Category
LLM Structure
The machine-readable architecture that makes a website legible to AI systems.
Start with Feeds vs. Structure: How LLMs Actually Read Your Website
Explore LLM StructureApplied LLM Research
How AI systems discover, cite, and recommend websites.
I run controlled experiments and publish what holds up in practice.
This site is organized around three foundational categories: LLM Structure, LLM Visibility, and LLM Discoverability.
Framework
Start with the foundations.
Each category defines a different part of how websites become usable inside AI systems. Start with the foundational essay, then explore the full category.
Category
LLM Visibility
The conditions that let AI systems access, parse, understand, and verify a website.
Start with What Is LLM Visibility?
Explore LLM VisibilityCategory
LLM Discoverability
The point where a model decides a website is the right destination for a user ready to act.
Start with What Is LLM Discoverability?
Explore LLM DiscoverabilityEvidence
Published findings from live experiments.
The work here is not speculative. It is grounded in observed model behavior, controlled tests, and published findings.
Research
AI Citation Research: Key Findings from WhatsMyArtWorth.com
A 28-day experiment proved a brand-new site can earn ChatGPT citations in 33 days. Six findings on AI conversion rates, page weight, schema data, and the completion gap.
Read the articleWhat The Research Shows
- New websites can begin earning citations faster than most teams assume when their structure is usable by models.
- Visibility and discoverability are separate conditions. A model can understand a site without recommending it.
- Machine-readable structure affects whether models can parse, trust, and route users to the right destination.
The point of the research is to move past theory and document what changes model behavior in practice.
Method
A practical research method.
The work follows a simple pattern: define the category, test live behavior, and publish what survives contact with the real world.
Controlled experiments
Tests are designed around concrete variables so changes in citations, routing, and recommendation behavior can be observed.
Observed model behavior
The focus is on what models actually do with websites, not just what platform documentation suggests they might do.
Published findings
Results become categories, essays, and frameworks that other people can use to understand what is changing.
Archive
Browse the full writing archive.
Explore essays, category foundations, and research notes in one place.