YouTube Thumbnail A/B Test: How Test & Compare Works

Most thumbnail advice is opinion. YouTube's A/B test is the one tool that measures real viewers on your own video, which makes it worth knowing exactly what it can and cannot tell you. This guide sets out how it works from YouTube's own pages, how to read a result that says "Inconclusive", and what to do when your video is too small to test.

What is YouTube's thumbnail A/B test?

YouTube Help says creators can test and compare up to 3 different titles and thumbnails, and that at the end the option with the highest watch time is shown to all viewers. You choose one of three test types: title only, thumbnail only, or title and thumbnail.

Viewers see the options at the same time, not one after another. YouTube Help calls its tests true, concurrent A/B/C tests, and says many third-party tools run tests sequentially and often optimize for click-through rate, so their results can disagree with YouTube's. It also says YouTube may keep a small share of viewers as a control group that only sees the default option, and that this group is left out of the calculation.

Who can use it, and what cannot be tested?

A video that later becomes a Short loses access to its earlier tests and cannot be tested again. Uploads of up to three minutes that are square or taller count as Shorts, so check the shape of a short video before you plan a test. Our Shorts length and size guide has the rules.

How do you run a test?

1. 01

   Pick the video

   Sign in to YouTube Studio. For a new video, choose Create, then upload it. For an existing video, open Content and select it.

2. 02

   Open A/B Testing

   Under the title box or under Thumbnail, click A/B Testing, then choose Title only, Thumbnail only, or Title and thumbnail.

3. 03

   Upload up to 3 options

   Add the thumbnails or titles to compare, then click Done.

4. 04

   Leave the video alone

   YouTube Help says that if the title or thumbnail is changed during the test, the test stops automatically and has to be restarted.

5. 05

   Review or stop it

   Open Analytics, then the Reach tab, and under How your A/B test is going choose Manage test. You can review results or stop the test at any time.

How does YouTube pick a winner?

The result is based on watch time share. YouTube Help gives its reason: a good title and thumbnail also help a viewer understand what a video is about so they do not waste time clicking the wrong one, so tests are optimized for overall watch time over other metrics such as click-through rate.

Why is there no winner, and what should you do?

YouTube Help says it is normal not to get a Winner and gives two reasons: the difference between the options was too small to measure, or the video did not get enough impressions. More views make a winner likelier.

It also explains why the same video can give different results on different runs. A test is a real-world experiment with chance in it, like flipping a coin, and the audience changes over time: early impressions are more likely to come from viewers who already know your channel, and later ones are more likely to include people who have not watched it. Read one result as evidence, not as a verdict.

Three steps follow from that, all from the Help page's tips:

1. Test options that are clearly different. Options that are too similar make tests run longer.
2. Give the test time. It can take a few days or up to 2 weeks depending on impressions and how recently the video was published.
3. Start with older videos. YouTube Help recommends this to limit the effect on a channel's overall views, then to move to videos that matter most for your strategy.

What should you change between options?

YouTube Help does not say what to vary, so this part is our suggestion. Change one idea per option so a result points to something. The options below differ in a way a viewer can see at a glance.

Do not chase clicks. YouTube Help says to avoid trying to raise click-through rate with clickbait thumbnails or titles, because clickbait tends to show high click-through rate but low average view duration, and YouTube recommends a video when it is relevant to viewers and their watch time shows they find it interesting.

How do impressions and click-through rate fit in?

You still see click-through rate in Analytics, and it helps you read a test. YouTube Help says half of all channels and videos have an impressions click-through rate between 2% and 10%, that new videos and videos with fewer than 100 views can see a wider range, and that a video with many impressions, for example on the home page, naturally has a lower rate. It suggests comparing rates between videos over the long term and keeping the traffic source in mind.

An impression is counted when the thumbnail is shown for more than 1 second and at least 50% of it is visible. Impressions are counted on YouTube search, the home page, feeds and Up Next recommendations, and not on external websites and embeds, YouTube Kids, YouTube Music, email or notifications, so a test measures how a thumbnail does inside YouTube.

What if your video is too small to test?

A test needs impressions, and YouTube does not publish a minimum. On a small channel, expect more Performed Same and Inconclusive results. You still have options:

• Check how a thumbnail looks before you upload. Our Thumbnail Preview Simulator shows your image with your title in search results, the home feed, the suggested sidebar and on mobile, in light and dark mode. A design that reads at small size is a fair start.
• Learn from your own videos. Compare the click-through rate and average view duration of similar videos over time, and treat a swap on one video as a hint. It is not a controlled test, because the audience changes between the before and after periods.
• Study a thumbnail you admire. Our Thumbnail Downloader guide covers saving one for study and the rights that apply.

Do most thumbnails clear the 720p bar?

YouTube Help says that if the resolution of any thumbnail in a test is lower than 720p (1280 x 720), all the test thumbnails are downscaled to 480p (854 x 480), and that it recommends the highest resolution, 3840 x 2160 for videos. We checked how common that is. The Data API lists a thumbnail size called maxres, described as 1280 x 720 and available for some videos, and larger ones such as 1920 x 1080 and 3840 x 2160.

We read the thumbnail sizes of 2,361 long-form videos, meaning longer than 3 minutes, from the latest 50 uploads of 115 channels drawn at random from the ones our site tracks, on September 26, 2026, using 48 API quota units. Our tracked set leans toward larger channels.

Overall, 771 videos (32.7%) exposed a thumbnail 1920 pixels wide or more, and 91 (3.9%) exposed one of 3840 pixels. The other 151 videos, 6.4%, had no 1280-pixel version, and 148 of them topped out at 640 pixels wide. The API does not say why a size is missing, so read this as a sign that a small share of videos, mostly on smaller channels, were uploaded at lower resolution. Our thumbnail size guide has the full rules.

FAQ

Related reading