Subjective Impression (G-Eval)
The AI-evaluated quality score that measures how helpfully and prominently your brand was presented in an AI answer.
Plain Language Definition
Subjective Impression, measured by the G-Eval methodology, is an AI-calculated grade that scores how prominently, clearly, and helpfully your website was referenced in an AI-generated answer. It uses a language model to evaluate another language model’s output — scoring dimensions like relevance, logical flow, and diversity of coverage. It is the qualitative complement to PAWC’s quantitative measurement.
Technical Definition
LLM-evaluated quality score applied to AI-generated content passages, assessing dimensions including query relevance, brand prominence, logical flow contribution, factual grounding, and citation diversity — generating a composite quality index via structured multi-criteria prompting of a judge model.
Why This Matters for AI Search Visibility
G-Eval scores capture aspects of citation quality that word count cannot — whether the AI presented your brand favorably, logically, and helpfully versus awkwardly or minimally. High PAWC with low G-Eval suggests mechanical inclusion without genuine recommendation quality.