GEO: How to Get ChatGPT, Gemini and Perplexity to Recommend You
GEO in practice: how to get ChatGPT, Gemini and Perplexity to cite your business. AI crawlers, quotable content, brand mentions and a 90-day plan.
Read
Google AI Overviews and AI Mode cite pages that are indexed and eligible to show a text snippet. There is no submission form, special markup or “AI file”: sources are picked by the same ranking systems that power regular results, except the model splits each question into many sub-queries and cites whichever pages best answer each thread. Sound technical SEO, content built on first-hand experience and a clear structure are what move the needle.
AI Overviews are Gemini-generated summaries that Google shows above the organic results — but only when its systems decide a summary adds something beyond a list of links. AI Mode is a separate tab where you hold a conversation with Search: ask a complex question, follow up, compare options. In October 2025 Google expanded AI Mode to more than 35 new languages, bringing it to over 200 countries and territories.
Both are now mainstream. At I/O in May 2026, Google said AI Overviews had passed 2.5 billion monthly active users and AI Mode 1 billion (Sundar Pichai’s keynote). The two features are also merging: from an AI Overview you can ask a follow-up and continue the conversation in AI Mode with the context preserved (Google’s Search I/O 2026 update).
| Feature | AI Overviews | AI Mode |
|---|---|---|
| Where it appears | Above the results on a standard results page | A dedicated tab in Search and in the Google app |
| When it triggers | Automatically, when the system thinks it helps | When the user switches to it or follows up from an AI Overview |
| Typical queries | Complex informational questions | Multi-step questions, comparisons, planning |
| Sources | Links beside and inside the summary | Links in the answer, usually a broader set of sites |
Google’s guide to optimising for generative AI features (published May 2026) spells it out. Answers are built with retrieval-augmented generation, also called grounding: the core ranking systems fetch relevant, fresh pages and the model writes a response with clickable links to them. That is why Google insists that optimising for AI Overviews and AI Mode is, from its point of view, still SEO rather than a separate discipline.
The second half of the mechanism is query fan-out: the model fires off a set of related queries in parallel to gather more information. A question such as “how do I get my online shop ready for Black Friday” might fan out into threads on server performance, discount codes, email campaigns and returns logistics. A page that nails one of those threads can earn a citation even if it ranks nowhere near the top for the main question.
The data backs this up. In July 2025 Ahrefs found that 76% of pages cited in AI Overviews ranked in the top 10 for the same query (Ahrefs, July 2025). Its March 2026 rerun — 863,000 keywords and 4 million URLs — put the figure at 37.9%, with 31% of cited pages ranking outside the top 100 (Ahrefs, March 2026). The authors note they improved citation detection, so part of the gap is methodological. The direction is still clear: ranking for the head term helps, but covering the subtopics matters just as much.
Google’s “AI features and your website” documentation states there are no additional technical requirements. A page needs to meet the same conditions as for regular Search:
Watch your preview controls. In Google’s May 2025 guidance, John Mueller warns that more restrictive directives will limit how your content features in AI experiences. It’s a classic migration mistake: a noindex or nosnippet left over from staging goes live and nobody notices for weeks.
The most useful part of Google’s guide is its mythbusting section on popular “AEO/GEO” tactics. For Google Search, these are unnecessary or even risky:
| Tactic | Google’s position | Do this instead |
|---|---|---|
| An llms.txt file | Google Search doesn’t use it; it neither helps nor hurts | Get robots.txt, your XML sitemap and indexing right |
| Special “AI schema” | Doesn’t exist; structured data isn’t required | Use schema for rich results, matching the visible content |
| Chunking text into small pieces | Unnecessary; there’s no ideal page length | Write for people: clear sections, headings, paragraphs |
| Separate pages for keyword variants | AI systems understand synonyms and overall meaning | One thorough page instead of ten variants |
| Arranged, inauthentic mentions | “Not as helpful as it might seem” — ranking and spam systems still apply | Earn real mentions: trade press, case studies, reviews |
| Mass-producing pages for questions | May breach the scaled content abuse policy | Fewer pages, each with genuine value |
That doesn’t make llms.txt or structured data useless everywhere — other AI systems may use them. We cover the differences in our article on structured data, llms.txt and robots.txt for AI crawlers.
Google rewards “non-commodity” content: a distinct point of view, first-hand experience, data nobody else has. A generic “10 tips for choosing X” loses to a piece in which someone shows what worked in practice and why. These are the rules we follow for expert content:
Here is the difference in practice, answering “Do I need structured data to appear in AI Overviews?”:
Weak: “Structured data is a topic surrounded by plenty of myths. A lot depends on your specific situation, so it’s best approached case by case.”
Strong: “No. Google doesn’t require structured data for AI Overviews and there is no special schema for AI. Use it for rich results, and make sure the markup matches what’s visible on the page.”
The second version answers in its first word, then gives the fact and the condition. It is also quicker for a human to read — which is the whole point: Google’s guide boils everything down to whether visitors would find the content satisfying.
Google’s guide names two data sources beyond your website: Google Business Profile for local and service businesses, and Merchant Center product feeds for retailers. Both can make your services and products visible in AI responses as well as in standard results. If you run an online shop, a current feed with prices, availability and images is no longer optional.
ChatGPT, Perplexity and other AI assistants rely on their own retrieval, so visibility there calls for a slightly different strategy — see our guide to generative engine optimisation (GEO) for ChatGPT, Gemini and Perplexity.
Two independent analyses are quoted most often:
Google’s counterpoint is that clicks from results pages with AI Overviews are “higher quality”, with users spending more time on site. The practical upshot: expect fewer clicks on informational queries at the same rankings. How-to content now mainly builds visibility and trust, while revenue is measured on your service and offer pages — through form fills, calls and enquiries.
Since 31 August 2026, every site has a Generative AI performance report in Search Console. It shows how often links to your site appeared in AI Overviews and AI Mode, broken down by page, country, device and date. Keep a few caveats in mind:
An example from our own backyard: on olafmotion.com, the motion design portfolio of OrbitBuilder’s co-founder, Search Console shows 127,000 impressions and 6,190 clicks over 16 months at an average position of 8.7, with daily clicks climbing from near zero to 100–170 in October and November 2026. We read figures like these on three levels: the standard Performance report, the AI report (which pages get cited) and enquiries from the contact form. Only all three together tell you whether rising impressions turn into clients.
On top of that, spot-check 20–30 key questions by hand once a month and note who gets cited. Google also warns against tools that claim access to “internal” Search metrics — none of them have it.
| Control | Where | What it does |
|---|---|---|
| “Search generative AI” setting | Search Console → Settings | Excludes the site from AI Overviews, AI Mode and generative AI in Discover; no impressions or traffic from those features, no effect on rankings elsewhere in Search; usually takes effect in 1–2 days (Search Console Help) |
nosnippet |
meta robots or X-Robots-Tag header | Blocks text snippets in results, including AI features |
data-nosnippet |
HTML element attribute | Excludes a specific part of the page from snippets |
max-snippet |
meta robots | Caps snippet length |
noindex |
meta robots | Removes the page from Search entirely |
Google-Extended |
robots.txt | Limits use of your content for training Gemini models and grounding outside Search; per Google’s crawler documentation it doesn’t affect inclusion in Google Search |
For most businesses opting out makes little sense: you lose visibility exactly where customers increasingly search. Consider it for paywalled content or data you’d rather not share.
nosnippet.Want to know whether your site stands a chance of being cited in Google’s AI answers? Our SEO and GEO service covers the technical side, your content and your Search Console data — get in touch and we’ll walk you through a plan.
No. Sources in AI Overviews and AI Mode are selected by Search systems using the same signals as organic results. There is no submission form either — the requirements are being indexed, snippet-eligible and not opted out of AI features in Search Console.
No. Google's 2026 guide states that Google Search doesn't use llms.txt and the file neither helps nor hurts visibility. You can keep one for other AI tools, but don't expect any effect in Google.
No, although ranking well helps. In Ahrefs' March 2026 analysis, about 38% of cited pages ranked in the top 10 for the same query and the rest ranked lower. That's down to query fan-out: the model also looks for answers to related sub-queries.
Open the Generative AI performance report in Search Console, available to all sites since 31 August 2026. It shows impressions by page, country, device and date. If you can't see it, your site either has too few AI impressions or has opted out.
According to Google, no: the "Search generative AI" setting in Search Console isn't used as a ranking signal for other parts of Search. You will, however, lose all impressions and traffic from AI Overviews, AI Mode and generative AI features in Discover.
AI answers draw on the Search index, so Google first has to reprocess the page. Google says this can take anywhere from a few days to a few months. You can speed things up by requesting indexing in the URL Inspection tool.
Insights

GEO in practice: how to get ChatGPT, Gemini and Perplexity to cite your business. AI crawlers, quotable content, brand mentions and a 90-day plan.
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.