Bilibili Channel Analyst

This skill enables Claude to perform deep, data-driven analysis of Bilibili channels. It moves beyond simple observation to structured traffic funnel modeling and operational reviews.

Workflow

1. Data Extraction (The Foundation)

When a Bilibili space URL is provided:

• Spawn a browser agent to extract ALL video data.
• Task Prompt: "Go to [URL]/video. Extract all videos across all pages. For each video, get: Title, Views (播放), Comments (评论), Date, and Duration. Save as a JSON array of objects: [{"title": "...", "views": "...", "comments": "...", "date": "...", "duration": "..."}] in channel_data.json."
• Note: If the channel has many pages (>10), suggest using the Bilibili API https://api.bilibili.com/x/space/arc/search?mid={mid}&ps=50&pn={pn} for faster extraction.

2. Advanced Multi-dimensional Analysis

Run the bundled script scripts/analyze_bilibili.py on the extracted JSON data to generate:

• Topic Clustering: Categorize videos based on keywords in titles.
• Performance Correlation: Average views/comments per category, and correlation between video length and performance.
• Traffic Funnel Metrics: Visibility (Exposure proxies) -> Click (Views) -> Interaction (Comments).
• Keyword Weight Analysis: Identify which words in titles drive the most views.

3. Reporting Structure

Produce a comprehensive report using the following three sections:

A. Traffic Funnel Analysis (流量漏斗分析)

• Exposure: Analyze reach and visibility.
• Conversion: Click-through rate (CTR) proxies.
• Engagement: Interaction rate (Comments/Views). Identify bottlenecks (e.g., "High views but low interaction").

B. Operational Review (运营复盘)

• Wins: Identify top-performing categories and "hot-topic hijacking" successes.
• Losses: Identify underperforming content (e.g., conceptual vs. practical content) and traffic cannibalization issues (e.g., high-frequency posting).
• Diagnostics: Why certain videos failed (e.g., "Anxious Developer" persona mismatch).

C. Actionable Optimization Plan (整改建议)

• Title Templates: Provide specific, data-verified formulas (e.g., [Model] + [Hardware] + Data实测 + Result).
• Thumbnail Strategy: Visualization of performance metrics (Tokens/s, GPU usage).
• Content Roadmap: Shift from "Broad News" to "Deep Tutorials" or "Practical Case Studies."
• Engagement Hacks: CTA (Call to Action) strategies for the first 60 seconds of video.

Output Format

Always provide a concise summary in chat, followed by the full Markdown report as a file.

Created by Ecommerce Mind (TL) for Bilibili Content Strategy.