How does YouTube's thumbnail Test & Compare work?

When you edit a video's details in Studio on a desktop, the thumbnail section may offer "Test & Compare". Upload two or three versions, and YouTube splits impressions between them over the following days. At the end, it reports which version earned the largest share of watch time and applies it, or tells you the result was too close to call. You can also stop the test early and choose one yourself.

The choice of watch time over click-through is deliberate. A thumbnail can win clicks by promising something the video does not deliver, then lose viewers in the first seconds. Measuring watch time rewards the version that brings in people who actually stay. That makes the test useful, but it also means a thumbnail with a slightly lower click-through rate can win, which surprises people reading only the CTR column.

To get a meaningful answer, test genuinely different ideas, not tiny changes. A face with an expression against a product close-up is a real test; the same design in two shades of red is not. The test also needs impressions, so videos on small channels with few views may never reach a conclusion. If the feature is missing on your channel, compare manually: note a video's click-through rate over a stable week, swap the thumbnail, and look at the following week, remembering that other things such as traffic source also change.