[인턴] 2차 예고편
Test the pattern with caution
Why this decision is credible
Related public videos used as the local benchmark set.
Views divided by public publish age. Useful directionally, not private Studio data.
Likes and comments benchmarked against the related sample. Current engagement: 0.86%.
This video views compared with the average related video. Subscribers are not required.
Why this video will spread or stall
Built from public video data and content signals. Private retention, revenue, and real traffic-source data require YouTube Studio access, so those sections are clearly modeled as analyst inference.
Do not copy directly yet
Decision first: use the metrics below as evidence, but start from whether this idea is worth your next upload.
The content does not yet give viewers or the algorithm a strong enough reason to click, stay, and share.
- Mobile-friendly title length
- Uses specificity through numbers
- Viewer payoff could be more explicit
- Description is thin for search support
Public-data inference, not private YouTube Studio retention.
Hook is the likely biggest leak: 32% more viewers may leave around 0:00-0:15.
Show the payoff, conflict, or final result in the first shot before explaining background.
The title/opening promise may be too broad, so cold viewers can leave before context lands.
Public metadata suggests the viewer job is reasonably clear after the hook.
The first content beat needs escalation; flat pacing here usually creates the largest audience leak.
Comment intensity hints that enough viewers cared to react beyond passive watching.
If the payoff is not explicit, late retention likely depends on personality rather than structure.
The ending may be watched by a small core; generic CTA will not recover earlier losses.
Turn the diagnosis into a reusable workflow
Save the brief, monitor the video/topic, and create an alert so this analysis becomes an ongoing research asset.
🎯 What To Do Next
Packaging strategy
High impactTest a tighter title with one concrete outcome, a number, or a stronger curiosity phrase. This is the fastest improvement lever for underperforming packaging.
Audience response
Quick winEngagement is only 0.86%. Add a clearer opinion, stronger hook, or a more specific viewer prompt in the first 30 seconds.
Distribution timing
High impactThis upload time was not ideal. Re-test the same style closer to Tuesday 12:00 AM windows that better match viewer activity.
Search and description support
Quick winImprove description depth with clearer summary text, timestamps, or related links so the video has more search and session-supporting context.
📈 Performance Trends
Views Trend
Projected growth based on current velocity
Likes Trend
Like accumulation pattern
Engagement Trend
Combined engagement metrics
⚡ VELOCITY METRIC
views/day📈 ENGAGEMENT BREAKDOWN
percentile: 56%✍️ TITLE ANALYSIS
Score: 70/100Read this as the explanation layer. The action plan above tells you what to change first; this section explains why the video likely behaved this way.
💡 Replication Checklist
Save the title and thumbnail pattern before assuming the topic was the only reason it worked.
Reuse the format only if the promise also fits your niche, tone, and viewer expectation.
Change one thing first — title, topic angle, or thumbnail — so you know what actually moved performance.
Use 3-5 comparable winners before turning one successful video into a content rule.
👥 Engagement Intelligence
AI Recommendations
📊 Advanced Video Analytics
Content Rhythm
Dropoff Trend
Modeled Comparison
First 2 minutes show lower dropoff than category average — strong opening hook. Consider adding engagement elements at 50-60% mark to boost completion.
Modeled completion rate of 35.0% is performing well. To improve further, consider adding end-screen CTAs or suspense in the final 20%.
Traffic Sources
Estimated 62.5% mobile-weighted viewing. Optimize for vertical readability and mobile thumbnails.
Desktop and TV combined are modeled at 21.5%. Longer content may perform well here.
Mobile is the top modeled source at 62.5%. Treat this as a planning proxy.
