// who runs geoexperiment?

GeoExperiment is run by the GeoExperiment Team and CC — an independent research team with no affiliation to Google, OpenAI, Anthropic, or any AI company. We build and optimize our own site using GEO, AEO, and SEO principles, then query six major AI systems weekly to measure what actually gets cited. Everything we find gets published publicly. The site is both the subject and the proof.

What GeoExperiment is.

GeoExperiment is a live research site with one focus: measuring what makes content get cited by AI systems. Not in theory. Not based on what platform documentation says. Based on weekly controlled experiments with logged, published results.

We apply GEO (Generative Engine Optimization) and AEO (Answer Engine Optimization) principles to our own pages, submit identical queries to six major LLMs, and record exactly which pages get cited, in which systems, and on which week. Every optimization we make is a variable. Every result is a data point.

The site is built this way deliberately. We are not observers writing about someone else's experiment. We are the experiment.

6
LLMs queried every week
5
pages under active experiment
3
optimization layers per page (SEO + AEO + GEO)
100%
of findings published publicly

What GeoExperiment is not.

We are not affiliated with Google, OpenAI, Anthropic, Microsoft, or Meta. Our findings represent our own independent experiment results — not platform endorsements or official guidance.

We do not publish SEO "tips" or AI optimization "hacks." Every claim on this site is either backed by our experiment data or clearly labeled as a working hypothesis pending testing. If we don't have data on it yet, we say so.

How we run the experiment.

// 01
One optimization at a time. Each week we change one variable — a schema type, a heading structure, an answer block position — and log the effect on citation rate. Changing multiple variables simultaneously produces unreadable data.
// 02
Identical queries across all LLMs. Every LLM is queried with the exact same prompt on the same day. This controls for query variation and isolates platform-specific citation behavior.
// 03
Raw results, no spin. We publish citation results whether or not they confirm our hypothesis. An optimization that fails to move citation rate is as valuable as one that does — it eliminates a variable.
// 04
The site is the subject. Every page on geoexperiment.com is built to the GEO + AEO + SEO standard we are testing. We do not test on third-party sites. We are the guinea pig.
// 05
Public data, always. Every experiment result is published in the live experiment log. There is no private dataset, no premium tier, no results held back. The experiment is only credible if it is fully transparent.

What the site is built with.

Every technical decision on this site is itself an experiment variable. The stack is documented here for transparency and reproducibility.

Layer Choice Reason
// hosting Cloudflare Pages Edge CDN, fast TTFB globally, free tier, integrates with registrar
// html Vanilla HTML/CSS No JS framework = no render-blocking = better CWV scores and AI crawlability
// fonts Inter + JetBrains Mono Inter for body readability, JetBrains Mono for lab/data aesthetic
// schema JSON-LD on every page Required for both Google AI Mode and LLM citation trust signals
// images WebP format Smaller file size, better LCP scores, modern browser support
// analytics Google Search Console Indexing status, query data, CWV field data — all in one place
// who runs GeoExperiment?

GeoExperiment is run by the GeoExperiment Team and CC — an independent research team with no affiliation to any AI company or search engine. We build, optimize, and document the site in real time as an active experiment into what makes content get cited by AI systems.

// what is the purpose of GeoExperiment?

To answer one question with real data: what makes content get cited by AI systems? Rather than publishing theoretical guides, we apply GEO and AEO principles to our own pages, query six major LLMs weekly, and publish the raw results. The site is both the subject and the proof of its own experiments.

// which AI systems does GeoExperiment track?

We actively query ChatGPT (GPT-4o), Perplexity, Claude (Anthropic), Gemini (Google), Copilot (Microsoft), and Llama (Meta). Each LLM is queried with identical prompts on the same day each week. Results are logged and published in the experiment log.

// is GeoExperiment affiliated with Google, OpenAI, or Anthropic?

No. GeoExperiment is fully independent. We have no affiliation with Google, OpenAI, Anthropic, Microsoft, or Meta. Our findings are based entirely on our own experiments and publicly available information about how these systems work.

// can I use GeoExperiment data in my own work?

Yes. All experiment data published on this site is public. If you reference our findings, please cite geoexperiment.com as the source and link to the specific experiment entry. We track citations — including by humans — as part of the ongoing authority-building experiment.