Matrix · Quality evaluation
AuthorityEntity & E-E-A-T Signalsmedium effortmedium volatility
The myth
If a page is retrieved, it will get cited.
The receipt says
Platforms run LLM-as-a-Judge rubrics — helpfulness, reliability, cross-source corroboration — and drop retrieved sources that fail before synthesizing an answer.
Impact
7.0
How much this moves citations
Consensus
6.0
How much the industry agrees
Evidence
58%
Strength of underlying research
Engines
4
ChatGPT, Claude, Gemini, Google AIO
Nuance
Corroboration is the actionable part: a claim no other credible source supports is far likelier to be dropped.
Primary source
Platforms apply LLM-as-a-Judge rubrics before citing a source
Independent analysisLumar
Lumar's search quality analysis outlines how Anthropic, OpenAI and Google score external sources on helpfulness, reliability and cross-source corroboration before including them in a synthesized answer.
https://www.lumar.io/blog/best-practice/llm-search-quality-source-selection/ ↗Published 2026-02-01 · Evidence 58%
Applies to
ChatGPTClaudeGeminiGoogle AIO