How will AI create new airline revenue careers?

Core airline RM jobs will be unchanged by AI, but typical career paths may evolve

Consultants rejoice! Implementing Artificial Intelligence (AI) at airlines is going to be tough. Managing it will be even harder. Airlines are going to need lots of human experts.

In the upcoming era of AI-powered airline technology, some issues will still need to be resolved by humans rather than out-of-the-box AI platforms. Legacies will stand. Decisions will need to be taken. AI will help airlines, but like every corporate asset will need to be well managed.

Among other things, the humans at airlines will ensure that AI platforms are working as they should. Different platforms may indicate different outcomes and make different recommendations. So human managers and analysts will also need to decide which of the many plausible strategic paths generated by an AI their airline should choose.

Even AI suppliers recognise this. OpenAI, the company behind the well-known ChatGPT platform, are now offering consulting packages starting at $10 million (£7.5 million).

They send in their engineers to figure out how to integrate AI with legacy technology and established ways of doing things. Not the actions of a company expecting to turn the world of industry on it’s head.

AI is going to be good news for many airline employees too. For people working in Commercial Division AI will create new career paths and tweak the old ones.

I have written before about how consumers adopting Large Language Models (LLM) is impacting e-Commerce today. In this article I am going to explore how I think AI is going to impact the Revenue Management career paths at airlines tomorrow.

Coming along at the same time as AI is the industry’s move to offer & order. Both of these trends will impact airline careers together.

Now read on…

Airline Revenue Management (RM) is perhaps one of the original AI platforms

RM generates a demand forecast. It is pretrained on historical data, often from airline Revenue Accounting. And it transforms the demand forecast into a set of inventory availability that picks up pre-defined price points to send a price to travel shoppers.

Generative. Pretrained. Transformer. In other words, RM is a species of GPT a bit like ChatGPT but more specific and less powerful.

It takes a lot of effort to get RM teams up and running. An airline that is serious about the discipline might spend 2% of it’s annual revenue on RM opex and 1% on RM capex. RM generates a lot of activity and a lot of work.

Every airline is organised differently. But somewhere within the orbit of a typical RM Department are five areas. AI will have a different impact on each.

First up is core RM, demand forecasting, inventory management and pricing

When it comes to day-to-day core RM, there are three organisational approaches.

1. Some airlines specialise their managers and analysts in one of these

2. Other airlines combine them all into single roles so one person handles all three

3. A third option is to combine demand and inventory, with separate pricing specialists.

Shortly after I arrived at Qatar Airways we shifted from option three to option one. This was part of a bigger RM change – the airline was busy adopting an over-priced and over-rated origin-destination RM system.

The core RM job is about managing what is in effect an AI system. Analysts and managers already have plenty of tools to generate all the standard charts and tables. As the airline industry adopts more advanced AI platforms in RM the job itself will not change.

AI will however cause some tasks in core RM to become more heavily automated than they are today. Existing RM platforms can observe changes in demand and translate them into price changes through inventory availability.

“Super-analysts” like Fetcherr’s Large Market Model used by Delta will be able to carry out these processes more quickly and thoroughly than today’s standard systems.

Large Market Models may also be able to scan the press and social media to identify special events, TV broadcasts and viral consumer trends that boost or reduce demand for air travel. They will get price changes out to the market faster.

But for the Demand Analysts and their Managers handling the demand forecast, there will still be many questions that AI Large Market Models cannot answer. Examples include:

1. Is meeting this month’s revenue budget more of a priority than next month’s or next quarter’s?

2. One piece of analysis (generated by AI) says we should prioritise high prices and another (also generated by AI) says we should prioritise lower prices to drive volume – which strategy do we think is more realistic and what should we do?

3. How should the airline trade off the varying levels of cost and risk associated with different overbooking strategies?

For all of these questions and many more AI can provide indicators of the answers. It cannot though decide which set of trade-offs the airline should choose. So Demand Analysts’ jobs are safe.

Jobs are safe for Inventory Analysts too, which we called Flight Analysts at Qatar Airways. I was first recruited in 2007 with the job title Flight Analyst but it turned out that I never actually joined that team.

The Flight Analysts and their counterparts at other airlines do not just ensure that seats are available for sale. They also handle weight and balance issues and set policies for rebooking and refunds when things go wrong.

AI can tell Inventory Analysts about the assumptions they can make for weight and balance. Or the costs and benefits associated with certain schedule change policies. But they cannot tell the Inventory Analysts exactly what they should do. The human element will remain.

In Pricing though AI could change many things. Today’s Pricing Analysts manage complex fare products with price levels and associated rules. It is likely that Large Market Models will automate much of the work in benchmarking and fare sheet management.

But at the same time as Large Market Models are entering the RM world to automate much of today’s pricing work, airlines are committing to move to offer & order. In this new framework, airlines are hoping to become money-spinning travel retailers.

They will build digital marketplaces like Turkish Airlines Holidays that will need all kinds of attractive travel content. In principle, airlines who get really good at retailing could offer flights for free.

So I expect that as AI automates the pricing process, pricing specialist will need to become more retailers. As well as price levels, they will need to design and develop product bundles.

Identifying which products might bundle well together given the airline’s customer profiles will be a job well suited to AI. Pricing Analysts will need to become good prompt engineers to get the optimum product designed in place.

This may not come naturally. British Airways has a track record of failing to understand what their customers want.

But there is one part of Pricing that will not change following wider adoption of AI. This is working with other teams, often in Sales and Marketing, to convince them that Pricing’s recommendations are sound.

Sales and Marketing will have their own AI. There is no guarantee that the other AI will give the same answers as RM’s AI. The dance where each department negotiates based on their own preferences will continue in the AI era.

Great Pricing Analysts today are highly skilled in navigating this dance floor. They will continue to need this skill even when AI is taking care of making the charts and the tables. Pricing Analysts who can transition into a retailing mindset will find their jobs safe as AI becomes widespread.

Second is RM systems development, which builds, tests and implements RM systems and ensures they are aligned with other airline processes like distribution and ticketing

Revenue Managers who enjoy coding and building software often love systems development. This part of the discipline has three broad components, which every airline splits into many more tasks:

1. Large-scale software development to implement software the airline has either built in-house or sourced from a third-party