Run a thumbnail A/B test
before you hit publish
Compare 2 to 6 ideas with an explainable preflight score. Improve the weak signals, then validate your strongest three with real viewers on YouTube.
Free core test · No channel connection · Local scoring
- Scoring runs in your browser
- No YouTube login
- Originals stay local
- Open scoring algorithm
From creative draft, to pre-publish check, to real audience validation.
- Create thumbnail and title ideas
- Shortlist 2 to 6 concepts here
- Validate the final 3 on YouTube
Pre-publish workflow
How it works
Drop in 2 to 6 covers + titles
Upload JPG, PNG, or WebP files from your device. Core scoring stays in your browser.
Get an explainable 12-signal preflight score
Each concept gets a 0 to 100 relative score across 5 visual and 7 title signals, plus a clear list of what to fix.
Shortlist the strongest three
Use the ranking to improve your ideas, then take up to three finalists into YouTube Studio for a real audience test.
Every cover scored across 5 dimensions
A local canvas pass reads saturation, contrast, warm-color ratio, brightness, and skin-tone presence, then maps them onto five scores. You see exactly which dimension is holding a cover back, so you can fix the one detail that turns a scroll-stopper into a skip.
- Scoring runs in your browser
- Original image stays local
- Five visual dimensions, seven title angles
Your title scored across 7 angles
Hooks, emotion, conflict, length, structure, keywords, readability. The scorer reads the title and cover as one unit, and flags the style it sees: shock, question, number.
- Scores title and cover together
- Detects the title style
Then it ranks the set and tells you what to fix.
A repeatable preference model
The same inputs produce the same relative ranking across 10,000 deterministic model iterations. The chart is a comparison index, not audience research or a CTR forecast.
The top candidate, explained
Lift cover contrast so the focal subject still reads at feed size.
Variants ranked by preflight score
See which concept leads this set, why it leads, and what to revise. Shortlist the strongest options, then validate them with YouTube's native test.
Fixes ranked by impact
Not a vague score. A short list of what's dragging each variant down, in order. Fix the biggest leak first, then retest.
Start free. Choose Creator when you need to compare revisions.
See full plans →Free
Run the complete pre-publish workflow today.
- 5 visual + 7 title signals
- Relative ranking + focused fixes
- No login for core local scoring
Creator
For creators who review several revisions together.
- Grouped revision history
- Side-by-side revision comparison
- No-watermark decision reports
Studio
A team-workflow research direction that is not available for purchase.
- Creator workflow if Studio ships
- Batch + shared review under research
- No feature or launch-date promise
Questions, answered straight
The short version of what the tool does and does not do.
What does the preflight score measure?
It combines 5 pixel-based cover signals with 7 title signals, then compares every option in the same set. The result is a heuristic ranking for revision work, not a forecast of your YouTube click-through rate.
Can the score predict performance?
No. It has not been calibrated against your channel's real results, so we do not claim an accuracy rate. It is deterministic: the same inputs produce the same ranking. That makes it useful for comparing revisions, but it does not create new evidence about audience response.
When should I run a thumbnail test?
Before you publish. Use the preflight to narrow a larger batch and fix obvious weak signals. Then put your strongest options into YouTube's native test, which uses real watch-time data after publication.
How many covers can I compare?
Up to 6 covers and titles in a single test, today. Drop in anywhere from 2 to 6 variants and the scorer ranks all of them.
Does it work for YouTube Shorts?
It can score a Shorts cover and title as a still-image concept. That remains a heuristic preflight, not a native YouTube Shorts experiment or a measurement of Shorts performance.
Can I test the title and thumbnail together?
Yes. We score the title and thumbnail as a single unit, because that's how viewers actually see them in the feed.
Is this the same as YouTube's "Test & Compare"?
No. YouTube's native test uses real watch-time data after publication with up to 3 options. This tool compares up to 6 ideas before publication with a deterministic heuristic, so it is best used to build that final shortlist.
Do I need to connect my YouTube channel?
No, and you never will. The tool never touches YouTube, asks for no OAuth, and needs no channel access. It reads only the images and text you drop in for scoring.
Are my thumbnails uploaded to a server?
Core scoring happens locally in your browser. Account sync sends a downsized preview plus title and score data only after sign-in and an explicit save.
Choose your next step
After the quick score, choose the evidence you need next
Need the whole process? Start with the workflow. Use the methodology to inspect a score, and compare it with YouTube's live test before putting finalists in front of viewers.
Stop guessing which cover to ship.
Run the preflight before you publish. Scoring stays in your browser and never connects to YouTube.