Growth

How to A/B Test YouTube Thumbnails with Test & Compare

How YouTube's built-in Test & Compare feature works, who can use it, how to set up a thumbnail test, and how to read the results without fooling yourself.

Article cover: How to A/B Test YouTube Thumbnails with Test & Compare
In this article
  1. Who can use it
  2. Setting up a test
  3. How the winner is chosen
  4. Designing thumbnails worth testing
  5. Reading the results honestly
  6. No access yet? Test manually

Guessing which thumbnail will perform best is hard, even for experienced creators. YouTube’s built-in Test & Compare feature lets you upload several thumbnails for one video and see which one viewers respond to.

Who can use it

Test & Compare is available in YouTube Studio on a computer for channels with access to advanced features. If you don’t see the option, check that your channel is verified and has advanced features enabled under Settings → Channel → Feature eligibility. YouTube rolls features out gradually, so availability can vary.

Setting up a test

  1. Open YouTube Studio on a computer and go to Content.
  2. Open the video’s Details page.
  3. Under Thumbnail, choose Test & Compare.
  4. Upload two or three thumbnails (1280 × 720 each).
  5. Click Done and save.

YouTube then shows the versions to different viewers and reports the result in the video’s analytics. Tests usually run for days to a couple of weeks, depending on how many impressions the video gets.

How the winner is chosen

YouTube compares versions by watch time share — the share of total watch time each thumbnail generated. You’ll see one of three outcomes:

ResultWhat it means
WinnerOne thumbnail clearly performed better; YouTube applies it
PreferredOne did somewhat better, but the difference is less certain
NoneNo meaningful difference — pick the one you like

Designing thumbnails worth testing

A test only teaches you something if the versions are genuinely different. Change one big thing per version:

  • Subject: your face vs. the product vs. the result.
  • Text: a question vs. a number vs. no text at all.
  • Colour: bright background vs. dark background.

Testing three nearly identical images usually ends with “None”. Before setting up a test, line up your options in our thumbnail preview tool to remove weak ideas early.

Reading the results honestly

  • Don’t stop early. A version that leads after a day can easily lose later.
  • One test is one data point. Look for patterns across several videos before changing your whole style.
  • Keep notes. Save each set of thumbnails with the result. Over a few months you’ll learn what your audience prefers, which is more valuable than any single winner.

No access yet? Test manually

You can still compare thumbnails without the feature: change the thumbnail on an older video, wait at least a week, and compare Impressions click-through rate and average view duration in Analytics with the previous period. It’s less precise, because traffic changes over time, but it still reveals big differences. See how to change a YouTube thumbnail for the steps.

Frequently asked questions

Why does Test & Compare measure watch time instead of clicks?

A thumbnail that gets many clicks but makes viewers leave quickly is not really better. Watch time share rewards thumbnails that bring in viewers who actually watch.

Can I test thumbnails on Shorts?

Test & Compare has been available for long-form videos. Check YouTube Studio for the current list of supported video types, as YouTube expands features over time.

Grab any YouTube thumbnail in seconds

Paste a link and download every available size, up to 1280×720.

Open the downloader
Free Thumbnail Downloader Team

We build small, fast tools for YouTube creators and write guides based on testing them ourselves.

Keep reading