VisibilityOS
notableReviewed Sat Oct 31

Controlled SIGIR study isolates competitive first-citation selection across six LLMs

Vishwakarma, Kumar and Jamidar report 252,000 controlled paired-source trials across six LLMs and 18 content factors. In a two-source simulated retrieval context, topical relevance and source order most strongly affected which source received the first citation.

Context

Vishwakarma, Kumar and Jamidar report 252,000 controlled paired-source trials across six LLMs and 18 content factors. In a two-source simulated retrieval context, topical relevance and source order most strongly affected which source received the first citation.

What changed

The study supplies controlled evidence on first-citation preference among already provided documents, separating this stage from organic web retrieval, indexing and ranking.

Why it matters

AI visibility analysis should distinguish retrieval and candidate ordering from subsequent citation selection; content relevance, price detail and recency can matter once a source is in the candidate context.

Visibility OS interpretation

Visibility OS interprets the result as evidence for stage-specific citation-selection measurement, not proof of a general live-search ranking factor. Do not convert experimental odds ratios into predicted real-world traffic.

What remains uncertain

Will effects replicate in production engines with more than two sources, live retrieval, familiar brands and changing model versions? How much does first-citation preference predict answer absorption or referrals?

Evidence claims

Strong Evidencesupports

In a controlled two-document RAG testbed with 252,000 trials across six LLMs and 18 manipulated factors, Vishwakarma et al. found that topical relevance and candidate list position dominated first-citation selection; this does not test live web retrieval.

Controlled SIGIR study isolates competitive first-citation selection across six LLMs | Visibility OS