How to Appear in ChatGPT, Gemini and Perplexity

People no longer rely solely on a list of links. They now ask ChatGPT, Gemini, Perplexity and Google’s generative features to compare solutions, explain a subject and sometimes recommend a professional. This creates a new question for businesses: how can you become a source that these systems can understand, cite and link to?

In short: there is no magic button or tag that guarantees a citation in an AI-generated answer. To improve your chances, your website must be accessible to the relevant crawlers, indexable, technically clear, rich in verifiable information and recognised by other sources. AI visibility extends SEO; it does not replace it.

SEO, AEO and GEO: what do these terms mean?

Traditional SEO aims to make a page visible in search engines. AEO, or Answer Engine Optimisation, focuses on supplying answers that can be easily extracted. GEO, or Generative Engine Optimisation, more broadly describes the optimisation of content that generative artificial intelligence tools may use or cite.

In practice, these disciplines share the same foundations:

• permitted crawling;
• correct indexing;
• useful and understandable content;
• a consistent business identity;
• evidence and sources;
• a good mobile experience;
• credible external mentions.

Google officially states that no additional technical optimisation is required to be eligible for its AI search features: established SEO best practices remain the foundation. In particular, a page must be indexed and permitted to appear with a snippet.

1. Allow the right crawlers to access your website

Content that cannot be accessed cannot be analysed properly. Start by checking your robots.txt file, noindex tags, server protections and CDN rules.

For ChatGPT search, OpenAI recommends not blocking OAI-SearchBot if you want your content to be eligible for discovery, summaries, citations and links. This crawler is separate from GPTBot, which relates to the potential use of content for training. You can therefore make different choices for each one.

Example allowing access to the search crawler:

User-agent: OAI-SearchBot
Allow: /

This rule makes the website eligible for crawling; it guarantees neither use nor ranking. Also check that your server or firewall is not rejecting the crawler’s IP addresses while the robots.txt file appears to allow them.

2. Resolve conventional indexing problems

Generative engines rely heavily on web search and discovery systems. A poorly indexed website therefore starts at a significant disadvantage.

Check the fundamentals:

• correct HTTP status codes and pages available without signing in;
• consistent canonical tags;
• an up-to-date XML sitemap;
• logical internal linking;
• main content present in the rendered HTML;
• a complete mobile version;
• reasonable loading times.

If certain pages remain invisible, begin with the detailed diagnosis: why a website does not appear on Google.

3. Give the clear answer before adding detail

The easiest answers to understand provide the essential point quickly and then add nuance. A good page should not force the reader - or a machine - to work through fifteen paragraphs before discovering the answer.

For every important topic:

• provide a short answer below the heading or at the beginning of the section;
• use descriptive headings instead of creative but vague ones;
• keep each paragraph focused on one main idea;
• turn criteria, steps and mistakes into lists;
• define technical terms the first time you use them;
• add frequently asked questions that your customers genuinely ask.

This structure also improves the human reading experience, especially on mobile. The aim is not to write for a robot, but to remove ambiguity.

4. Describe precisely who you are and what you do

An AI system needs to connect your brand with an activity, a geographic area and supporting evidence. Repeat the essential information consistently across your strategic pages:

• trading name;
• profession and specialist areas;
• city and service area;
• the audiences you help;
• services offered;
• experience and working method;
• official contact details.

This information should match your Google Business Profile, professional networks, reputable directories and partner websites. A business described differently everywhere is harder to recognise as one consistent entity.

5. Publish specific, verifiable information

Generic claims such as “quality service” or “innovative solution” provide little substance. Content that is worth citing is generally more precise:

• a method explained in several steps;
• usual timescales and the factors that affect them;
• price ranges with clear assumptions;
• reasoned comparisons;
• first-hand observations;
• client cases with context, work carried out and outcome;
• simple definitions;
• checklists that can be used immediately.

Whenever you mention a figure, legal requirement or technical development, give its date and link to a reference source. A transparent page is more trustworthy than one that makes every claim without evidence.

6. Demonstrate real experience

AI engines can rephrase information that is already available everywhere. Your advantage lies in what they cannot legitimately invent: your first-hand experience.

Show real projects, the constraints you encountered, the trade-offs you made, before-and-after examples and observed results. Explain what does not work as well as the situations in which a particular solution is unsuitable. This nuance makes content more credible and useful.

For VulcainDesign, years of experience, projects delivered in Toulouse and clearly explained technical choices are far more distinctive than a succession of generic articles about “the importance of digital technology”.

7. Use semantic and accessible HTML

Clear code helps browsers, screen readers, search engines and software agents. Use a logical heading hierarchy, descriptive links, genuine buttons, form labels associated with their fields and relevant alternative text for images.

Structured data in JSON-LD can also clarify the type of item being presented: an organisation, local business, article, breadcrumb trail, service or person. However, it must match the content that visitors can actually see. Its presence guarantees neither a rich result nor an AI citation.

In other words, machine readability and human accessibility often rely on the same solid technical foundations.

8. Build a reputation beyond your own domain

A company calling itself an expert on its own website provides a self-interested claim. When customers, partners, local media, professional associations and specialist publications also mention it, its existence and reputation become easier to verify.

Work towards legitimate links and mentions through:

• complete professional profiles;
• carefully selected local directories;
• partners and suppliers;
• detailed customer testimonials;
• interviews and contributed articles;
• studies or resources that other websites genuinely want to cite.

The goal is not to accumulate hundreds of weak backlinks, but to earn consistent confirmations from real, relevant sources.

9. Keep information up to date

An older article is not automatically poor content. However, a page that presents discontinued tools, outdated prices or superseded regulations as current information loses its value.

Review the following regularly:

• dates and numerical data;
• external links;
• screenshots and interfaces;
• services and prices;
• legal information;
• biographies and contact details.

Display an updated date only when a genuine review has taken place.

10. Measure visibility without inventing an “AI ranking”

Generative answers vary according to phrasing, location, date and context. One manually tested prompt is therefore not a reliable metric.

Instead, measure:

• referral traffic from AI platforms;
• pages that are cited or visited;
• enquiries that mention ChatGPT or another assistant;
• impressions and clicks in Search Console;
• growth in brand mentions;
• conversions generated by informational content.

OpenAI explains that reference URLs from ChatGPT search may include the parameter utm_source=chatgpt.com, which makes them easier to track in an analytics tool.

What does not work

• publishing dozens of generic, automatically generated articles;
• artificially repeating “ChatGPT”, “AI” and your city name;
• adding structured data that does not match the page;
• inventing reviews, statistics or references;
• blocking every crawler and then expecting citations;
• buying hundreds of links unrelated to your business;
• believing that an llms.txt file guarantees visibility.

An isolated file or trick cannot replace a clear brand, an accessible website and content that deserves to be recommended.

Frequently asked questions

Can you guarantee that a website will be cited by ChatGPT or Gemini?

No. Platforms select sources through their own systems, and an answer may change from one query to the next. Serious optimisation can improve eligibility and understanding, but never guarantee a citation.

Do you need separate content for Google and AI systems?

Not necessarily. Clear, original, structured and verifiable content serves both. Certain adaptations - direct answers, definitions, tables and FAQs - can nevertheless make extraction easier.

Does blocking GPTBot prevent a website from appearing in ChatGPT search?

OpenAI distinguishes GPTBot, which relates to potential model training, from OAI-SearchBot, which relates to search. Their permissions can therefore be configured separately.

Make your website visible to people and AI engines

The best strategy is to create the website you would want to find as a prospect: fast, precise, reassuring, accessible and supported by evidence. VulcainDesign can audit the technical structure, content, structured data and consistency of your online presence. Let’s discuss your visibility in conventional and generative search.

Useful sources: Google Search Central - AI features and your website and OpenAI - publishers and developers FAQ.