Research & whitepapers
The paper library.
How AI models choose what to recommend, written down: frameworks you can grade any vendor with, the Index methodology in full, and data papers from the frozen monthly editions. All free, no card.
The Coverage Illusion
Research study No. 1: track one AI model and the number one you see is wrong in 61.4% of categories, measured from the frozen August 2026 edition.
Read the paper
How AI Actually Chooses What to Recommend
The mechanism behind every AI recommendation: two memories, chunk-level retrieval, and the answer-versus-citation split.
Read the paper
The AI Visibility Audit: The Complete Framework
The published six-layer framework for auditing AI visibility. Use it to grade any vendor, including us.
Read the paper
The Crawler Question
Which AI bots visit your site, the five jobs they do, and whether you should let them in.
Read the paper
Is Your GEO Tool Measuring Anything Real?
Seven questions to ask any GEO tool before you pay for it, with a scorecard. We volunteer to be graded.
Read the paper
How CiteHawk Measures AI Recommendations
Methodology v1.1: the methodology paper as published with the July 2026 edition, covering how the AI Index collects, scores, and freezes its monthly records. The live methodology page carries the current version.
Read the paper
The AIs Don't Agree
Three pages on where the eight AI models split, from the frozen July 2026 edition of the AI Index.
Read the paper
Who AI Recommends: Category Spotlight
A two-page spotlight on who AI recommends in a category, generated from the public Index edition. Sample: digital marketing agencies.
Read the paper
Vertical Intelligence Reports
Deep reports on what AI recommends in 11 service verticals, from law firms to roofing companies. July 2026 edition.
Read the paper
The AI Recommendation Report
The flagship report on what AI models recommend and why, built from the frozen August 2026 edition.
Read the paper