How to get into ChatGPT and Google AI answers: a GEO guide for businesses

What is GEO (generative engine optimisation)?

GEO (Generative Engine Optimisation) is optimisation for AI search: preparing your website and your company's information online so that ChatGPT, Perplexity, Gemini, Copilot and Google's AI Overviews can find your pages, understand them and cite them. It's not a replacement for SEO, but an extension of it.

The term was coined in 2023 by researchers from Princeton University and other institutions in the paper ‘GEO: Generative Engine Optimization’. On their test set, they found that source links, quotations and statistics in the text could increase a page's visibility in AI answers by up to 40%. These are lab results, not a promise for any particular website, but the direction is clear: text with verifiable facts and sources was more visible in AI answers. If you'd rather not do this work yourself, we offer it as an AI search optimisation (GEO) service.

How do ChatGPT, Perplexity and Google AI choose their sources?

To build an answer, AI search systems look for pages in their own or their partners' search indexes, pick a few of the most relevant ones and write an answer from them, with links. So the first condition is that your page is in the index and the crawler can read it. Each system has its own crawlers.

Google states in its documentation that there are no additional requirements for AI Overviews: a page must be indexed and eligible to be shown in search with a snippet. In other words, GEO doesn't work without an SEO foundation.

Where should you start with GEO?

First, make sure AI search crawlers can reach your website, your pages are in the Google and Bing indexes, and the main text is visible in the HTML without JavaScript. This is one-off work and usually takes up to a month.

1. robots.txt. Check that OAI-SearchBot, PerplexityBot, Claude-SearchBot, Googlebot and Bingbot are not blocked. A common mistake is accidentally blocking the search crawlers along with the training crawlers.
2. Bing Webmaster Tools and IndexNow. Verify your website and enable IndexNow – Bing then learns about new and updated pages faster.
3. Sitemap. Your sitemap.xml file must be submitted in Google Search Console and Bing Webmaster Tools.
4. Text without JavaScript. Some AI crawlers don't run JavaScript, so text that only JavaScript displays is invisible to them, and animated counters may show ‘0’.
5. Speed and errors. Slow pages, pages returning errors and duplicate pages are less likely to make it into the index.

A minimal robots.txt example: AI search crawlers are allowed to crawl the website, except for the admin area.

User-agent: OAI-SearchBot
User-agent: PerplexityBot
User-agent: Claude-SearchBot
Allow: /
Disallow: /administrator/

User-agent: *
Disallow: /administrator/

Sitemap: https://example.lt/sitemap.xml

A crawler that has its own group in the file no longer reads the general ‘User-agent: *’ rules, so any restrictions have to be repeated in its group too.

How do you write content that AI will cite?

AI usually cites not a whole page but a short passage that answers the question on its own. So open every important page and section with a 40–60-word answer, phrase your headings as customer questions, and replace empty adjectives with facts: ‘from’ prices, turnaround times, terms and names.

• Answer first. The first sentence after a heading is a definition or a direct answer; the details follow.
• One section, one question. A passage must make sense without the rest of the page: instead of ‘this service’, name the service.
• Numbers with units and conditions: ‘from 15 working days’, ‘12-month warranty’, ‘up to 5 users’.
• Tables and lists. AI often conveys comparisons, prices and timelines more accurately when they're presented in tables rather than long paragraphs.
• Date and author. State who wrote the text and when it was last updated.
• Sources. For every external figure, a link to the source and the year.

Compare two openings of the same page. AI is unlikely to cite the sentence ‘We are a professional team offering top-quality solutions’ – it doesn't contain a single fact. The sentence ‘Building a business website takes 10–25 working days, prices start from €499 and there's a 12-month warranty’ answers three questions at once.

Do you need structured data?

Structured data (schema.org markup in JSON-LD format) helps search engines understand exactly who your company is, what services or products it offers and how much they cost. Google says no special markup is needed for AI features, but standard markup is still useful.

The most commonly used types: Organization or LocalBusiness for the company, Service and Product for services and products, Article for articles, FAQPage for FAQs and BreadcrumbList for the breadcrumb trail. The key rule: mark up only what is visible in the page text.

Why do mentions on other websites matter?

AI systems often base an answer on several sources, so companies mentioned by several independent sources are more likely to be named, while contradictory information makes that less likely. Your own website isn't enough: what directories, customers and partners say about you matters too.

• Make your name, address, phone number and business description consistent across your website, Google Business Profile, Bing Places and Lithuanian business directories.
• Collect customer reviews where others can see them: on business profiles and in directories.
• Aim for mentions on partners' websites, industry portals and in the media.
• Don't buy links from dubious websites – it's a risk to both your SEO and your reputation.

Do you need an llms.txt file?

An llms.txt file is a short description of your website for AI systems, with links to the most important pages. It was proposed by Jeremy Howard in September 2024, but it isn't yet an official standard, and Google says AI Overviews don't require any new AI-specific files.

So treat llms.txt as an inexpensive extra step: it's quick to prepare and does your website no harm, but on its own it won't improve your visibility in AI answers. Content, indexing and mentions on other websites have the biggest impact.

How do you measure whether AI mentions your company?

AI visibility can be measured in three ways: by asking AI systems the same set of questions every month, by tracking visitors from AI assistants in Google Analytics 4 reports, and by analysing AI crawler visits in the server logs.

1. Draw up 15–20 questions that customers ask before choosing a supplier, e.g. ‘how much does … cost’, ‘what's the difference between …’, ‘which company should I choose for …’.
2. Ask them in ChatGPT, Perplexity, Gemini and Copilot every month, and note whether your company is mentioned and which sources are cited instead of you.
3. In GA4 reports, track traffic from chatgpt.com, perplexity.ai, gemini.google.com and copilot.microsoft.com (some of these visits are recorded as direct traffic, so the real figure is higher).
4. In the server logs, check which pages are crawled by OAI-SearchBot, PerplexityBot and other crawlers.

Record a baseline before you start – then after a few months you'll see the real change. A convenient way to establish the starting point is an SEO audit (from €49), which includes a check of how ready your website is for AI search.

What not to do in pursuit of AI visibility

Don't trust promises that ChatGPT will definitely mention your company: answers vary with the wording of the question, the user, the location and the time. Also avoid tricks that can harm your regular search rankings.

• Don't hide text meant only for crawlers, and don't put anything in your markup that isn't on the page.
• Don't generate dozens of AI-written texts without facts or fact-checking – such pages are rarely cited and undermine trust in your website.
• Don't block all AI crawlers with a single setting if you want to be cited.
• Don't believe that a single file or plugin will be enough.

Which GEO tasks should you do, and when?

The technical foundation is sorted out in the first month, content and mentions are built continuously, and more stable changes in AI answers are usually seen after 3–6 months.

GEO is built on the same foundation as search engine optimisation (SEO), so it makes sense to plan the two together.

Sources

• Aggarwal P. et al., ‘GEO: Generative Engine Optimization’, 2023 (KDD 2024): arxiv.org
• Google Search Central, ‘AI features and your website’ (updated 2025): developers.google.com
• Google Search Central, Google crawlers and Google-Extended: developers.google.com
• OpenAI, overview of its crawlers (OAI-SearchBot, GPTBot, ChatGPT-User): developers.openai.com
• Perplexity, overview of its crawlers (PerplexityBot, Perplexity-User): docs.perplexity.ai
• Anthropic, overview of its crawlers (ClaudeBot, Claude-SearchBot, Claude-User): support.claude.com
• The llms.txt proposal (2024): llmstxt.org

Want to know whether AI can see your company? In the enquiry form, give us your website address and 3–5 questions your customers ask. Within 1 working day we'll tell you whether AI search crawlers can reach your website and where it's worth starting.