Travel SEO in the world of AI “answer engines”

For publishers and content creators, the furor surrounding the launch of OpenAI’s ChatGPT was certainly a shock. But for businesses dependent on Google’s organic search traffic, the initial hubbub on what generative AI would do to (human) content creation was just the warm-up act. 

The real game changer came next, once Google’s big AI move became clear. Harnessing generative AI trained on the billions of websites in its index, Google intends to turn itself from a search engine into an “answer engine.” Instead of the familiar pages of blue links to third party websites, Google will now provide an AI generated response and a chatbot interface.

The upshot is clear: expect a world with vastly fewer organic search clicks once Google starts to answer informational queries itself. This could be cataclysmic for the travel industry, where the top of the marketing funnel is dominated by Google search: itinerary and trip planning, destination research, experience and activity recommendations, there aren’t many travel queries that won’t be answered by the new AI interface.

There are massive legal, regulatory, social and economic questions around the shift that we are witnessing. What is the legal basis for training large language models (LLMs) on other peoples’ content without attribution? What happens to Google’s travel revenue when half the industry has gone bust?

Fortunately, these questions are all well beyond my pay grade. I’ll happily stay in my lane and confine ourselves to the implications for travel publishing and marketing.

Instead of millions of pages of search results, we’ll have a tiny number of “read more” citations within or alongside the AI answers that link out to a handful of source websites. Getting featured in those citations will be the new SEO, driven by an entirely new set of ranking factors which have been outlined over 2023 by keen Google watchers, notably the excellent Marie Haynes.

People-first content

First, Google is going hard on “helpful content” – or how well a page actually answers a query vs how well it’s optimized for SEO.

I think Google would be pursuing this regardless of the AI revolution, as it’s a common complaint that SEO has ruined the search experience, and the web in general.

If you’ve ever tried to read a recipe, a product review, or a typical travel blog article you’ll know what I’m talking about. For the last decade, the SEO game has been largely about crowding ever more text onto a page, forcing readers to root through thousands of words and dozens of vaguely related keyword-optimised subheadings to find the answer they’re looking for.

Those kinds of SEO tactics are going to age like milk, and a series of algorithm updates have already started to hammer sites that are getting this wrong. 

In order to stand any chance of making it into an AI citation, content will need to directly answer the user’s question, with minimal fluff or SEO waffle.

Travel sites and publishers who rely on organic traffic will need to have a good think about how well their content actually answers specific queries, potentially removing content that could be seen as unhelpful. Interestingly, this may be as much about UI/UX as content design, presenting the key information faster and more clearly, rather than buried deep down a colossal wall of text.

Credibility and expertise

The second indicator of a changing landscape was a small but significant tweak to what Google calls E-A-T, or “Expertise, Authoritativeness, and Trustworthiness”. This is Google’s system for determining what constitutes content credibility – i.e. whether the author is a qualified expert, and if they’re talking from their own personal knowledge.

Around the time of GPT3’s big launch in December 2022, Google added a second ‘E’ for “Experience” to the guidelines. I don’t think they’ve explicitly said that this precludes AI-generated content, but I’m not sure ChatGPT is currently able to get on a plane to Tanzania to research the latest safari camp openings, so it’s a safe bet that this is aimed at surfacing human-produced content based on direct personal expertise.

Once again, travel sites and businesses who want to thrive in this new world will need to think carefully about the provenance of their website content. For years desk researched blog articles churned out by non-expert copywriters have been considered a perfectly acceptable SEO tactic. Now, lots of people are going all-in on generative AI, which is essentially the same thing on steroids.

Google has clearly said that there’s no rulepenalizingAI-generated content in the search results, but that’s a long way from saying it will perform well when they’re trying to surface personal experience and expertise.

Originality amd perspectives

Related to the previous point: has there ever been a need for 56,600,000 results for the query “how to get to Machu Picchu”? Especially when every single one is repeating the same basic information?

This is precisely the sort of query that an AI answer engine can reasonably deal with. So what might humans be needed for? A recent change to some of Google’s search filters, adding an option to look for “perspectives” gives us a clue: original and unusual viewpoints, qualified by the E-E-A-T signals above.

An AI can easily regurgitate the handful of ways to get to Machu Picchu. Only an expert in Peru who is talking from decades of personal experience is qualified to say: “maybe, just don’t bother. I personally prefer Choquequirao instead, here’s why…”

It remains to be seen how far this will factor in the new AI answer engine interface. But if you assume that straight informational queries will be answered by bots regardless, it can’t hurt to focus on content that runs against the grain and offers unique and differing perspectives from everything else.

Adding new knowledge

Taking that point a step further is the concept of “information gain.” This is a patent from 2020 for a system that looks for content that adds genuinely new information, rather than repeating what’s widely known.

To extend the previous example, the existing hike to Choquequirao is already comprehensively described by hundreds of thousands of web pages. Information gain means someone physically visiting Peru’s Apurimac Valley, finding a brand new route, and creating original content (text, video or audio) that describes it all.

It could be new data, original market research, something that adds genuinely new knowledge to an existing topic, or simply a novel viewpoint or opinion that goes against the consensus.

This is something a robot cannot do, and it’s probably something that Google wants more of – not only to continue to train its LLMs, but also because users want a variety of opinions and viewpoints, especially when it comes to big, complex topics like international travel.

So I think it’s another reasonable bet that to compete in the world of an AI answer engine, content that genuinely adds new information to a subject (especially if it satisfies E-E-A-T criteria) should stand a better chance of doing well. 

SEO strategy or a suicide note?

With the above points taken together, the direction of travel becomes fairly clear.

Sites that have prospered by using non-experts to rehash existing knowledge without adding anything to the sum total of online knowledge (the majority of the SEO game for the last 12 years) are in for a world of pain.

Businesses and publishers who are capable of adding new knowledge and original perspectives, grounded in real-world experience, might stand a chance of surviving in this new era of search.

But is it worth it?

Dancing to Google’s tune is rarely fun, but SEOs have been adapting to the ever shifting goalposts for decades now. The difference this time is that the rules of the game have fundamentally changed. There was always a faustian pact between website owners trading access to their content in return for search traffic; now it seems that we’re all going to help train their LLMs while losing an as-of-yet unknown percentage of the audience share.

It is of course possible to block Google (and, recently, OpenAI) from indexing your content, but the nuclear option doesn’t really help if you’re already dependent on search traffic. Instead, travel businesses and publishers will continue to adapt as they always have, paying close attention to the explicit and implicit signals on what Google is looking for in this brave new world. 

And despite all this, I think Google will still have some form of symbiotic relationship with the web: people still want to see other sources of information, and Google still needs (some) publishers to exist – if nothing else than to continue to train their AIs.

Most interestingly, the approach outlined above is precisely the sort of thing that specialist travel companies and expert (vs generalist) bloggers – long the SEO underdogs – are perfectly placed to do. These are the people who know their destinations best; they understand the pitfalls of mainstream tourism marketing and have all the unique and contrarian perspectives to share.

After years of being cut out of the SEO action, is it wishful thinking to wonder if there’s a potential golden age for expert human-created content, provided it truly adds original perspectives and new knowledge to the world? 

Time will tell!