Skip to content

YouTube thumbnail A/B testing

YouTube lets you test up to three thumbnails on one video from YouTube Studio on a computer: open the video, click A/B Testing under "Thumbnail" and upload your options. Within two weeks, the one with the highest watch time is shown to everyone.

YouTube thumbnail downloaderUpdated 3 min read

Who can use it

  • A computer. A/B testing is only in YouTube Studio on a computer.
  • Advanced features. Your channel needs advanced features. You may have them already through your channel history, or you can unlock them by verifying your identity with an ID or a video.
  • A long-form video. You can't test Shorts, scheduled live streams or Premieres. Premieres can be tested once they end and become normal videos. Live archives can be tested.
  • The right settings. Videos made for kids, for mature audiences, or set to private can't be tested. If you test age-restricted content without marking it as 18+, you lose access to the feature.

Run a test

  1. Sign in to YouTube Studio.
  2. Pick the video. For a new one, click Create, then Upload videos. For an existing one, click Content in the left menu and click the video.
  3. In the Title box or under "Thumbnail", click A/B Testing.
  4. Choose Thumbnail only, Title only or Title and thumbnail.
  5. Upload up to three thumbnails.
  6. Click Done.

Don't change the title or thumbnail while the test runs. If you do, the test stops and you'll have to start it again.

Make every option at least 1280 × 720

If any thumbnail in a test is under 1280 × 720, YouTube shows all of them at 854 × 480. Make each option 1280 × 720 or larger, at 16:9. YouTube recommends 3840 × 2160 for thumbnails in general.

Drop each image into the resizer below. It starts on YouTube's thumbnail preset, crops to 16:9 around the point you choose and downloads a JPG or PNG. It runs in your browser, so your images aren't uploaded anywhere.

Social media image resizer

Open the full tool

How the result is picked

YouTube shows the options to different viewers and compares their watch time share, not clicks. It may keep a small control group who only see the default thumbnail; their views don't count in the result. You'll get one of three outcomes:

  • Winner. One option clearly did better. It's shown to all viewers.
  • Performed same. The options did about the same. Pick the one you like.
  • Inconclusive. No clear difference. The first thumbnail you uploaded stays.

To follow a test, open the video's Analytics, go to the Reach tab and click Manage test under "How your A/B test is going". You can stop a test from there at any time. Once a test ends, you can still change the thumbnail by hand.

Tips from YouTube

  • Start with older videos, so a weaker option doesn't hurt your channel's views as much.
  • Make the options clearly different. Similar thumbnails make the test run longer, and may not produce a winner.
  • Give it time. A test can take from a few days to two weeks.
  • Use high-resolution images for every option.

Running the same test twice can give different results. YouTube puts this down to chance and to changes in who's watching: early viewers are more likely to know your channel already.

Checking a thumbnail before you publish

An A/B test needs real viewers to give a result. Before you publish, YouTube's tips suggest thinking about who the video is for and where they'll find it, and keeping the design simple. After publishing, check click-through rate on Home and Suggested in the first 24 hours. See YouTube thumbnail guidelines for the rules and tips, and why a YouTube thumbnail looks blurry if an option looks soft.

Sources

Questions

How long does a YouTube thumbnail test take?

A few days to two weeks. It depends on how many impressions the video gets, how recently it was published and how different your options are.

Can I A/B test a Shorts thumbnail?

No. YouTube doesn't let you A/B test Shorts, scheduled live streams or Premieres. A Premiere becomes eligible once it ends and turns into a normal video, and live archives can be tested.

Does YouTube pick the winner by click-through rate?

No. It picks the option with the highest share of watch time, so a thumbnail that gets clicks but disappoints viewers doesn't win.

Why didn't my test have a winner?

The options may have been too similar to make a measurable difference, or the video didn't get enough impressions. You'll see "Performed same" or "Inconclusive" instead, and the first option you uploaded stays.