What is an LLM citation audit?
It is a structured review of the URLs and domains cited in AI-generated answers, the claims those sources support, which brands benefit, and what evidence gaps or inaccuracies require action.
Free evidence worksheet
LLM citation audit template
Map the exact pages shaping AI answers, the claims they support, the brands they benefit, and the evidence gaps worth fixing.
Published:
Audit pages, not domain counts
A citation matters because of what it supports
Start from stored answers, not a backlink export. For every visible citation, record the exact URL, its role in the answer, and whether the page actually supports the generated claim.
Aggregate only after reviewing the evidence. A domain-level total can hide that one outdated page is repeatedly shaping inaccurate comparisons or that one independent guide drives several prompts.
Audit fields
Preserve the source-to-claim connection
01
Prompt and provider
Preserves the question and answer surface that produced the citation.
02
Cited URL and domain
Identifies the exact page and source owner instead of counting a domain without context.
03
Source class
Separates owned, independent editorial, directory, community, competitor, and unknown sources.
04
Claim supported
Records the statement or decision the source appears to support in the answer.
05
Brand and competitor effect
Shows whether the citation supports your brand, a competitor, both, or neither.
06
Accuracy and freshness
Flags outdated, contradictory, inaccessible, or weakly supported source material.
07
Action and owner
Turns a source gap into a specific correction, publication, outreach, or monitoring task.
Common questions
Use the resource without overstating the result
Should all cited domains be treated equally?
No. Record the exact page, source type, supported claim, freshness, independence, and relevance. A generic directory mention and a detailed independent evaluation are different evidence.
What if an answer mentions the brand without citing it?
Record the mention in the answer-level tracker and leave the claimed-domain citation field empty. Mentions and citations are separate signals and should not be substituted for one another.
Does earning a link guarantee an AI citation?
No. The template documents observed citations and source opportunities; it does not imply that publishing or earning a link will cause a future model to retrieve or cite it.
Evidence before conclusions
See a complete benchmark before running your own
Review the questions, stored answers, citations, competitors, coverage, and five prioritized actions in the sample report.