LLM-first airline retailing
It is going to be harder than it sounds, but the rewards could be high
ChatGPT can save you money when you book flights. It can also help you find the most convenient times, figure out which airline has the best meals and consider the places and times you might want to go to. It can even tell you which prompts to use to achieve all the above.
Large Language Models (LLM) are poised to turn the way travel shoppers buy flights and other travel products head over heels. In 2012, travel shoppers typically visited eight sites before buying. By 2022 this had increased to 35 according to Google, a tech giant.
With LLMs things are moving back in the opposite direction. Now, savvy shoppers might only need to visit two or three sites before all their travel is defined.
Travel buyers have their favourite LLMs and many will turn to prompting before they even consider going to an airline’s website.
Artificial Generative Intelligence (AI) is evolving so fast that it is hard to know exactly where things stand today. Yesterday’s ideas are already chip paper. Tomorrow’s may have moved on from today’s. Which means it is hard to offer a definitive state of the industry.
However the direction of travel has become clear. In a world where deft shoppers are LLM-first, airlines can no longer rely on traditional ways of doing things.
This presents significant challenges. Legacy distribution platforms and Online Travel Agents can no longer be relied upon to generate traffic.
Amadeus, Sabre and the other travel vendors will struggle to align their legacy platforms with the way travellers will buy in the future. Identifying new technology platforms and partners will be costly for airlines.
But the LLM-first environment also presents opportunities for airlines.
A travel shopper who is considering flights with a carrier may be much more likely to buy once they start their Internet Booking Engine experience. In the words of sales bros, they are “further down the funnel”.
Accordingly travel shoppers who arrive at an airline after an LLM recommendation may be less price sensitive and more inclined to purchase ancillaries or travel packages. This could translate into easier revenue.
To give airlines a sense of urgency, LLM-driven retailing could also offer a compounding first-mover advantage as they are recommended to AI users who come back to sources the AI reads and recommend it to others, which the AI passes on etc… etc…
There are four things that airlines need to achieve to sell flights effectively when travel shoppers are LLM-first. Read on…
First up is actually getting recommended by AI.
When I first started my airline career with Qatar Airways back in 2007, everything was about keywords. Airlines competed with each other to bid up the price for “flights to Paris”, “cheapest flights from Doha” and “half term holidays on the beach”.
When consumers are shopping for travel using AI, things are not so different at a fundamental level. They still want to go to Paris or the beach, and they still want a good deal. Air fares from a distant or overseas city remain irrelevant for most shoppers.
What has changed is the portal and the presentation. In an LLM-driven world, airlines can no longer throw money at buying keywords.
They have to be recommended by well-regarded third-party websites. These are the places that create the content that LLMs consolidate and include:
1. Dedicated review sites such as the comments section of Trip Advisor and the many air travel bloggers and vloggers
2. Frequent-flyer interest sites, such as FlyerTalk
3. Traditional travel journalism
4. Emerging travel marketplaces, like what Turkish Airlines Holidays may become.
Rather than buying search engine keywords, airlines will need to put resources into ensuring there is plenty of long-form text and other content for LLMs to learn from.
Generating such content will be much harder than simply buying keywords, although if done well need not be as costly. The content will need to be organic, original and insightful. It cannot be designed by committee and cannot be pumped out of a PROS optimiser.
Airlines will need to get used to working with third-party content creators. Their talents may not be well aligned with the top-down corporate structure of large organisations.. A startup who can consolidate the process of content creation and management could be successful.
All that said, airlines will have one big advantage over other industries. Their barriers to entry are high. So even if airlines do little, LLMs will still learn about them at least at a basic level.
The second thing airlines need to do in the LLM-driven travel ecosystem is redesigning the booking process.
The entire airline shopping and booking process is based on a ‘from -> to’ stackable process. This was the best that could be done in the 1960s when aviation IT was being built. But it never really captured the spirit of what travel shoppers wanted.
Gentle reminder: most articles are now for paying subscribers. If you have subscribed already, thank you for supporting me. If you are subscribing now, I am offering 25% off here.