- Home
- Skills
- Research & Analysis
- YouTube Insights
More screenshots
Works with the AI Agents you already use
YouTube Insights
It extracts real performance metrics directly from your logged-in session to build a data-backed growth strategy.
$9.99
YouTube Insights
Example session with this skill installed
Analyze my YouTube Shorts and tell me what's working: which titles, themes and lengths get the most views and subscribers, and how they compare with my Instagram.
- Read your context and instructions
- Compiled the youtube insights
- Generated the document
Summary
I read YouTube Studio's Advanced mode table and Content list (read-only), tagged each Short and ran the analysis, matching each video with the Instagram run. Example below: a fictional sample channel, @sample_history, 18 Shorts from Oct 2 to Nov 9.
All 5 hits have a question as the title. Question titles: median 2,545 views vs 605 for a bare name.
| Metric | Value |
|---|---|
| Views | 28,800 (median 1,050 per video) |
| Watch time | 369 h (44 s per view) |
| Stayed to watch (didn't swipe) | 52% median |
| % viewed | 66% median |
| Hits (≥2× median) | 5, together 59% of views |
| Subscribers gained | 70 |
| Traffic | Shorts feed 86%, YouTube search 9% |
What the numbers show
| Finding | Evidence | Confidence |
|---|---|---|
| Question titles perform best | median 2,545 views (8 videos) vs 605 for “Name only” (4); 5/5 hits are questions | medium |
| The Shorts feed decides | 86% of views come from it | strong |
| Staying beats everything | stayed to watch: 63% on hits vs 49% on the rest; correlation with views +0.99 | strong |
| Longer Shorts lose share | length × % viewed correlation −0.77 | strong |
| Same video, different fate | 13 videos on both platforms; YouTube × Instagram views correlation +0.48. Vikings: 3,150 on YouTube vs 760 on Instagram | strong |
The one flop (under 25% of the median) is left out of group comparisons because it never got distributed.
What to change
Title it as the question the video answers. Turn “Name: explanation” and bare names into the question a curious viewer would type.
2. Win the first second. The feed only pushes Shorts people don't swipe away.
3. Stay near 60 s. Longer Shorts lose a bigger share of the video.
4. Use YouTube search. 9% of views already come from search; questions match what people type.
Next videos
- Why did the Hindenburg burn in 1937? (disasters + question title)
- Did medieval people really never bathe? (myths as questions: the Vikings Short got 3,150 views)
Test for the coming weeks
| Test | How | What to measure |
|---|---|---|
| Same Short, two titles | Re-title a “Name only” Short as a question after 14 days | Views and search share in the next 7 days |
Deliverables
report.html: KPIs, views per video, median views by theme, title format, era, length, weekday and uploads per day, stayed-to-watch and % viewed scatter plots, YouTube × Instagram table, correlations and a sortable table of every videoanalysis.jsonto compare with the next run (--previous)
Next steps
- Review
tags.json(theme, era, title format) and re-run if anything is off. - Re-title two “Name only” Shorts as questions.
- Run the analysis again in 2 weeks.
report.html
PDF · document
Example file from a real run - the skill writes it into your workspace.
Connects securely to your tools. The creator never sees your data.
About this skill
Stop guessing why your videos flop. This skill audits your YouTube Studio data to pinpoint exactly which titles, hooks, and formats drive views versus what causes viewers to swipe away. It extracts real performance metrics directly from your logged-in session to build a data-backed growth strategy. ## What it does - **Metric extraction** pulls views, watch percentage, subscriber net change, and CTR directly from YouTube Studio. - **Content tagging** categorizes videos by theme, era, hook type, and title format to identify patterns. - **Cross-platform mapping** syncs YouTube performance with Instagram data to see which content travels best. - **Pattern detection** identifies "outliers" (2x median views) and "flops" (under 25% median) to isolate success variables. - **Report generation** builds an interactive Artifact dashboard with charts and specific next-step recommendations. ## How it works 1. **Collect** data by navigating YouTube Studio in Chrome to scrape Advanced Mode tables and Content lists. 2. **Normalize** raw JSON and text files into a structured `videos.json` using Python scripts. 3. **Tag** content to assign dimensions like `theme` or `hook_type`, allowing for categorical analysis. 4. **Analyze** the dataset to calculate medians, identify performance tiers, and compare against previous runs. 5. **Render** a standalone HTML report and publish it as an interactive Artifact for review. ## Frameworks & tools Python 3.12+ (standard library only), YouTube Studio (web), Claude in Chrome (optional) and small read-only JavaScript snippets that read the Studio tables. ## Why this beats prompting it yourself Manual prompts cannot access the deep "Advanced Mode" tabular data or maintain long-term theme consistency across dozens of videos. This skill ensures your analysis is grounded in multi-run historical data rather than surface-level observations. ## Use cases - Identify why specific Shorts have high "swiped away" rates. - Determine which title formats (e.g., questions vs. statements) yield higher CTR for your niche. - Compare performance of the same video file across YouTube and Instagram. - Generate data-backed ideas for your next content batch based on high-performing themes. ## Known limitations Requires a logged-in Chrome session via `chrome-browser`. Large channels with 50+ videos may require manual CSV exports for full accuracy.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Research & Analysis.
Automatic collection needs an agent with browser access and a logged-in YouTube Studio session; otherwise download the Advanced mode CSV. The Studio Content list shows dates without the time, so posting hours are not compared. With few videos per group, findings are hints to test. Reports are in Brazilian Portuguese.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 6 days ago
- Software
Python 3.12+
Git
- Optional
- Google Chrome