Terms
AI visibility
Whether and how a company appears in AI answers when people ask about its industry: in ChatGPT, Perplexity, Copilot, Gemini and Google AI Overviews.
Generative Engine Optimization (GEO)
Generative Engine Optimization (GEO) is the optimisation of content, technical signals and external authority with the aim of being mentioned, cited or recommended correctly, and more often, by generative search and answer systems. The term comes from a 2023 research paper (Aggarwal et al., KDD 2024). None of these measures guarantees inclusion in an AI answer. GEO has nothing to do with geography.
AEO, LLMO, AI SEO
Answer Engine Optimization (AEO), LLM Optimization (LLMO) and AI SEO: other names for the same work.
AI search
Search features that return a written answer instead of a list of results: Google AI Overviews and AI Mode, ChatGPT with web search, Perplexity, Copilot and Gemini.
Mention, citation, recommendation
A mention is your name in the answer. A citation is a reference to your website. A recommendation says you are a suitable option for the question. Only the last moves buying decisions. The visibility audit reports all three separately.
Share of answers
The share of answers to a fixed set of questions in which a company appears. The metric that replaces position in this field. Only a measurement if questions, runs and engines are stated.
Sample and spread
The same question does not get the same answer every time. So every question runs several times, in clean sessions. The spread across runs is the threshold: only a movement beyond it counts as change.
How answer engines find their sources
Google: the same search, summarised
AI Overviews and AI Mode sit on the Google index. Google says no special files or markup are needed (Google Search Central, May 2026).
ChatGPT: its own index, a short look
ChatGPT runs its own index. In default mode it opens no page in most answers and works from the title and a snippet around the main heading. JavaScript content is blank to it (Resoneo, 26,900 pages, 2026).
Perplexity: live and source-heavy
Fetches sources fresh for every question and cites visibly. In Germany 7% of AI users use it, Copilot 43% (Bitkom Research, n = 1,003, 2026).
What follows for a website is covered under technical GEO implementation.
What demonstrably correlates
The research is young and mostly vendor-published. We cite only studies with a published method. Two of the four findings concern work outside your own website, which PicoBit AI does not offer.
0.66
correlation between brand mentions on the web and visibility in Google AI Overviews. Backlinks: 0.22.
Source: Ahrefs, 75,000 brands, May 2025 (correlation, not causation)
84%
of all citations in ChatGPT, Claude and Gemini point to editorial content.
53.5%
citation rate for brands with a minimal review profile. Without one: 1%.
Source: Seer Interactive, 804,491 AI responses, 1,926 brands, March 2026
+30 to 40%
visibility from sourced statistics, quotations and citations in your own text. Keyword stuffing: worse than nothing.
Source: Aggarwal et al., KDD 2024, 10,000 queries, 2023 conditions
Supported, unsupported
The evidence supports
- Crawler access per bot: mechanical prerequisite
- Content without JavaScript, clear first characters: shown for ChatGPT
- Sourced figures and citations in the text: peer-reviewed
- Mentions in trade media and review platforms: strongest correlation, outside the website
- Consistent company data: plausible, cheap
The evidence does not support
- llms.txt: called unnecessary by Google
- Schema markup as a citation lever: no effect in a controlled study
- Keyword stuffing "for the AI": counterproductive
- Mass AI-generated content: a risk
- Bought mentions: called ineffective by Google
Common questions
Isn't AI visibility just SEO?
For Google, yes. Not for ChatGPT, Perplexity and Copilot: different indexes, retrieval paths and sources. About two thirds of the work is SEO. The last third is new; more under GEO agency.
How big is the effect?
Direct traffic from AI assistants is small, under 2% of desktop visits (Datos/SparkToro, Q1 2026). The bigger effect is before the click: enquiries never made because another name came up.
How long does it take?
Perplexity reacts in days, Google with your search visibility, ChatGPT slowest. Experience, not promises. That is why we measure trends over months.
Should we block AI crawlers?
Decide per bot. Blocking the training crawler costs nothing in search. Blocking the search crawler makes you invisible in ChatGPT.
Where do you stand in these answers?
Give us one question your customers ask. We look it up.
A reply within one working day, from the person who measures.