Add the first thumbnail
Upload a local design or import a public thumbnail to compare the full title-and-image package.
How to use the thumbnail preview
Compare up to four thumbnail variants with one video title across approximate home, search, suggested, and compact feed surfaces. Local files stay inside the browser tab. The lab describes dimensions and presentation risks without claiming to predict click-through rate, watch time, or a winning thumbnail.
Judge the title and image as one package
Upload two to four genuinely different directions, then add the working title, channel label, duration, view label, and publication age. Switch between wide and compact surfaces to see what survives when the image shrinks and the title truncates. Use grayscale to inspect value contrast and the squint test to check whether the main subject and broad composition remain recognizable without fine detail.
Change one packaging hypothesis at a time when possible. For example, compare a close crop with a wider evidence-led frame, or compare a result-focused phrase with no thumbnail text. If every element changes between variants, the review may reveal a preference but cannot explain which decision caused it.
Continue with a real YouTube experiment
This lab is a pre-publish visual review, not a live A/B test. It has no impression, click, retention, or watch-time data and does not rotate thumbnails to viewers. YouTube's native Test and Compare feature can run a concurrent experiment with eligible public long-form videos and evaluates variants using watch-time share rather than a third-party visual score.
Treat the exported PNG as a review sheet for collaborators, not statistical evidence. After publishing, use YouTube's native experiment when eligible and interpret the result with traffic source, impressions, click-through rate, watch time, and viewer satisfaction together. A thumbnail that earns a click but breaks the video's promise is not a durable packaging improvement.
Questions about this tool
Is this a real thumbnail A/B test?
No. It is a pre-publish visual comparison. A real experiment needs viewer impressions and outcome data; use YouTube's native Test and Compare feature when the video is eligible.
Are uploaded thumbnails sent to Free Creator Toolkit?
No. Local JPG, PNG, and WebP files are represented with temporary browser object URLs and are released when removed or when the tab closes.
Why does the preview differ from YouTube?
YouTube changes layouts and adapts them by device, account, language, and experiment. These surfaces are approximate inspection contexts, not pixel-identical replicas.