Test Reader
Your thumbnail test numbers, turned into a decision
Thumbnail A/B Test Analyzer
Write your briefSample result, from a real run
Made from this brief: YouTube A/B test (Test & Compare), video 'The quietest NAS I've ever built', ran 11 days, 3 thumbnail variants. Variant A (control): 68,410 impressions, 4,120 clicks, 6.02% CTR, average view duration…
The Numbers
| Variant | Impressions | Clicks | CTR |
|---|---|---|---|
| A | 68,410 | 4,120 | 6.02% |
| B | 67,980 | 4,690 | 6.90% |
| C | 68,150 | 3,540 | 5.19% |
Verdict: Real
- Winner: B at 6.90%
- Gap: 0.88 percentage points (14.6% relative)
- z = 4.29 against a 1.96 threshold at 95% confidence
- Sample floor: cleared
The test shows a real difference between the variants. Variant B, with a face and an arrow, performed better than the control, Variant A, despite the channel's historical hostility to face-and-arrow thumbnails.
What would settle it
Nothing - this one is settled.
Benchmark
6.90% CTR is a strong performance, meaning the thumbnail is effective at enticing viewers to click, especially considering the channel's baseline CTR of 5.1%. This suggests that the thumbnail style used in Variant B could be a good direction for future videos.
What each thumbnail was doing
- Variant A (control) - Object / Result, leaning on social proof implicitly through the display of a finished project. The thumbnail shows the finished white NAS cube on a wooden desk, shot in daylight, with the text '28 dB' in the corner in brass on slate.
- Variant B - Face-Driven, leaning on curiosity gap and relatability through the use of a face and an arrow pointing at a key metric (the sound meter reading). The text 'QUIETER THAN YOUR FRIDGE' adds a comparative element to pique interest.
- Variant C - Object / Result, leaning on before/after contrast by showing the old noisy Dell tower on the left and the new white cube on the right, with a vertical divider.
Why Variant B pulled ahead
Variant B likely pulled ahead because it effectively used a face to convey a personal, relatable experience, combined with an arrow that directly points to a key metric (the sound meter reading), making the thumbnail more engaging and informative at a glance. The text 'QUIETER THAN YOUR FRIDGE' adds a layer of curiosity by making a comparative claim that viewers might find interesting or hard to believe, encouraging them to click to learn more. Despite the channel's historical hostility to face-and-arrow thumbnails, this specific execution seems to have resonated with the audience.
What hurt Variant C
Test 1 - Safe: Simplified Face-Driven
- Hypothesis: A simplified version of Variant B's face-and-arrow approach will retain its effectiveness while improving legibility and reducing visual noise.
- Archetype: Face-Driven
- The one change: Remove the arrow and the sound meter reading, focusing on the face and a simplified textual claim ("QUIETEST NAS").
- Everything else held: Background, lighting, and overall composition remain the same as Variant B.
- Overlay copy: "QUIETEST NAS"
- Watch: CTR and average view duration.
- Brief to paste into AI Thumbnail Factory:
A close-up of my face looking at the NAS, with a slight smile and a hint of surprise, on a clean and neutral background. The text "QUIETEST NAS" is overlayed in a simple, bold font, either in white or a light color that contrasts with the background.
Test 2 - Bold: Emphasizing the Benefit
How it works
Add your impressions and clicks per variant
One line per thumbnail. Paste it straight out of YouTube Studio: "A: 2847 impressions, 312 clicks". CTR percentages alone are not enough: we need the raw counts to tell a real difference from noise.
We draft it
The app works through its steps and hands back:
- Test Readout
- Diagnosis
- Next Tests
Refine and export
Change the brief and run it again, or open the full run to read, copy and download each file.
Questions
What is a credit?
The unit a run is priced in. Each app has a fixed price in credits, shown on its Generate button, and a run costs exactly that. Nothing is metered by length or time.
Am I charged if a run fails?
No. Starting a run holds the credits rather than spending them. The hold becomes a charge only when the run succeeds; a failed or cancelled run releases it.
Can I edit the brief and run it again?
Yes. Change any input and press Generate again, or open an earlier run and choose Adjust and run again to start from a copy of its brief. Uploaded files need attaching again. Each run is new and costs the applet's price; the earlier run stays as it was.
Where are my results kept?
Every run you start is listed under Your runs at /create/runs. Each one opens its own page, where you can read, copy and download the files it produced.
More apps
All Thumbnails apps- Thumbnail MakerRanked thumbnail concepts and finished image candidatesTry it · 12 credits

- Channel BannerMatching channel art built from one brand directionTry it · 14 credits

- Cover MakerVertical graphics from a quote, takeaway or hookTry it · 10 credits
- Voice GuideYour writing samples, turned into a reusable voice guideTry it · 4 credits