Google AI Overviews and AI Mode: How to Get Cited in AI Answers
How to get cited in Google AI Overviews and AI Mode: technical requirements, query fan-out, llms.txt myths and the new Search Console AI report, sourced.
Read
GEO (generative engine optimisation) is the work of making sure AI assistants such as ChatGPT, Gemini, Perplexity and Google’s AI Overviews can find your business, understand what it does and cite it in their answers. In practice it comes down to four things: letting AI search crawlers in, strong visibility in regular search, content that is easy to quote (direct answers, numbers, sources) and consistent mentions of your brand beyond your own website. There is no paid slot in ChatGPT’s answers, and no single trick that gets you in.
The term “Generative Engine Optimization” was introduced in a research paper by Aggarwal and colleagues, accepted at KDD 2024 (arXiv: GEO: Generative Engine Optimization). The authors describe GEO as a way to increase how visible content is in responses produced by LLM-powered search engines.
SEO competes for a link’s position in a list of results. GEO competes for a place inside the answer itself: a passage of your content or your company’s name, ideally with a link to the source. The audience is huge: in February 2026 OpenAI said ChatGPT had 900 million weekly active users (TechCrunch).
| SEO | GEO | |
|---|---|---|
| Goal | A high ranking for the page | Your brand and content appearing in the AI answer |
| Unit | The whole page (URL) | A passage: paragraph, definition, table, figure |
| Where you see results | Google, Bing | ChatGPT, Perplexity, Gemini, AI Overviews, AI Mode, Copilot |
| Metrics | Rankings, clicks, CTR | Mentions and citations, referral traffic from AI assistants |
| Key signals | Content, links, technical health, user experience | All of that, plus quotability and consistent brand information across the web |
Do not treat GEO as a separate discipline that replaces SEO. It extends it: same foundations, different presentation of the result.
An AI assistant has two sources of knowledge. The first is the data the model was trained on, which you influence only indirectly and with a long delay. The second is live search: the assistant runs queries, reads the pages it finds and builds its answer from them. Here you have real influence, because you decide whether crawlers can read your site and what they find there.
| Tool | Where fresh data comes from | What to set up |
|---|---|---|
| ChatGPT | Its own search, via the OAI-SearchBot crawler | Do not block OAI-SearchBot. According to OpenAI, sites that opt out are not shown in ChatGPT search answers. GPTBot is used for model training and is controlled separately (OpenAI: bots) |
| Perplexity | Its own index, via PerplexityBot | Allow PerplexityBot. Perplexity says it is used to surface and link websites in results, not to train foundation models (Perplexity: bots) |
| Google AI Overviews and AI Mode | Google’s index, using “query fan-out”, i.e. many related searches at once | Focus on indexing and solid SEO. Google: “There are no additional requirements to appear in AI Overviews or AI Mode” (Google Search Central) |
| Gemini | Google Search and content from its index | The Google-Extended token controls use of your content for Gemini training and grounding, but does not affect inclusion in Google Search (Google: crawlers) |
The practical lesson: businesses that block every AI bot “to be safe” often remove themselves from exactly the answers they care about. You can block training (GPTBot, Google-Extended) and still let search crawlers in. They are independent settings.
utm_source=chatgpt.com to its links, which makes that traffic easy to isolate (OpenAI: Publishers and Developers FAQ).The best-documented guidance comes from the GEO paper itself. The authors tested different ways of rewriting content on their GEO-bench query set and measured how often, and how prominently, a page appeared in the generated answer (Aggarwal et al., KDD 2024):
To be fair, these are results from an experiment on a specific set of queries, not a guarantee of how any particular assistant behaves. The direction matches what we see in practice, though, and it translates into a short set of rules:
When someone asks an assistant for “the best company for X”, the model assembles its answer from many sources: rankings, directories, reviews, articles and profiles. If your business appears consistently across them, with the same name, description and scope of services, it is easier to recognise as a distinct entity and to recommend.
If you are changing your company name, this layer is especially fragile. We cover it in our company rebranding guide.
A minimal robots.txt that lets in the ChatGPT and Perplexity search crawlers looks like this:
User-agent: OAI-SearchBot
Allow: /
User-agent: PerplexityBot
Allow: /
Add schema.org structured data (Organization or LocalBusiness, Service, Article, Product) to describe your business in machine-readable form. Google stresses two things here: structured data must match the visible text on the page, and no special markup or new AI files are needed to appear in AI Overviews. “You don’t need to create new machine readable files, AI text files, or markup” (Google Search Central). So treat llms.txt as an optional extra for other tools, not a requirement for visibility in Google.
We go into implementation details in our guide to structured data, llms.txt and robots.txt for AI bots. If you want a site built for SEO and GEO from day one, see our website design and development service.
Below are real Google Search Console results for two sites in our portfolio. This is Google search data, not a count of ChatGPT citations. We show it because this foundation decides whether an AI assistant has anything to cite: AI Overviews and AI Mode draw on Google’s index, and ChatGPT and Perplexity read the pages their crawlers find on the web.
| Site | Period | Clicks | Impressions | Average CTR | Trend |
|---|---|---|---|---|---|
| olafmotion.com | 16 months, with clear growth from August 2026 | 6.19k | 127k | 4.9% (average position 8.7) | Daily clicks from around 0 to 100–170 in October and November 2026 |
| shadowbrand.pl | Since July 2026 | 1.68k | 52.2k | 3.2% | Daily impressions from around 0 to 1.2k–1.6k in November 2026 |
olafmotion.com is the motion design studio of Olaf Klimasz, OrbitBuilder’s co-founder. Both examples have one thing in common: growth came after several months of steady work, not from a single change. The sites started collecting impressions for a wide range of queries, including long, question-style ones, which are exactly the kind people ask AI assistants.
Want to know what ChatGPT, Gemini and Perplexity say about your business today? See how our SEO and GEO service works, or get in touch via our contact page. We will start with a test on your customers’ real questions.
GEO is the optimisation of content and brand presence for AI-powered search engines and assistants such as ChatGPT, Perplexity, Gemini and Google’s AI Overviews. The aim is for the assistant to find your page, understand it and cite it in its answer. The term was introduced in a research paper by Aggarwal and colleagues, accepted at KDD 2024.
There is no official paid placement in ChatGPT’s answers. You influence visibility by making your site accessible to OAI-SearchBot, by publishing specific, high-quality content and through mentions of your brand elsewhere. Treat any offer that “guarantees” a spot in AI answers as a red flag.
No. GEO builds on the same foundations: indexing, valuable content, technical health and authority. Google says no additional optimisation beyond normal SEO best practice is needed for AI Overviews and AI Mode. GEO adds an emphasis on quotability and consistent brand information across the web.
It depends on your goal. If you want to appear in AI answers, do not block search crawlers such as OAI-SearchBot and PerplexityBot. You can opt out of model training separately by blocking GPTBot or Google-Extended, because these are independent settings.
In Google Analytics 4, check referrals from chatgpt.com, perplexity.ai and gemini.google.com. According to OpenAI, ChatGPT adds utm_source=chatgpt.com to its links. Combine this with a regular test of the same questions in AI assistants, because not every recommendation leads to a click.
Not for Google’s AI Overviews and AI Mode. Google says you do not need to create new machine-readable files or AI text files. You can add llms.txt as an extra for other tools, but site accessibility, content and structured data remain the foundation.
Insights

How to get cited in Google AI Overviews and AI Mode: technical requirements, query fan-out, llms.txt myths and the new Search Console AI report, sourced.
Read
A step-by-step SEO audit checklist: indexing, redirects, content, Core Web Vitals, structured data and AI search visibility, with tools and priorities.
Read
How to rank a Kraków business in Google and Maps: Business Profile setup, compliant reviews, location pages and structured data — a practical guide.
ReadWe grow rankings in Google and optimise for AI search engines. No “#1 in a week” promises — just Search Console data you can verify.
We use essential cookies and, with your consent, analytics (Google Analytics, Hotjar) and marketing cookies to understand how you use the site and improve it. Details in our privacy policy.