What the Data Shows
Living research pages built from patents, API leaks, court testimony, and large-scale citation studies. I update each page as new evidence emerges. Each one reports the accumulated weight of the data.
How AI Search Decides What to Cite, and What It Ignores
A synthesis of AI citation mechanics across ChatGPT, Perplexity, Google AI Mode, and Gemini. What the systems cite, why the traditional SEO metrics predict so little, and how the retrieval pipeline works.
15+ studies across Ahrefs, Semrush, DEJAN AI, Profound, AirOps, and others
The Reality Gap: What Google Says vs. What Google Measures
Google's public guidance and its internal ranking documentation disagree. Each public statement sits next to the API module and the sworn testimony that contradict it.
Google API leak (14,014 attributes), DOJ antitrust testimony, spokesperson statements
The Zero-Click Paradox: Fewer Clicks, Higher Conversion
Traffic volume is collapsing but traffic value is concentrating. Why the traditional CTR curve is dead and why the clicks that survive the AI filter convert at 5x the historical rate.
SparkToro, Semrush, Ahrefs, Seer Interactive, Superprompt, Pew Research Center
Content Engineering: Building the Agent-Shaped Web
The shift from writing SEO content to building extractable data structures. How to architect pages for LLM ingestion, query fan-out, and agentic retrieval.
DEJAN AI, iPullRank, Semrush, Wellows, Growth Marshal, Seer Interactive, Surfer SEO
Topical Authority: From SEO Folklore to Confirmed Signal
The 2024 API leak confirmed topical authority as a real, multi-signal system: siteFocusScore, siteRadius, topic embeddings, NsrChunks, and ClusterUplift. What the system measures, how it affects AI citation, and how to act on it.
Google API leak (14,014 attributes), Semrush, Surfer SEO, Graphite, iPullRank, Digital Bloom
Programmatic SEO Architecture: When Data Is the Product
What separates a programmatic build that ranks from one that Google suppresses. Data granularity, template architecture, Google's detection systems, rendering strategy, and why 96.55% of all indexed pages get zero traffic.
Google patents, API leak (14,014 attributes), Wise, Zillow, NerdWallet, G2, Ahrefs, Surfer SEO
Content Decay: The 31% Refresh Threshold and the Dual Decay Curve
Content visibility half-life has collapsed to 3-6 months. A controlled 14,987-URL study shows only major content expansions (31%+ change) produce ranking gains. AI citations turn over 70% in 2-3 months.
RepublishAI (14,987 URLs), Seer Interactive, Authoritas, HubSpot, Google API leak, patent US8549014B2
In-Page Information Architecture: Structuring Content for Three Audiences
Site architecture tells search engines how pages relate. In-page architecture determines how humans scan, how crawlers parse, and how AI systems extract. The patterns that serve all three overlap more than practitioners realize.
NN/g, Baymard Institute, DEJAN AI, Shashko (42,971 citations), iPullRank, WebAIM, Google API leak
Unhelpful Content: The Patterns That Trigger Suppression
Google does not measure helpfulness. It detects unhelpfulness across 13 independent classifiers. Mapped from the 2024 API leak, patents, and HCU recovery data to the ranking pipeline stages where each pattern fires.
Google API leak (14,014 attributes), Panda/N-gram/NavBoost patents, SpamBrain taxonomy, Lily Ray HCU analysis
Living Documents
These pages are continuously updated as new studies, patent filings, and platform changes emerge. Every claim carries a source. The certainty language reflects the weight of the evidence, and not my conviction.