Generative Engine Optimisation: How Brands Get Cited by AI
Generative engine optimisation (GEO) explained: how brands get mentioned and cited by ChatGPT, Gemini, Perplexity and Copilot, and how to measure it.
In February 2026, OpenAI said ChatGPT had reached 900 million weekly active users 1. Three months later Google said AI Mode had passed 1 billion monthly users 2. A growing share of product research, supplier shortlisting and "what is the best..." questions now ends in an AI-written answer that names a handful of brands and cites a few sources. Generative engine optimisation (GEO), also called answer engine optimisation (AEO), is the practice of making sure your brand is one of them.
This article explains what the research says actually drives AI citations, how GEO differs from SEO, and how to measure it with the tools now available.
Where GEO came from
The term comes from a 2023 research paper by Princeton University and partners, which introduced a benchmark for testing how content changes affect visibility in generative engine responses 3. The authors reported that optimisation could raise a source's visibility in those responses by up to 40%, while noting that what works varies by domain 3.
Industry analysts saw the shift coming. In February 2024, Gartner predicted that traditional search engine volume would fall 25% by 2026 as people turned to AI chatbots and virtual agents 4. Gartner's advice then still holds: "content utility and quality still reigns supreme," and companies need "unique content that is useful to customers and prospective customers" 4.
What actually drives AI citations
Much GEO advice is a checklist: add FAQ blocks, summary boxes, author bios and schema markup. A large study published in October 2026 by researchers at the University of Toronto tested those assumptions. It tracked 90 days of daily answers from ChatGPT, Gemini and Perplexity across 651 consumer queries (143,249 answers) and a smaller B2B software market 5. Its findings are worth reading closely:
- Retrieval comes first. Every page ChatGPT cited had first been returned by ChatGPT's own search step, and retrieval explained about 32% of the variation in citations between pages on the same site 5. If the AI's search does not find your page, nothing else matters.
- AI search is not Google search. 97% of the pages ChatGPT retrieved were absent from Google's top 50 results for the same query 5. Ranking well on Google does not guarantee AI visibility.
- Matching the question matters. Within a website, a page whose best paragraph closely matched the query earned about 15.5% more citations, and a better title match about 14.1% more 5.
- Formatting tricks did not hold up. FAQ blocks, summary boxes, author boxes, schema markup and page speed looked helpful when comparing different websites, but none had a significant effect when comparing pages within the same site 5. The apparent benefit came from the kinds of sites that use them.
- Brand knowledge is a major factor. ChatGPT wrote known brands into its search queries 67% of the time, against 11% for unknown brands, and 46% of its brand mentions had no supporting cited page 5.
- Citations drift. Only 40% of the pages ChatGPT cited for a query were still cited two weeks later, and 14% after ten weeks 5.
The authors describe these as associations rather than proven causes 5. The practical reading is still clear: AI visibility depends on being findable by each engine's retrieval system, on directly answering the question, and on the model already knowing your brand.
How GEO differs from SEO
GEO builds on SEO rather than replacing it. Pages still need to be crawlable, indexed and useful. The differences lie in what you optimise for and how you measure it.
| SEO | GEO / AEO | |
|---|---|---|
| Goal | Rank a page and earn the click | Be mentioned or cited inside the answer |
| Unit of success | Position, clicks, sessions | Citations, brand mentions, share of answers |
| Main levers | Keywords, links, technical health | Retrievability, direct answers, brand presence across the web |
| Stability | Rankings change gradually | Citations shift week to week 5 |
| Measurement | Mature (Search Console, analytics) | New and partial (citation and impression reports) |
The biggest strategic difference is that an AI answer can recommend a brand without linking to it. Because nearly half of ChatGPT's brand mentions in the Toronto study had no supporting citation 5, a brand's reputation in reviews, press coverage, forums and comparison content becomes part of the optimisation task, not just its own website.
Why it is worth the effort
AI referrals are still small next to organic search for most sites, but they are growing quickly and they convert. Adobe reported that AI-sourced traffic to US retail sites rose 393% year on year in the first quarter of 2026, and that in March 2026 this traffic converted 42% better than non-AI traffic, a reversal from March 2025 when it converted 38% worse 6. Adobe also found US retail product pages scored lowest of the page types it tested for machine readability, averaging 66% 6.
How to measure GEO
Measurement has improved in 2026 but remains incomplete:
- Microsoft Bing Webmaster Tools. Its AI Performance report, launched in public preview in February 2026, shows how often your pages are cited in Microsoft Copilot, Bing's AI summaries and selected partner integrations, with sample "grounding queries" the AI used to find your content 7. It does not show clicks or traffic 7.
- Google Search Console. Since August 2026, all sites can see impressions from AI Overviews, AI Mode and Discover's AI features, but not clicks 8.
- Analytics referrals. Segment sessions from chatgpt.com, perplexity.ai, gemini.google.com and copilot.microsoft.com to track volume and conversion rate over time.
- Prompt tracking. Run a fixed set of category and comparison prompts on each engine regularly and record which brands are named and cited. Because citations drift 5, measure share over weeks rather than reacting to single snapshots.
A practical GEO plan
- Fix retrievability. Make sure AI crawlers are not blocked, key pages render without heavy scripts, and product and service pages are machine readable 6.
- Answer real questions directly. Map the prompts buyers use and give each a page or section whose opening paragraph answers it plainly 5.
- Publish what others cannot. Original data, pricing logic, comparisons and expert commentary give engines a reason to cite you 34.
- Build brand presence off-site. Reviews, earned media, partner pages and community discussion feed the model's knowledge of your brand 5.
- Cover Bing as well as Google. Copilot citations depend on Bing's index, and Bing now reports them 7.
Key takeaways
- AI assistants now reach hundreds of millions of people every week, making brand presence inside answers a marketing priority 12.
- Recent research suggests retrieval, query match and brand knowledge drive citations far more than formatting tricks 5.
- AI search results differ heavily from Google's rankings, so SEO success does not transfer automatically 5.
- AI referral traffic is small but growing fast and, in retail, now converts better than other traffic 6.
- Bing and Google both report AI citations or impressions in 2026, though neither yet reports clicks 78.
What this means for your brand
Start with a baseline: a set of 30 to 50 prompts that reflect how your buyers search, tracked monthly across ChatGPT, Gemini, Perplexity and Copilot. Connect Bing Webmaster Tools and Search Console reporting to the same dashboard as your analytics referrals. Then prioritise the pages that answer your highest-value prompts, and treat PR, reviews and partner content as part of search, not separate from it. Expect results to move week to week and judge progress over quarters.
Sources
- ChatGPT reaches 900M weekly active users - TechCrunch, February 2026
- 100 things we announced at Google I/O 2026 - Google, May 2026
- GEO: Generative Engine Optimization - Aggarwal et al., arXiv, June 2024
- Gartner Predicts Search Engine Volume Will Drop 25% by 2026, Due to AI Chatbots and Other Virtual Agents - Gartner, February 2024
- What Drives AI Citations? Content, Retrieval and Model Knowledge in Generative Engine Optimization (GEO) - Chen and Koudas, University of Toronto, arXiv, October 2026
- AI traffic grows but retail sites lag in AI search visibility - Adobe, April 2026
- Bing Webmaster Tools officially adds AI Performance report - Search Engine Land, February 2026
- Google Search Console AI Reports Rolled Out Worldwide - Search Engine Journal, August 2026
Figures and product details are as reported by the sources listed, at the time of writing. Platforms change quickly, so check the latest documentation before acting on any single number.
