RAG Retrieval Quality Analyzer

Analyze the retrieval quality of our RAG system and suggest improvements.

## Sample Queries and Results
{{query_result_pairs}}

## Current Retrieval Configuration
- Embedding model: {{embedding_model}}
- Top-K: {{top_k}}
- Similarity threshold: {{similarity_threshold}}

## User Feedback
{{user_feedback_summary}}

Perform a structured analysis:

1. **Relevance Assessment**: Rate each result set (1-5)
2. **Failure Pattern Analysis**: Identify common retrieval failures
3. **Root Cause Diagnosis**: Why are failures occurring?
4. **Improvement Hypotheses**: Ranked by expected impact
5. **A/B Test Proposals**: Experiments to validate improvements

Provide specific, actionable recommendations with expected lift.