To be cited by AI assistants, a travel brand needs five things: a connected schema graph identifying the company and what it sells, answer-first content with 40–60 word capsules under question-shaped headings, server-rendered HTML that does not hide content behind JavaScript, open crawler access for GPTBot, PerplexityBot and Google-Extended, and third-party mentions corroborating what the site claims.
Why this matters now
A traveller planning Egypt used to open Google and scroll. Increasingly they open an assistant and ask a direct question: who should I book a Nile cruise with as a first-time visitor? The model returns two or three names and a reason for each.
Gartner expects traditional search volume to fall by around 25% by 2026 as this behaviour spreads. Whatever the exact figure, the direction is not in dispute, and the consequence for an operator is stark: in that format there is no page two. You are either named or you are not.
In an AI answer there is no page two. You are named, or you are absent.
The unusual thing about this moment is how open the window is. Across Egyptian travel we find very few operators doing any of this deliberately, which means the cost of being early is low and the advantage lasts longer than it will once the market catches up.
How assistants choose who to cite
Simplifying considerably, three signals dominate.
Can the model read you? If your content only appears after JavaScript executes, or your robots.txt blocks the crawlers, you are not a candidate. This is the single most common failure, and it disqualifies otherwise excellent sites.
Is your claim extractable? Models lift passages. A page that answers a question directly in its first two sentences is far easier to quote than one that builds to the answer over six paragraphs of atmosphere.
Does anyone else say the same thing? Models weight corroboration heavily. A claim that appears only on your own site is weaker than one echoed in a directory, a review platform, a press mention or a forum thread.
The five things to fix
1. Build a connected schema graph. Not scattered snippets — a single @graph where Organization, Service, CreativeWork and Person nodes reference each other by @id. This is how a machine learns that your company is a licensed Egyptian tour operator that sells Nile cruises, rather than inferring it from prose.
2. Rewrite key pages answer-first. Under each question-shaped heading, put a 40–60 word direct answer, then elaborate. Comparison tables and numbered steps are lifted particularly readily.
3. Serve HTML, not promises. Server-side rendering or static generation, so the content exists in the initial response. We have audited beautiful React travel sites that were entirely invisible to every AI crawler.
4. Open the doors. Allow GPTBot, PerplexityBot, Google-Extended and ClaudeBot in robots.txt, add an llms.txt describing your site, and verify it works rather than assuming. Also submit to Bing Webmaster Tools — ChatGPT’s search leans on the Bing index, so being absent there removes you from a large share of citations.
5. Get corroborated. Directories, review platforms, genuine press, and communities where travellers actually discuss Egypt. This is slower than the technical work and it is what separates brands that get cited from brands that merely could be.
How to measure it
There is no rank tracker for this, so measurement is more direct: take your money queries, run them through ChatGPT, Perplexity and Google’s AI Overviews monthly, and record whether your brand and domain appear. It is a citation count, not a position.
Expect different speeds. Perplexity typically reflects changes within one to two weeks. ChatGPT often lags four to eight weeks. AI Overviews follow your organic authority, so they move on ordinary SEO timescales.
The mistakes we see most
- Blocking the crawlers to protect content. For a tour operator this trades away visibility to protect a tour description. It is rarely the right trade.
- Client-side rendering with no fallback. Common on newer builds, and completely disqualifying.
- Schema that describes pages instead of the business. Article markup on everything and no Organization entity.
- Writing for length. Long preambles bury the answer. Models quote the sentence that answers the question, so put it first.
- Treating it as separate from SEO. Most of the technical work overlaps. Organic authority still feeds AI Overviews.
If you want to know where you currently stand, our free audit includes a check on whether the assistants mention you for your own money queries. Usually they mention an OTA instead.
Our free audit covers search visibility, AI-answer presence, booking flow and what your OTA channel is costing you. Written, and yours whether you hire us or not.