How do you actually A/B Test YouTube Thumbnails?
For the past few months, my clients and I have been A/B testing YouTube thumbnails in early access. These are 7 new things I’ve learned (so far):
For the past few months, my clients and I have been A/B testing YouTube thumbnails in early access. These are 7 new things I’ve learned (so far):
1) Maximizing Effectiveness
Creating three vastly different thumbnails is the main principle when you test thumbnails. This way, you create the most significant results in your first A/B test to base future tests on.
Something I’ve been pushing to try is testing a thumbnail that matches the intro of the video as one of the three thumbnails in the test.
This approach can build trust with the audience, as they'll see that you're not engaging in clickbait. Over time, this trust could lead to quicker clicks on your future content, benefiting both your audience and your content
1) Maximizing Effectiveness
Creating three vastly different thumbnails is the main principle when you test thumbnails. This way, you create the most significant results in your first A/B test to base future tests on.
Something I’ve been pushing to try is testing a thumbnail that matches the intro of the video as one of the three thumbnails in the test.
This approach can build trust with the audience, as they'll see that you're not engaging in clickbait. Over time, this trust could lead to quicker clicks on your future content, benefiting both your audience and your content
2) Iterative Testing
After running your initial test, it's time to fine-tune the thumbnail by testing iterations of the winning version. These small changes can help optimize the thumbnail to achieve even better results.
Examples of things you could try to change are:
- Text on thumbnails
- Font styles
- Color combinations
- Facial expressions
- Objects
2) Iterative Testing
After running your initial test, it's time to fine-tune the thumbnail by testing iterations of the winning version. These small changes can help optimize the thumbnail to achieve even better results.
Examples of things you could try to change are:
- Text on thumbnails
- Font styles
- Color combinations
- Facial expressions
- Objects
3) A/B Testing Timeline
YouTube pushes your video to different audiences over time:
Directly after your upload: your Core Audience (subscribers or people who previously watched your content)
A few days after your upload: your Casual Audience (people who might be interested in your content based on their watch history)
This means you will likely gain the most insights from your tests if you run them a couple of times in different timeframes: one when you upload, one a few days after your upload and then extra tests a few weeks/months later.
Your initial test will most likely target your core audience, which might appreciate a different type of packaging than the general audience later on. For example, they might be more enticed to click on videos when the creator’s face is clearly recognizable, or if the thumbnail looks more authentic.
It is a smart idea to redo the A/B test after a couple of weeks to gain continuous insights into your audience behavior as trends, events, and audience interests change over time.
3) A/B Testing Timeline
YouTube pushes your video to different audiences over time:
Directly after your upload: your Core Audience (subscribers or people who previously watched your content)
A few days after your upload: your Casual Audience (people who might be interested in your content based on their watch history)
This means you will likely gain the most insights from your tests if you run them a couple of times in different timeframes: one when you upload, one a few days after your upload and then extra tests a few weeks/months later.
Your initial test will most likely target your core audience, which might appreciate a different type of packaging than the general audience later on. For example, they might be more enticed to click on videos when the creator’s face is clearly recognizable, or if the thumbnail looks more authentic.
It is a smart idea to redo the A/B test after a couple of weeks to gain continuous insights into your audience behavior as trends, events, and audience interests change over time.
4) A/B Testing with Smaller Audiences
Doing multiple tests is always a good idea when you’re A/B testing, and it’s even more important for smaller audiences.
With a smaller audience, the number of viewers engaging with each thumbnail variant is limited. This obviously leads to a smaller sample size, making it difficult to achieve statistically significant results.
For example, the variations in viewer engagement could be due to random factors rather than the effectiveness of a particular thumbnail. You'll only be sure your winning thumbnail is the most effective after you've run multiple tests with similar results.
4) A/B Testing with Smaller Audiences
Doing multiple tests is always a good idea when you’re A/B testing, and it’s even more important for smaller audiences.
With a smaller audience, the number of viewers engaging with each thumbnail variant is limited. This obviously leads to a smaller sample size, making it difficult to achieve statistically significant results.
For example, the variations in viewer engagement could be due to random factors rather than the effectiveness of a particular thumbnail. You'll only be sure your winning thumbnail is the most effective after you've run multiple tests with similar results.
5) Understanding Results
The winning thumbnail might not be the one that gained the most clicks during your test. YouTube’s tool is based on watch time share, which means it’s optimized for audience satisfaction (i.e., the winning thumbnail got people to click and to watch the video for a long time).
If you want to just optimize your thumbnails for clicks, you could try using third-party tools that give insights into CTR during tests, or look at the CTR of your winning thumbnail after the test's completion.
5) Understanding Results
The winning thumbnail might not be the one that gained the most clicks during your test. YouTube’s tool is based on watch time share, which means it’s optimized for audience satisfaction (i.e., the winning thumbnail got people to click and to watch the video for a long time).
If you want to just optimize your thumbnails for clicks, you could try using third-party tools that give insights into CTR during tests, or look at the CTR of your winning thumbnail after the test's completion.
6) A/B Testing Backlog
Thumbnail testing isn’t just a great way to optimize new content; it can also help revive previous content you’ve posted.
Focus on content that still receives a good number of impressions, has high watch time, and has current packaging that you can easily improve.
6) A/B Testing Backlog
Thumbnail testing isn’t just a great way to optimize new content; it can also help revive previous content you’ve posted.
Focus on content that still receives a good number of impressions, has high watch time, and has current packaging that you can easily improve.
7) Alternative Testing Methods You Could Try
Sometimes you might not be able to think of three unique and potentially enticing thumbnail concepts to test. Instead of using only two thumbnails to test, you could try to: - Test a video screenshot/simplistic picture - Test a new thumbnail style for your channel (minimalistic, memey, illustrations)
7) Alternative Testing Methods You Could Try
Sometimes you might not be able to think of three unique and potentially enticing thumbnail concepts to test. Instead of using only two thumbnails to test, you could try to: - Test a video screenshot/simplistic picture - Test a new thumbnail style for your channel (minimalistic, memey, illustrations)
Featured in
Featured in