What problem does it solve? Creators on Xiaohongshu (RED) often export backend analytics or screenshot their dashboard but cannot tell why a note underperformed or which content direction to double down on. This Skill turns raw exposure, click, engagement, and follower numbers into a clear diagnosis of which funnel layer is broken and what to change next. ## Core Features & Use Cases - Six-layer funnel diagnosis: Walks exposure, CTR, completion, engagement, follower conversion, and conversion layers in order, stopping at the first broken layer with concrete fix suggestions. - Multi-note pattern analysis: Uses medians, quantiles, outlier inspection, and grouped comparisons (topic type, title style, cover format) to surface repeatable patterns while enforcing sample-size honesty rules. - CSV/Excel processing script: The bundled xhs_notes.py script probes column names, maps Chinese column aliases, normalizes units (percent, 万), computes all rate metrics, and runs data-quality checks without loading full tables into context. - Use Case: A creator exports 20 notes of backend data and asks why growth stalled; the Skill computes CTR and engagement medians, finds that tutorial-style notes have 3x the save rate of review-style notes, and recommends one verifiable change for the next post. ## Quick Start Ask the AI to analyze your Xiaohongshu note data by pasting backend numbers, uploading a screenshot, or attaching an exported CSV/Excel file and requesting a funnel diagnosis.
