Techniques and Signals
Techniques and signals are what a publisher does, or is told to do, to be cited by an AI answer. Seventeen pages cover the ones being sold hardest, and each carries a verdict on the evidence rather than on the consensus.
The pattern that repeats across this category is a real mechanism attached to an unearned promise. Retrieval genuinely works on passages rather than whole documents, so content chunking is a real thing — but no engine documents a chunk-shaped ranking factor. Structured data genuinely helps a parser, and no operator has said schema markup influences whether an answer cites you. Separating the mechanism from the promise is most of the work on these pages.
The second pattern is measurement sold as visibility. A share of voice figure needs a denominator, and none of the surfaces publish one; an AI Overview rank has no position to rank in; attribution depends on referrer headers that several products do not send. Where a number cannot mean what it is sold as meaning, the page says so plainly.
None of that makes the category worthless. Crawler access is real and checkable, answer-shaped writing is a legitimate craft with a long history in featured snippets, and entity grounding has decades of retrieval research behind it. The verdict column separates the three cases.
| Page | Evidence | What it is |
|---|---|---|
| Generative Engine Optimization | Unsupported | What GEO claims and what the evidence supports |
| Answer Engine Optimization | Unsupported | Whether AEO is a discipline or a label |
| GEO vs AEO vs SEO | Unsupported | How the three terms relate, overlap and differ |
| AI Citations and Source Selection | Observed | How AI answers pick and credit their sources |
| Brand Mentions in AI Search | Unsupported | Whether unlinked mentions move AI answers |
| Schema Markup for AI Search | Unsupported | What structured data does and does not do |
| FAQ Markup and Answer Extraction | Unsupported | The markup is dead; the writing pattern is not |
| Content Chunking and Passage Retrieval | Observed | Retrieval matches spans of text, not URLs |
| Entity Grounding and the Knowledge Graph | Unsupported | Whether being a known entity gets you cited |
| E-E-A-T Signals in AI Search | Unsupported | A rater framework, carried into a silence |
| AI Crawler Directives and llms.txt | Unsupported | Which tokens matter, and which file nobody reads |
| Retrieval-Augmented Generation | Documented | The mechanism behind every cited AI answer |
| Query Fan-Out | Documented | One question, many searches, one answer |
| AI Visibility Tracking | Observed | Measuring whether AI answers cite or name you |
| AI Overview Rank Tracking | Unsupported | Tracking presence and citation inside AI Overviews |
| Share of Voice in AI Search | Unsupported | A borrowed fraction with no real denominator |
| AI Search Attribution and Analytics | Observed | Proving an AI answer sent the visit |