A visitor opens ChatGPT on their phone and types: “three-bedroom villa with sea views near Loulé, under 1.5 million”. Someone in Faro asks Perplexity for a good children’s dentist in town. Within seconds, the assistant replies with two or three names and a link for each.
GEO is the work of getting your business onto that shortlist. If your SEO is already in order, you are halfway there.
What GEO Is
GEO (generative engine optimisation) covers the practices that make an AI assistant more likely to read, understand and cite the pages on your site.
ChatGPT, Claude, Perplexity and Google’s AI Overviews answer with links in much the same way: they search an index, read a handful of pages and write an answer that cites its sources. To make it into that answer, your page has to be in the index, let the crawler in and state the answer in plain text.
AI Search Rests on SEO
Google is explicit. Its Search Central page on AI features says there are no additional requirements to appear in AI Overviews or AI Mode: a page must be indexed and eligible to show in Search with a snippet. It adds that you don’t need to create AI text files or special markup.
OpenAI says the same in its publishers FAQ: any public website can appear in ChatGPT search, as long as it doesn’t block OAI-SearchBot.
A slow site, missing from the index and short on useful text, stays out of AI answers for the same reasons it stays off Google’s first page. Start with our technical SEO checklist.
What to Do on Your Site
1. Let the Right Crawlers In
Each AI company runs several crawlers, each with its own job:
- OpenAI: GPTBot (training), OAI-SearchBot (ChatGPT search) and ChatGPT-User (user requests). OpenAI’s crawler docs state that sites which opt out of OAI-SearchBot will not be shown in ChatGPT search answers.
- Anthropic: ClaudeBot (training), Claude-SearchBot (search) and Claude-User (user requests). According to Anthropic, blocking the last two may reduce your site’s visibility in Claude’s answers.
- Perplexity: PerplexityBot (search) and Perplexity-User (user requests).
- Google: Google-Extended controls whether Gemini can train on your content and has no effect on Google Search.
Each rule stands on its own. You can refuse training (GPTBot, ClaudeBot) and stay visible in search:
User-agent: OAI-SearchBot
Allow: /
User-agent: Claude-SearchBot
Allow: /
User-agent: PerplexityBot
Allow: /
Check your firewall too: since July 2025, Cloudflare asks every new domain whether to allow AI crawlers and blocks them by default.
2. Serve HTML That Doesn’t Need JavaScript
In a December 2024 study, Vercel found that the crawlers from OpenAI, Anthropic and Perplexity don’t execute JavaScript. If your site only shows its text after JavaScript runs, those crawlers find an empty page.
Ask your developer for pages rendered on the server or generated as static HTML. The test takes a minute: open the page source (Ctrl+U) and look for the price, the address or the name of the service. If they aren’t there, ChatGPT can’t see them either.
3. Write the Answers Customers Look For
Assistants cite facts. “Bespoke solutions of excellence” answers nothing. “We service pools in Lagos, Portimão and Lagoa, from €60 a month” answers three questions.
Put on each service page what a customer would ask on the phone: price or price range, the area you cover, lead times, opening hours and what’s included. An FAQ built from real customer questions gives the assistant short sentences to quote. Keep your name, address and phone number identical on your site and your Google Business Profile.
4. Structured Data That Matches the Page
Structured data (Schema.org in JSON-LD) tells machines what kind of business you are, where you are and what you sell. For AI Overviews, Google asks for one thing: that this data matches the visible text on the page. Use the type for your sector (LocalBusiness, Restaurant, RealEstateAgent), FAQPage on FAQ sections and BreadcrumbList on inner pages.
5. An llms.txt File as an Extra
llms.txt is a proposal by Jeremy Howard, published in September 2024: a Markdown file at the root of your site (/llms.txt) that sums up the business and points to its most useful pages. It has a title, a short summary and lists of links:
# Barlavento Pools
> Pool maintenance in Lagos, Portimão and Lagoa.
## Services
- [Monthly maintenance](https://example.pt/maintenance): from €60 a month
Google has said you don’t need AI text files. An llms.txt costs little to build and serves the assistants and agents that read it. Generate it from your database so it keeps up with every change.
6. Tell Bing Every Time You Publish
IndexNow tells Bing, Yandex, Naver, Seznam and Yep whenever you add, change or delete a page. Bing’s index feeds Microsoft Copilot, and in 2023 Microsoft announced Bing as the search engine behind ChatGPT.
In Bing Webmaster Tools, the AI Performance report, in preview since February 2026, shows which of your pages Copilot cites.
7. Measure the Traffic AI Sends You
Three data sources are enough to start:
- Server logs: each crawler identifies itself in the user agent. You can see which pages OAI-SearchBot or Claude-SearchBot read, and how often.
- Analytics: according to OpenAI, ChatGPT adds
utm_source=chatgpt.comto the links users click. Set up a segment for that traffic in GA4 or Plausible. - Forms: record on every enquiry whether the customer came from an assistant.
A Real Example: ICON Property
In 2026 we rebuilt the site for ICON Property, an Algarve estate agency. The old site was an AngularJS app, and the listings only appeared after JavaScript had run. On the new site:
- An llms.txt generated live from the database, with every property for sale (price, bedrooms, areas and town), the contact details and the 16 municipalities the agency covers.
- A robots.txt open to 21 AI and search crawlers, including GPTBot, OAI-SearchBot, ClaudeBot, Claude-SearchBot, PerplexityBot, Google-Extended, Bingbot and Applebot.
- Server-rendered pages in four languages, with hreflang.
- Structured data for RealEstateAgent, priced offers, FAQPage and BreadcrumbList.
- IndexNow on every publish, so Bing hears the moment a property goes live, changes price or comes off the market.
- AI traffic tracking: which assistant reads which page, and which enquiries came through an assistant.
If someone asks ChatGPT for a villa in Loulé, ICON’s site already has the answer written down, with price, areas, town and a link to the listing.
Frequently Asked Questions
Does GEO replace SEO?
No. AI assistants find their sources in search indexes, and Google sets the same conditions for AI Overviews as for ordinary search. GEO adds steps on top of solid SEO.
If I block GPTBot, will I disappear from ChatGPT?
No. GPTBot is for model training. ChatGPT search uses OAI-SearchBot, and OpenAI treats the two rules separately.
Conclusion
AI assistants cite sites that Google and Bing can already find and read. Sort out your technical SEO and your local SEO first. Then let the right crawlers in, write the answers customers look for and measure what comes back.
Want to know how your site shows up in ChatGPT and Google today? Get in touch. You can also browse our services or request a quote.