translate directly into measurable commercial performance.

The Lufthansa case therefore illustrates the evolution of AI in airline revenue management particularly well: from forecasting demand, to optimizing prices, to creating increasingly contextualized offers in real time.

The ultimate objective is no longer simply filling the aircraft.

It is maximizing the economic value of every unit of capacity and every customer interaction.

PROS. (2022). PROS Drives Revenue Acceleration Across the Lufthansa Group Airline Network. https://pros.com/news/pros-drives-revenue-acceleration-across-the-lufthansa-group-airline-network/ 

PROS. (2026). AI-Fueled Innovation: Revenue Management. https://pros.com/learn/outperform-2024/ai-fueled-innovation-revenue-management/