Research Library
Research Papers

Citation behavior in LLM answer engines

An empirical study of which sources answer engines cite, and why.

The paper measures citation frequency across engines and finds strong biases toward well-structured, dated, and named content.

What the study found

Structured pages with named authors and explicit dates were cited multiple times more often than unstructured equivalents, even at similar ranking positions.

Content-farm patterns — unnamed authors, undated pages, boilerplate paragraphs — were systematically underrepresented in citations.