A/B Thumbnail Compare
Upload 2–4 candidate thumbnails, shuffle them into a simulated sidebar surrounded by generated filler tiles, and click whichever one your eye lands on first. Repeat a few times — a thumbnail that wins consistently across shuffles is a much stronger signal than a single glance.
1. Upload your candidates
Fill in at least two of the four slots below. Images are read locally in your browser and never uploaded.
2. Run the eye-test
Click the tile your eye lands on first. It reshuffles automatically for the next round. 0 rounds played
Upload at least two variants above to start the eye-test.
3. Tally
| Tile | Clicks |
|---|
How to run a quick eye-test
Keep everything else constant — same sample title, same surrounding tiles — and vary only the thumbnail image, which is exactly what this page does. Upload your real candidates, then click through several shuffled rounds. A thumbnail that keeps winning regardless of where it lands in the column is a genuinely stronger visual pull than one that only wins when it happens to sit at the top.
Why distractor tiles are included
A thumbnail never competes in isolation — on a real watch page or home feed it's surrounded by a dozen other videos fighting for the same half-second of attention. Testing a thumbnail alone, with nothing to compete against, tells you almost nothing about whether it stands out. The four distractor tiles here are plain CSS gradients generated in your browser (no stock photos, no other creators' work, no external requests) with neutral filler labels, just enough visual noise to make the comparison realistic without being a distraction in themselves.
Limitations of a self-test
You already know which tile is yours, so you are not a neutral tester — your eye may land on your own thumbnail out of familiarity rather than genuine visual pull. Treat this as a fast first pass, then, if the decision matters, show the same shuffled rig to a few other people and ask the same question cold. This measures instinctive visual attention, not click-through rate, topical relevance, or whether the thumbnail accurately represents your video — all of which still matter and aren't something any automated tool can judge for you.