CITED LABS

AI search and AI visibility research

ABOUT CITED LABS

Cited Labs is Cited AI’s research programme. We publish original tests of how AI engines choose which Australian businesses to name, with the prompts, dates, engines, sample sizes and limitations attached, so other people can check our work and cite it.

Original research

Which AEO agencies does AI recommend in Australia?

October 2026. 138 answers from ChatGPT, Gemini and Perplexity to 18 prompts. 161 agencies named, none in more than 11% of answers.

Case study: a Northern Rivers pet business

What we changed so AI engines can understand and recommend a Northern Rivers pet business.

Choosing an AI search agency: criteria and data

Eight criteria and provider types, built on the study above. Includes disclosure.

Benchmarks and frameworks

AI Visibility Benchmark for Australian brands

How we design a fair benchmark: prompt sets, competitor groups, change logs.

Australian AI Search Readiness Report 2026

The five readiness layers and where Australian businesses commonly stall.

AI Visibility Case Study Framework

How we document results without cherry-picking.

Australian AI Search Prompt Library

Real customer prompts by category for your own testing.

Experiments in progress

These pages set out experiment designs. We publish results only once a test is complete; until then they describe the method, not findings.

The llms.txt experiment

Does an llms.txt file change crawling or citations?

FAQ schema and AI visibility

A controlled test of FAQ markup.

Canonical URLs and AI retrieval

Which URL versions AI engines cite.

Evidence reviews

Who does AI recommend? What the evidence shows

Official documentation and published research across four engines.

What Princeton proved about getting cited by AI

The original GEO study, read carefully.

AI Overviews and disappearing clicks

What the Pew data shows.

How AI search chooses sources

Retrieval, reranking and citation, explained.

We do not publish planned studies as if they were results.
Study	Question	Status
Agency recommendations, re-test	Do the same 18 prompts produce the same agencies over time?	Planned
AI visibility by industry	How often AI names local businesses for mortgage broker, builder, dentist and buyer’s agent prompts in Australian cities	Prompt set in design
Which sources AI cites in Australia	Which domains ChatGPT, Gemini and Perplexity cite most for Australian local-service questions	Prompt set in design
ChatGPT vs Gemini vs Perplexity	How far the shortlists differ for identical questions	Uses the data above as it accumulates

How we run research

Fixed prompts, written down before testing.
Repeated runs on more than one engine.
Dates, locations and engine versions recorded.
Our own results reported alongside everyone else’s, even when they are poor.
Limitations stated on every study.

Full standards: research methodology and editorial policy. Journalists can request underlying data at hello@citedai.com.au.

Find out what AI says about you today

Start with the free check, or book a consultant-reviewed AI Visibility Audit (report within 3–5 business days).

Check my AI visibility Book a free audit