Is GEO worth it? Why AI visibility matters
AI assistants send few clicks but name brands in answers. What three October 2026 data points show, and how to start measuring your AI visibility.
- geo
- ai-search
- data
- measurement
Checked 22 October 2026.
Short answer: Yes, if your buyers ask AI assistants. The click numbers are small, but clicks are the wrong place to look: an assistant can name your brand, or a competitor's, in an answer without anyone visiting a website. Three October 2026 data points point the same way: measure mentions and citations, not only referral sessions.
What does the data on AI search traffic say?
| Source | Date | What it shows |
|---|---|---|
| Marie Haynes, "GEO is a distraction" | 21 Oct 2026 | AI referrals are a small share of Google organic traffic; visitors who convert, convert at a higher rate |
| Ahrefs, ChatGPT query fan-out study | 12 Oct 2026 | Only a minority of responses contained fan-out queries (extra web searches) |
| Google Search Console, Generative AI performance report | Worldwide since 31 Aug 2026 | Impressions in AI Overviews and AI Mode, no clicks |
Haynes titled her post "GEO is a distraction". We read the same numbers differently, and we say below why. Her numbers themselves are useful.
Why do click numbers miss most of AI visibility?
Haynes looked at 12 months of Google Analytics 4 data from 100 sites of varying sizes, more than 145 million sessions, published on 21 October 2026. ChatGPT sessions were 0.62% of Google organic sessions; all large language models (LLMs) together were 0.73%.
Those numbers measure one thing: people who clicked through. They cannot measure the reader who got an answer, saw your brand named and decided without visiting a page. That is zero-click behaviour, and referral analytics are blind to it. A small referral share therefore tells you how few people clicked, not how often brands were named.
Do AI visitors convert better?
In the same analysis, the conversion rate was 13.8% for ChatGPT visitors against 3.02% for Google organic visitors. An honest note on the base: it comes from the 55 of 100 sites that track conversions in GA4, and rests on 36 LLM conversions (Google organic: 964). Haynes does not define what counted as a conversion, and calls it her "small dataset". Treat it as a hint, not a benchmark. The hint points in a useful direction: the visitors who do arrive from AI have, on these sites, been more likely to act.
Why does being in the answer matter?
Ahrefs studied 66.6 million free, logged-out responses from GPT-5.6 Luna, collected in August and September 2026, and published it on 12 October 2026. It found that 14.4% of them contained fan-out queries, the extra web searches a model sends to build an answer (see query fan-out). Ahrefs corrected its own earlier wording: the text now says "More accurately, 14.4% contained fan-out queries", not that 14.4% "searched the web".
What does that mean for brands? In this sample, most responses contained no fan-out queries. A page you publish is therefore not the only route to a mention, and what a model already holds about your category matters as well. That is a reading of the figure, not something Ahrefs states. The sample is also only free, logged-out use, so it is not every ChatGPT user. What it supports is this: you cannot rely on one tactic, and you have to check what assistants actually say about you.
Does Google treat AI visibility as its own channel?
Since 31 August 2026, Search Console has had a Generative AI performance report (Search), rolled out worldwide. It shows impressions of your pages in AI Overviews and AI Mode, by page, country, date and device. It shows impressions only, no clicks, and is absent if a site has too few impressions. Google also provides a control to opt out of generative AI features.
So Google reports AI appearances separately from classic search clicks. That does not prove business value, but it shows that even Google measures AI visibility as something other than traffic. The report covers only Google's own surfaces, not ChatGPT, Perplexity, Gemini or Claude.
What is the case for starting now?
Put the three together:
- Clicks from AI are small, and clicks miss the mentions that happen without a visit.
- The visitors who do come converted at a higher rate in the one dataset we have, with caveats.
- Google itself now reports AI impressions as a separate figure.
Teams that only watch referral sessions see a number near 1% and may conclude nothing is happening. Teams that also track whether they are named learn earlier where they are missing and which sources the assistants draw on. We cannot quantify the commercial value of a mention, because no public data does, and we will not invent a figure. The reason to start is that measurement is cheap and the blind spot is real.
How do you start?
- Measure mentions. Write 10 to 25 questions your buyers really ask. Ask each assistant, and record whether you are named, in which position, and which competitors appear. Method: how to track your brand's visibility in AI answers.
- Fix what AI reads. Look at the sources that are cited: your own pages, directories, review sites, industry media. Make facts clear and consistent, publish answer-first pages and honest comparisons. See what GEO is and how ChatGPT chooses brands. Google's guidance says the same core ranking and quality systems apply, with no special markup for AI, so good SEO foundations count.
- Track over time. Answers change, so repeat the same questions weekly and compare. Add Search Console AI impressions and referral sessions as supporting numbers.
Nodpick is opening soon and will automate this tracking across assistants. You can join the waitlist. For the practical routine, read the GEO guide for small teams.
Frequently asked questions
Is GEO worth it? For businesses whose buyers ask AI assistants, the data supports measuring it: clicks are small, but mentions happen without clicks and cannot be seen in referral reports. How much a mention is worth is not publicly quantified.
How much traffic do AI assistants send? In Marie Haynes's analysis of 100 sites over 12 months, ChatGPT sessions were 0.62% of Google organic sessions and all LLMs together 0.73%. One dataset; other sites will differ.
Do visitors from AI convert better? In that analysis 13.8% (ChatGPT) against 3.02% (Google organic), but only among 55 sites that track conversions and based on 36 LLM conversions.
How does GEO compare with SEO in the data? SEO still drives far more measurable visits. GEO targets being named in the answer, which referral analytics miss. They overlap in practice.
How do I see my AI visibility in Google? Use the Generative AI performance report in Search Console: impressions in AI Overviews and AI Mode, not clicks.
Sources
- Marie Haynes, "GEO is a distraction", 21 October 2026: https://www.mariehaynes.com/geo-is-a-distraction/
- Ahrefs, "How ChatGPT Does Query Fan-Out (15 Million Queries Studied)", 12 October 2026: https://ahrefs.com/blog/query-fan-out-chatgpt/
- Google Search Central, Generative AI performance reports, June 2026: https://developers.google.com/search/blog/2026/06/gen-ai-performance-reports
- Google Search Console Help, Generative AI performance report: https://support.google.com/webmasters/answer/16984139?hl=en
- Google, new controls for website owners (updated 31 August 2026): https://blog.google/products-and-platforms/products/search/new-controls-website-owners/
- Google Search Central, AI optimization guide (updated 10 July 2026): https://developers.google.com/search/docs/fundamentals/ai-optimization-guide
Related: What is GEO?, How to track your brand's visibility in AI answers, How ChatGPT chooses brands, Zero-click.