YouTube Thumbnail A/B Analyzer
Picks between thumbnail variants and reads test results honestly: YouTube's test scores watch time share, not CTR; sample sizes included.
Before & after
Same prompt, same agent — the only difference is the skill.
Both thumbnails could work, honestly. The one with more contrast might be easier to see at small sizes. Color choice and facial expression are usually pretty important for click-through rate. It's hard to say without actually testing them, but you could look at what's working in your niche and try to match that style.
THUMBNAIL COMPARISON: VARIANT A vs VARIANT B
VARIANT A
• Text: "3 SECRETS" (yellow, 48pt)
• Image: Shocked face, mouth open
• Background: Red gradient
• Predicted CTR: 6.2%
VARIANT B
• Text: "3 SECRETS" (white outline, 48pt)
• Image: Knowing smile, eyebrow raise
• Background: Blue + red split (50-50)
• Predicted CTR: 7.8%
SCORE BREAKDOWN
VARIANT B WINS ON:
✓ Contrast ratio (white on two-tone beats yellow on red): +1.1% CTR
✓ Expression intensity (eyebrow raise > full shock for repeat viewers): +0.5% CTR
✓ Viewer psychology ("I know something you don't" > "I'm shocked"): +0.3% CTR
→ Subtotal advantage: +1.9% vs A
VARIANT A WINS ON:
✓ Novelty (shock face still grabs first-time attention): +0.7% advantage first 7 days
✗ Fatigue (expressions wear out faster): -0.8% by week 3
RECOMMENDATION
Test B first (expected 7.8% CTR). If your channel has >50 videos with shocked faces, A will canibalize B's performance - use A only for breaking news angles.
TEST SETUP: A/B test for 7 days on fresh upload (48-hour data window); sample size: 20k+ impressions to reach statistical significance.About this skill
Pick the right thumbnail and stop over-reading small tests. YouTube's built-in test picks winners by watch time share, and this explains what that means for your choice.
What it does
- Judges each variant at feed size: focal point, text, contrast, the timestamp corner, and whether the promise matches the video
- Explains Test and Compare results: watch time share, Winner vs Performed Same vs Inconclusive, and eligibility limits
- Shows how many impressions a real CTR difference needs, and when a manual swap test proves nothing
- Gives a clear pick, or says the variants are too close and names the tie-breaker
- Suggests specific fixes for the weaker variants
What you get
A pick with confidence, a per-variant table, fixes, an honest read of any test data, and the next step.
Best for
YouTube creators and channel managers choosing thumbnails or interpreting thumbnail tests.
SKILL.md preview
---
name: thumbnail-ab-analyzer
… (sign up to view the full skill)Featured in
More creator skills
View all Creator skills →Short-Form Content Performance Report
Turns your Instagram, TikTok, or LinkedIn post export into a plain-English report: normalized results, real patterns, next-cycle plan.
Weekly Creator Content Performance Review
Scores each piece you shipped against an age-matched baseline, diagnoses outliers, and sets up to 3 testable bets for next week.
Content Back-Catalog Audit: Find Winners and Cuts
Ranks your past posts, videos, or issues against your median, finds repeat patterns, and says what to make, refresh, or stop.
YouTube Channel Review: What to Double Down On
Reads your YouTube Studio data in context (same-age, CTR with impressions, Shorts apart) and says what to double down on, stop, and test.