The research arm of Discovered Labs. We run our own studies on the page and authority signals that earn visibility in Google search and in ChatGPT, Claude, Gemini and Google AI Overviews, and publish the method behind every statistic we report.
Meet the researchers
Liam Dunne
Co-Founder
Helped scale Instantly.ai to $20M ARR in three years. Publishes on AI search to 20,000+ marketers.
Ben Moore
Co-Founder & CTO
Ex-Stanford AI researcher. Led engineering at Aurora, built fraud detection at Stripe.
Joell Alex
Data Scientist, AI & ML
Builds LLM pipelines: text-to-SQL agents on LangGraph and Gemini, model observability with Prometheus and Grafana. Ran the 291,607-citation analysis behind the ChatGPT retrieval study.
Kruthika Janvekar
Data Scientist, AI & ML
ML engineer on LLM infrastructure: RAG pipelines, LoRA fine-tuning and FastAPI model serving, deployed on AWS with Docker and Kubernetes.
Cited by
For questions on the method, data requests, or press, email liam@discoveredlabs.com
We tested the LinkedIn study claim that individuals win AI citations: across 5,770 Google page-one pages, Perplexity cited companies more often than individual authors.
How ChatGPT retrieves in 2026: the search plan, the ranked list, live fetches, the Reddit block, review sites, and the vendor-site rule, with 291,607 citations re-measured.
Original research reveals how Reddit powers AI recommendations invisibly, and what this means for brands investing in AI SEO, LLM SEO, and generative engine optimization.
Original research grounded in 2M citation observations and 10K feature-engineered pages. Prompt-content alignment is the dominant lever; AI-perceived domain authority is the structural ceiling; on-page AEO best practices compound meaningfully on top.
A study of AI brand recall and category accuracy across 1,000 B2B SaaS companies, measured in model weights before web search.