Travel marketers look to improve AI visibility and prepare for agentic booking

When Travelier CMO Mario Gavira said “travel is horrible” at Phocuswright’s Travel Marketing AI Summit London AI summit last week, it was not a popular viewpoint among the gathered senior travel executives. But Gavira eventually got the crowd on-side when he explained he was talking about the industry’s uptake of agentic artificial intelligence (AI).

Gavira backed up his claim by pointing delegates to Cloudflare’s Agent Readiness scores, which give an indication of how agentic-capable websites are across industries.

“Travel is around 20% agentic. Basically, we are nowhere from an agentic point of view, and the reason we are nowhere is because we don't really know how it's going to work out,” he said.

“Do we actually want to allow agents to be the front store of our retail experience? And second, are we able to even track it and control it? The answer is we don't really know.”

During the session, Piero Sierra, chief AI officer at Skyscanner, said he believes travel’s embrace of agentic is “inevitable.”

“We're definitely going to see the rise of agents powering travel,” he said, noting that other industries will transition sooner.

“Travel will be one of the last things to go. Travel's very hard. It's personal, it's expensive,” he said, also suggesting that it will become common to use agentic AI for booking simple itineraries.

“If it’s a low-cost flight I could see somebody doing that …The scenario that's always shown in the demos is ‘I want a five-star vacation somewhere exotic. I have three kids, please plan it for me and here's my credit card.’ That's not going to work from a complexity perspective, but I also think travelers enjoy travel planning. The trust gap is so great that I think it's not going to be ever surmounted except for the simplest kind of itineraries.”

This was clearly demonstrated when the session moderator asked the audience to use their red/green emotion cards to show whether they had carried out an end-to-end booking using AI. Just a single delegate indicated they had.

From SEO to GEO

Uptake is far from there, but the travel sector needs to plan for a new world built on AI. According to Phocuswright research, 40% U.S. travelers and 28% of U.K. travelers are willing to let an AI assistant book flights and hotels.

The panel discussed at length how travel companies need to prepare for this by moving from search engine optimization (SEO) to generative engine optimization (GEO).

B2B AI specialist Demand-Genius has carried out extensive research into the behavior of AI at the awareness, consideration and conversion stages of the buyer journey and how AI cites brands during those stages.

“We observe some pretty clear trends,” said CEO and co-founder Tom Rudnai.

“AI doesn't go to retrieval until it gets to conversion. If you look at the retrieval rate, the citation rate, the mention rate of brands, it's 0% at awareness and consideration because AI is not a directory. That's the fundamental difference between [AI] and Google. It has no interest in linking off to your site unless it serves the user's intent. Where AI does invoke retrieval, there's an SEO element to it, but we still have to be quite careful with how we go about that because AI doesn't search for the thing that the user originally entered.

The temptation is to optimize for prompts as if they're keywords, warned Rudnai, because modern AI uses a technique called query fanout. For example, if you type in the prompt “best Paris hotel,” the AI will break that down into sub-queries which might include “Palace Status holders,” “Sawday’s top 10” and “popular business hotel La Defense.” The idea is that by exploring multiple potentials, the AI avoids incorrectly guessing user intent.

This complexity makes travel GEO much harder to game than in other categories.

“Your purchasing journey for buying toothpaste is: ‘I ran out of toothpaste. What toothpaste should I buy?’ So yes, then it's SEO,” said Rudnai. But travel is not toothpaste.

Travelier’s Gavira likened GEO to the winner-take-all scenario in The Hunger Games.

“You need to first be able to make the cut of being part of the shortlist of the LLM to actually be part of the answer. Then you need to shape the opinion of the LLM with all of your content strategy, not just on site but especially on third-party sites,” he said.

He said that GEO is made harder because there is no source of truth in the same way that Google Search Console and Keyword Planner do for SEO.

“With LLM visibility … there is no real prompt volume data, and it's something that, so far at least, Google is [not] willing to share,” he said.

Prioritizing brand investment

The panel agreed that brand will remain vital.

Gavira said, “With AI, search is becoming more a branding channel; 99.5% of people who see responses in chatbots will not click on any link, so you cannot track it. It becomes branding. You need to be able to build an attribution model that allows you to have confidence in the investment. I don't believe that it's going to become a direct performance channel.”

“Brand is so important in a world of GEO,” concurred Skyscanner’s Sierra.

“There's always a temptation to view AI as this kind of shiny new channel … but AI is actually a reflection of your brand and reputation across all your digital footprint and a really intelligent version of that. The biggest driver of AI visibility is your general brand spend.

“One of the big mistakes people make when they're trying to kind of report on the success of stuff they do on GEO is they'll do this kind of tactical work from an SEO perspective. They'll have a load of brand spend over here. They see their visibility go up and they attribute it to the tactical work they did, not the fact that their brand is just generally getting better and better. Brand investment is one of the best ways to carve out some kind of moat in this.”

And small players with weaker brands are not doomed to disappear just yet, according to Rudnai.

“If you want to get a foothold in any form of search, it's about being very strategic with where you start, the specific types of customers that you're going after and the specific channels because it's fragmented so much. If you're a big player, you can drown every one of the sources that AI reads with information about your brand. As a small brand, you don't necessarily have that. You have to get very clever, and it all comes back to marketing fundamentals. Who are we trying to reach? What are the sources that search is reading? You then push your chips in on a particular channel. There are ways through it, but it requires a lot more nuance.”