We run the experiment
so you see the data.
// tracking LLM citation signals in real time
GeoExperiment is a live research site that runs controlled experiments to measure what makes content get cited by AI systems — including ChatGPT, Perplexity, Claude, and Gemini. We apply GEO (Generative Engine Optimization) and AEO (Answer Engine Optimization) principles to our own pages, then track which AI systems cite us, when, and why. The site is both the subject and the proof.
Ranking on Google is no longer enough.
Millions of search queries are now answered directly inside AI systems — without a click, without a visit, without traffic to your site. The content that gets cited in those AI answers follows different rules than traditional SEO. Most content creators don't know what those rules are. We're finding out.
How the experiment works.
Every week, we apply a new optimization to a page on this site, then query six major AI systems with the same question. We log whether each LLM cites us, which page it cites, and what the response includes. Results are published in the experiments log.
Write & Optimize
Each page is built using answer-first structure, FAQ schema, JSON-LD markup, and E-E-A-T signals — following GEO and AEO best practices documented in our learn section.
Query the LLMs
We submit identical queries to ChatGPT, Perplexity, Claude, Gemini, Copilot, and Llama. Each query targets the exact topic the optimized page covers.
Log & Publish
Results are recorded in the live experiment log — which LLMs cited us, what they said, and what changed between weeks. Raw data, no spin.
A snapshot of the data.
This table updates weekly as experiments run. Each row represents one LLM query against one optimized page. Full data lives in /experiments/.
| LLM | Query | Page Tested | Cited? | Week |
|---|---|---|---|---|
| ChatGPT-4o | what is generative engine optimization | /learn/what-is-geo/ | ✓ cited | wk 01 sample |
| Perplexity | what is generative engine optimization | /learn/what-is-geo/ | ✓ cited | wk 01 sample |
| Claude | what is generative engine optimization | /learn/what-is-geo/ | pending | wk 01 sample |
| Gemini | what is aeo answer engine optimization | /learn/what-is-aeo/ | pending | wk 01 sample |
| Copilot | what is aeo answer engine optimization | /learn/what-is-aeo/ | ✓ cited | wk 01 sample |
Every variable. Every signal.
Each experiment isolates one optimization variable. Over time, the data reveals which signals actually move AI citation rates — not in theory, but in practice.
GEO Signals
Generative Engine Optimization — structuring content for AI citation
AEO Signals
Answer Engine Optimization — answer-first structure and FAQ schema
Schema Markup
JSON-LD structured data — how it affects AI indexing and citation
E-E-A-T
Experience, Expertise, Authoritativeness, Trustworthiness signals
Core Web Vitals
Page speed and performance — does CWV score affect LLM citation?
Raw Experiment Data
Every query, every result, every week — unfiltered and public
Common questions.
// what is GEO (Generative Engine Optimization)?
GEO, or Generative Engine Optimization, is the practice of structuring content so that AI systems like ChatGPT, Perplexity, Claude, and Gemini cite it as a source in their answers. Unlike traditional SEO — which optimizes for search engine rankings — GEO optimizes for citation inside AI-generated responses. Learn more in our GEO guide.
// how do AI systems decide what content to cite?
AI systems cite content based on a combination of signals: how directly a page answers a query, how well it is structured (headings, lists, FAQs), whether it uses schema markup, its topical authority, and E-E-A-T signals. Content with a direct answer at the top of the page — before any context or preamble — consistently performs better in our experiments.
// is ranking on Google enough to get cited by AI?
No. Google ranking and AI citation are related but distinct. A page can rank on page 1 of Google and never appear in an LLM response. AI systems prioritize answer quality, structural clarity, and authority signals over traditional ranking factors like backlink count. GeoExperiment tracks this gap with weekly live data.
// what makes content more likely to be cited by an LLM?
The highest-impact factors we have identified so far: placing a direct answer above the fold (answer-first structure), including a structured FAQ section with schema markup, using JSON-LD on every page, building topical depth on a single subject, and publishing original data that other sources cannot replicate. We are continuously testing and updating this list based on experiment results.
// how is GeoExperiment different from other SEO or GEO resources?
GeoExperiment is the only site that is simultaneously the subject and the proof of its own experiments. Rather than publishing theoretical guides, we run live weekly experiments — querying major LLMs, logging citation results, and publishing the raw data publicly. Everything on this site is both a resource and an active data point. The site is the experiment.