GEO (generative engine optimisation) is the work of getting your brand named, described correctly and recommended by AI tools such as ChatGPT, Gemini and Perplexity. It builds on SEO, but the goal is different: being in the answer, not just on the results page.
GEO in one sentence
SEO gets your page ranked. GEO gets your brand into the answer. When someone asks an AI "who is the best SEO expert in Melbourne?", GEO is the work that decides whether you are named, how you are described, and whether the AI links to you.
The three terms you will hear are related but not identical. SEO (search engine optimisation) is about ranking pages in Google and Bing. AEO (answer engine optimisation) is about being quoted and cited inside answer features such as Google AI Overviews and AI Mode. GEO is about the generative tools people talk to directly: ChatGPT, Gemini, Perplexity, Claude and Copilot. In practice I plan all three together, because they share most of the same inputs: a crawlable site, specific and current content, and a brand that is mentioned on credible sites.
The reason GEO exists as a discipline now is scale. OpenAI reported 400 million weekly active ChatGPT users in February 2025 and 900 million in February 2026. That is a very large number of people asking a machine for recommendations instead of scanning ten blue links.
How AI tools decide which brands to recommend
AI assistants draw on two sources: what they learned during training, and what they find when they run a live web search before answering. Brands that are mentioned consistently across credible sites, and whose own pages rank for the right questions, are recommended more often.
When you ask a question, most assistants now "fan out" into several sub-searches, read the top results, and write one answer from them. That is why classic SEO still matters: if your page is not found in those live searches, it cannot be cited. Ahrefs' research for its AEO course found that about 38% of Google AI Overview citations come from pages already in Google's top 10, and that Perplexity is the closest to classic Google, with roughly 29% of its citations ranking in the top 10.
The other half of the equation is consensus. Models learn who you are from how often, and in what context, your brand appears across the web. In Ahrefs' study of 75,000 brands, branded web mentions had a 0.664 correlation with appearing in AI Overviews, higher than backlinks or Domain Rating. In plain English: being talked about matters more than being linked to.
| Signal | What it means for you | Where it comes from |
|---|---|---|
| Live search ranking | Your page is found when the AI searches | Classic SEO |
| Branded mentions | The AI has "heard of you" in the right context | PR, reviews, lists, forums |
| Freshness | Recently updated pages are preferred | Content maintenance |
| Specific facts | The AI can quote you accurately | Official FAQ, data, named products |
What the numbers say about AI search in 2026
AI answers now sit on a large share of searches and they reduce clicks. Comscore data reported in September 2026 put AI Overviews on roughly 39% of US desktop Google searches in June 2026, up from about 26% a year earlier. Pew Research found people clicked a normal result in 8% of visits when an AI summary was present, versus 15% without one.
SparkToro's analysis of Similarweb panel data has put the share of Google searches ending without a click to any website at around two-thirds in early 2026. The practical point is not that search is dying. Google still handles more than five trillion searches a year. It is that the answer increasingly happens on the results page or inside a chat, so the brand named in that answer wins even when nobody clicks.
Five things to fix first
Start with the changes that remove blockers and give AI accurate facts to repeat. These five take days, not months.
- Let AI crawlers in. Check robots.txt is not blocking GPTBot, OAI-SearchBot, ClaudeBot, PerplexityBot or Google-Extended. Ahrefs found 5.9% of 140 million sites block GPTBot, making them invisible to ChatGPT.
- Publish an official FAQ. Real facts: what you do, where, for whom, prices or ranges, founding year, named people. This is what the AI repeats when someone asks about you.
- Refresh your most valuable pages. Ahrefs found 76% of ChatGPT's most-cited pages had been updated within 30 days. Make real updates, not just a new date.
- Earn mentions where AI looks. Reviews, "best of" lists, industry publications and relevant Reddit threads. About 44% of cited pages in Ahrefs' study were listicles.
- Fix what AI gets wrong. Ask the main assistants about your brand. Correct your own content first, then ask publishers to update theirs.
How to measure GEO
No single report shows AI visibility. Combine AI referral traffic in GA4, AI crawler activity in your server logs, and a "How did you hear about us?" question, then track share of voice against competitors monthly.
Expect the traffic numbers to look small at first. Conductor's 2026 benchmarks put AI referrals at roughly 1% of all website traffic, with ChatGPT sending the large majority of it. Ahrefs reported that AI visitors to its own site converted at around 23 times the rate of organic visitors. Small volume, high intent. Measure it properly and it earns its place in the plan.
| Measure | Tool | Check |
|---|---|---|
| AI referral visits | GA4 custom channel | Monthly |
| AI crawler hits | Server logs / CDN | Monthly |
| Self-reported source | Form question | Every lead |
| Share of voice | Fixed prompt list, 3–5 competitors | Monthly |
The three kinds of AI visibility
Visibility is a range, not a switch. Ahrefs describes three states: cited with a link, mentioned without a link, and not visible. Each is worth something different, and most brands sit in the middle state far more often than they realise.
Cited with a link is the easiest to measure and the only one that sends traffic directly. It is also the rarest: Ahrefs found links appear in about 28% of AI mentions on average, ranging from 52% on Perplexity to 11% in AI Overviews. Mentioned without a link is word of mouth at scale. It sends no visit you can see, but it shapes the decision, and every credible mention is another example the models learn from. Not visible is the state you fix first, because you cannot improve what you do not appear in.
For a brand selling a product or service, the priority is being named for category questions ("best", "near me", "for small business") so buyers click through or search for you by name. For a publisher, being cited builds authority in the model over time. Knowing which state you are in, per platform, is the starting point of every plan I write.
Common questions about GEO
Three questions come up in almost every first call. The short answers: no, GEO does not replace SEO; yes, small brands can win because AI answers favour specific, current sources over big ones; and results usually show first in Perplexity and Google AI features, then in ChatGPT.
"Do we need a separate GEO agency?" Usually not. The inputs are the same as good SEO plus earned mentions and a different measurement setup, so one plan and one team is simpler and cheaper. "Can we opt out of AI Overviews?" You can remove your pages as a source, but Google still generates the answer from competitors' pages, so opting out rarely helps a brand that wants to be found. "How long does it take?" Technical fixes show within weeks; mentions and content build over two to three months; the training-data effect compounds over a year or more.
The honest caveat: AI answers are probabilistic. The same question can produce a different answer on a different day, and Ahrefs found roughly 45% of AI Overview citations change on refresh. The goal is to raise your odds consistently across a topic, not to "rank number one" for a single prompt.
What to do this month
Run the five fixes above, set up measurement, and pick ten buyer questions to track. In 90 days you will know where you stand and what moved.
Quick answers
SEO gets your page ranked. GEO gets your brand into the answer. When someone asks an AI "who is the best SEO expert in Melbourne?", GEO is the work that decides whether you are named, how you are described, and whether the AI links to you.
AI assistants draw on two sources: what they learned during training, and what they find when they run a live web search before answering. Brands that are mentioned consistently across credible sites, and whose own pages rank for the right questions, are recommended more often.
Start with the changes that remove blockers and give AI accurate facts to repeat. These five take days, not months.
No single report shows AI visibility. Combine AI referral traffic in GA4, AI crawler activity in your server logs, and a "How did you hear about us?" question, then track share of voice against competitors monthly.
Visibility is a range, not a switch. Ahrefs describes three states: cited with a link, mentioned without a link, and not visible. Each is worth something different, and most brands sit in the middle state far more often than they realise.
SOURCES
- OpenAI, weekly active user announcements (400M, February 2025; 900M, 27 February 2026), via TechCrunch / Yahoo Finance
- Ahrefs Academy, Answer Engine Optimization course (2025): citation, freshness and traffic studies
- Comscore, AI Overviews share of US desktop Google searches, July 2025 vs June 2026 (as reported by industry press, 2026)
- Pew Research Center, "Google users are less likely to click on links when an AI summary appears" (July 2025; 900 US adults, 68,879 searches, March 2025)
- SparkToro and Similarweb, US Google zero-click study, January–April 2026 (68.01%; 60.45% in 2024)
- Conductor, AI Search Benchmarks 2026 (21.9M queries; AI referral share of web traffic)
- OpenAI, "Overview of OpenAI crawlers" (GPTBot, OAI-SearchBot, ChatGPT-User)
- Google, "Our latest AI features in Search" and robots.txt / Google-Extended documentation




