Generative Engine Optimisation: How to Get Cited by AI Search in 2026
What the original GEO research actually found, what Google says about AI Overviews and AI Mode, where llms.txt really stands, and a measurement plan for Indian brands.
The short answer
Generative engine optimisation (GEO) means making your content easy for AI answer engines to retrieve, trust and cite. In practice: stay indexed and crawlable, put specific answers with sources and statistics near the top, publish original expertise, keep brand facts consistent across the web, and measure citations directly. Google says no special files, chunking or markup are needed for AI Overviews or AI Mode.
Key takeaways
- The term GEO comes from a KDD 2024 paper that found adding quotations, statistics and citations could raise a source's visibility in generative answers by up to about 40 percent.
- Google's 2026 guide says you do not need llms.txt, AI text files, content chunking, special schema or an AI-specific writing style to appear in its AI features.
- AI Mode launched in India in June 2025 and added Bengali, Kannada, Malayalam, Marathi, Tamil, Telugu and Urdu in October 2025, after Hindi.
- Search Console's generative AI performance reports, announced in June 2026, had rolled out to all sites by late August 2026, showing AI Overviews and AI Mode impressions.
- A monthly prompt panel across ChatGPT, Gemini, Perplexity and AI Mode is still the most direct way to track citations.
Generative engine optimisation (GEO) is the work of making your content easy for AI answer engines such as ChatGPT, Gemini, Perplexity and Google's AI Overviews and AI Mode to find, trust and cite. It sits on top of SEO rather than replacing it, because almost every one of these systems grounds its answers in a web index. The practical recipe is unglamorous: be crawlable, answer early, back claims with evidence, publish what others cannot copy, keep your brand facts consistent, and measure citations directly.
A lot of GEO advice in 2026 is confident and unsourced. This guide sticks to what the original research and the platforms themselves say. For the Gulf view, our UAE colleagues have published a guide to AI search and GEO in the UAE that pairs well with this one.
What did the original GEO research actually find?
The original study found that adding citations, quotations and statistics made a source noticeably more visible in AI-generated answers, while keyword stuffing did not help.
The term comes from "GEO: Generative Engine Optimization" by Pranjal Aggarwal, Vishvak Murahari, Tanmay Rajpurohit, Ashwin Kalyan, Karthik Narasimhan and Ameet Deshpande, accepted to KDD 2024. The abstract reports that GEO methods can boost visibility by up to 40 percent and that their effect varies across domains. A detailed review of the study adds the specifics:
- Benchmark: GEO-bench holds 10,000 queries drawn from nine sources across 25 domains.
- What worked: Cite Sources, Quotation Addition and Statistics Addition gave relative improvements of roughly 30 to 40 percent, with fluency improvements close behind.
- What did not: keyword stuffing was among the weakest methods and, in the Perplexity test, scored about 10 percent below the unedited baseline.
- Who benefits most: lower-ranked sources gained disproportionately, which is good news for smaller Indian brands competing with national publishers.
Two cautions. The experiments rewrote sources inside a controlled pipeline, not the open web. And the findings describe content qualities, not loopholes: be specific, evidenced and readable.
How do Google AI Overviews and AI Mode choose sources?
Google says a page only needs to be indexed and eligible to show a snippet; there are no extra technical requirements for its AI features.
Google's Search Central page on AI features and your website states this directly. Both features may use "query fan-out", running several related searches across subtopics, which is why they can surface a wider set of pages than the classic top ten. Controls are the usual ones: nosnippet and max-snippet limit what is shown, robots.txt governs Googlebot, and the Google-Extended token limits use in Google's other AI systems.
Google's guide to optimising for generative AI features, last updated in July 2026, goes further. It says you do not need machine-readable AI text files or Markdown copies, artificial chunking, a separate writing style for AI, special schema markup, or manufactured brand mentions. It recommends original, expert-led "non-commodity" content, solid technical SEO, good images and video, and up-to-date Merchant Center and Business Profile data.
Are AI Overviews and AI Mode available in Indian languages?
Yes. Google said in May 2025 that AI Overviews reached more than 200 countries and territories and more than 40 languages. AI Mode launched in India in June 2025, added Hindi in September 2025, and on 8 October 2025 added Bengali, Kannada, Malayalam, Marathi, Tamil, Telugu and Urdu. On 27 March 2026 Google India said Search Live in AI Mode also works in Gujarati and Odia, among others. For brands in Kerala, Karnataka or Tamil Nadu, AI answers can now be generated in the language customers actually use. Our piece on regional language content in India covers how to create it well.
Do you need an llms.txt file?
Not for Google Search. llms.txt is optional, cheap, and aimed at agents and developer tools rather than rankings.
llms.txt was proposed by Jeremy Howard on 3 September 2024 as a Markdown file at a site's root that gives language models a short, curated map of its most useful content. It is not a web standard. Google's 2026 guide says Search does not use such files and they neither help nor hurt visibility. Chrome's Lighthouse 13.3 did add an experimental agentic browsing audit that checks for the file, but Search Engine Journal noted that audit concerns browser agents, not search rankings or citations. Add one if it costs an hour; never expect it to move citations.
What actually improves your chances of being cited?
Crawl access, early answers, evidence, originality and consistent entity facts do most of the work.
- Be crawlable by the engines you want. OpenAI's crawler documentation says sites that block OAI-SearchBot will not be shown in ChatGPT search answers, while GPTBot covers model training, so you can allow one and block the other. Perplexity's crawler page lists PerplexityBot for search results. Our technical SEO checklist includes a robots.txt template.
- Put the answer near the top. A 40 to 70 word direct answer, followed by detail, is easy for any system to extract and quote.
- Add evidence. Statistics with sources, named standards and short attributed quotations are the edits the GEO study found most effective.
- Write non-commodity content. First-hand process, original comparisons and decision rules are harder to replace with a summary.
- Use structured data honestly. Organisation, LocalBusiness, Article and Product markup help engines understand entities, but must match visible content and are not required for AI features.
- Keep entity facts consistent. Brand name, founding year, cities, founders and services should match across your site, Google Business Profile, LinkedIn, directories and press.
- Earn real mentions. Coverage, reviews and expert quotes help. Paid or fake mentions do not, and Google's guide says so.
Common GEO mistakes we see in India
- Blocking every AI bot in robots.txt by default, then wondering why ChatGPT never cites the site.
- Hiding key facts such as prices, cities served or fees inside PDFs or images.
- Different founding years or addresses on Justdial, LinkedIn and the website.
- Publishing hundreds of near-identical AI-written pages, the opposite of non-commodity content.
How is GEO different from traditional SEO?
GEO keeps SEO's foundations but changes the unit of value from a ranked page to a quoted passage or fact.
| Dimension | Traditional SEO | GEO |
|---|---|---|
| Goal | Rank a page and earn the click | Be retrieved, quoted and cited inside an answer |
| Unit of value | The page | The passage, fact or entity |
| Main signals | Relevance, quality, links, page experience | The same, plus extractable answers, evidence and entity consistency |
| Measurement | Rankings, clicks, impressions | AI impressions, citation share in tested prompts, referral traffic from AI tools |
| Crawler control | Googlebot, Bingbot | Also OAI-SearchBot, GPTBot, PerplexityBot and the Google-Extended token |
How do you measure GEO?
Use four layers: Search Console's AI report, a prompt panel, referral analytics and server logs.
- Search Console AI report: Google introduced Search Generative AI performance reports in June 2026. At launch they showed only impressions, how often your URLs were served as sources in AI Overviews, AI Mode and AI-powered Discover, with no clicks or queries, according to Collective Measures. By 31 August 2026 the reports had rolled out to all sites worldwide. Clicks from AI features still sit inside the Web search type of the main Performance report.
- Prompt panel: 30 to 100 real buyer questions, run monthly in ChatGPT, Gemini, Perplexity and AI Mode, recording whether you are cited, mentioned or absent, and who is cited instead.
- Referral analytics: segment sessions from chatgpt.com, perplexity.ai, gemini.google.com and similar referrers in GA4, and watch conversion rate, not just volume.
- Server logs: confirm that search-oriented AI crawlers actually fetch your key pages.
What should you do in the first 30 days?
- Week 1: check robots.txt and server logs for OAI-SearchBot, PerplexityBot and Googlebot access, and list your ten most valuable pages.
- Week 2: audit brand facts across your site, Google Business Profile, LinkedIn and the main Indian directories, and fix every mismatch.
- Week 3: rewrite the ten pages answer-first, adding sourced statistics, a comparison table and your own method or decision rules.
- Week 4: set up the prompt panel and GA4 referral segments, record a baseline, and repeat monthly.
Treat the prompt panel like a ranking report: noisy, directional and useful over months. To build depth that engines draw on repeatedly, see our guide to topical authority and content strategy.
How we approach GEO at Old Port
We start with an entity audit, checking that the facts about a business match everywhere, then fix crawl access, then rewrite priority pages answer-first with sources and original method. Only after that do we build the prompt panel, so we measure a site that deserves to be cited. Our SEO and GEO service runs this in India, and our UAE team offers the same SEO and AI search visibility service for Gulf markets.
Frequently asked questions
What is generative engine optimisation?
Generative engine optimisation (GEO) is the practice of making your content easy for AI answer engines such as ChatGPT, Gemini, Perplexity and Google AI Overviews to find, trust and cite. It builds on SEO fundamentals such as crawlability and quality, and adds a focus on direct answers, evidence, original expertise and consistent brand information.
Does Google require special markup to appear in AI Overviews?
No. Google states there are no additional requirements for AI Overviews or AI Mode beyond being indexed and eligible to show a snippet. Its 2026 guide adds that special schema, AI text files, chunking and AI-specific writing are not needed. Structured data can still help rich results, but it must match visible content.
Is llms.txt a ranking factor?
No. llms.txt is a September 2024 community proposal for a Markdown file that points language models to key content. Google's 2026 guidance says Search does not use such files, so they neither help nor harm visibility. Chrome Lighthouse has an experimental agent audit that checks for it, but that concerns browser agents, not rankings.
Is Google AI Mode available in Malayalam and Kannada?
Yes. AI Mode launched in India in June 2025, added Hindi in September 2025, and on 8 October 2025 Google added Bengali, Kannada, Malayalam, Marathi, Tamil, Telugu and Urdu. In March 2026 Search Live in AI Mode also gained Gujarati and Odia. Regional-language content is now directly relevant for businesses in Kerala and Karnataka.
Can Search Console show AI Overviews performance?
Yes, partly. Google introduced Search Generative AI performance reports in June 2026, and by late August 2026 they had rolled out to all sites worldwide. They show how often your URLs appear as sources in AI Overviews and AI Mode as impressions, but not clicks or queries. Clicks remain inside the main Web search report.
How can I tell if ChatGPT cites my website?
Build a fixed panel of real buyer questions and test them monthly in ChatGPT, Perplexity, Gemini and Google AI Mode, logging whether your site is cited or mentioned. Add GA4 segments for referral traffic from AI tools, and check server logs to confirm OAI-SearchBot and similar crawlers are fetching your important pages.
Sources
- arXiv: Aggarwal et al., GEO: Generative Engine Optimization (KDD 2024)
- blckalpaca: The Princeton GEO study, methodology, results and critique
- Google Search Central: AI features and your website
- Google Search Central: Optimizing your website for generative AI features on Google Search
- Google Search Central Blog: Introducing Search Generative AI performance reports in Search Console (June 2026)
- Collective Measures: Google Search Console introduces Search Generative AI performance reports
- Search Engine Journal via ATDB: Google Search Console AI reports rolled out worldwide (August 2026)
- Google Blog: AI Overviews expansion, May 2025 update
- TechCrunch: Search Live comes to India, AI Mode gets more languages (October 2025)
- Google India Blog: Search Live comes to more Indian languages (March 2026)
- llmstxt.org: The llms.txt proposal
- Search Engine Journal: Google's llms.txt guidance depends on which product you ask (May 2026)
- OpenAI: Overview of OpenAI crawlers
- Perplexity: Perplexity crawlers
Cite this article
Jacob, A. I. (2026). Generative Engine Optimisation: How to Get Cited by AI Search in 2026. Old Port India. https://oldportindia.com/insights/generative-engine-optimization-guide/ Plain-text version