YouTube Thumbnail Testing: Contrast Checks, Mobile Previews and A/B Experiments
This page is about testing thumbnails, not their dimensions. For sizes and layout, see the YouTube thumbnail design guide. Here, the goal is a repeatable process for checking whether a thumbnail works before and after you publish, using text images from the free Brat Text Generator.
Pre-publish checks
| Check | How | Pass condition |
|---|---|---|
| Thumbnail size | Shrink to about 168 px wide | Text is still readable |
| Grayscale | Convert to black and white | Text still stands out |
| Mobile | View on a phone at arm’s length | You can read it in a second |
| Dark mode | View on a dark background | No clashing edges |
Text rules that test well
- Three words or fewer.
- Large, high-contrast type. See the colors guide.
- One focal point.
Running an A/B test
YouTube offers a built-in thumbnail testing feature in some channels. If you have it, upload two or three versions that differ in only one way, such as text color or word choice, and let the test run to completion. If not, change the thumbnail after a few days and compare click-through rate over equal periods, keeping in mind that timing affects results.
Reading results
| Metric | Meaning |
|---|---|
| Impressions | How often it was shown |
| Click-through rate | Clicks divided by impressions |
| Average view duration | Whether viewers got what the thumbnail promised |
A high click-through rate with a low view duration means the thumbnail overpromises.
Keep a test log
Record what you changed, the dates and the results. Over ten tests you learn what works for your audience.
Avoid misleading thumbnails
Thumbnails must reflect the video content. Misleading thumbnails violate platform policies and hurt long-term trust.
Planning a thumbnail before you film
A thumbnail is easier to design when you know the single promise of the video. Write it as three words, such as the result or the question, and generate that text first. Then plan a plain background and one visual focus, so the image reads in under a second.
Testing at small sizes
On a phone, thumbnails in a list are small, and on a TV they are large. Check the design at about 160 pixels wide and again at full size. Keep the main text away from the bottom-right corner, where the video length label may sit.
Comparing versions
If your platform lets you test thumbnails, compare two versions using the same title. Otherwise change one element, such as the background colour, and compare click-through over the same number of impressions.
A worked example: testing two headlines
A small channel uploads a video about budget meal prep. Version A of the thumbnail says “cheap meals”. Version B says “5 meals, $20”. Everything else is identical. After the test period, B has a higher click-through rate and a similar average view duration, so B is the winner. The creator notes that specific numbers beat vague phrases for this audience and tries a similar change on the next video.
Troubleshooting a test
| Symptom | Likely cause | Fix |
|---|---|---|
| No clear winner | Too few impressions or variants too similar | Run longer or change one bigger thing |
| High clicks, low watch time | Thumbnail overpromises | Make the thumbnail match the video |
| Results swing day to day | Timing and traffic sources vary | Compare equal periods |
| Thumbnail unreadable on phone | Text too small | Increase size, cut words |
What to test
Change one variable at a time: headline wording, text color, number of words, or the focal element. Testing two things at once makes results unclear.
Keeping honest data
Write down the date, the change and the numbers. Look for patterns after at least five tests.
When not to test
Very small channels may not get enough impressions for tests to be meaningful. In that case, rely on the pre-publish checks and review after a week.
Designing thumbnails that get read
Thumbnails are small, so use at most three words, large type and strong contrast. Place the text away from the bottom-right corner, where the video duration label appears.
Testing
View the thumbnail beside other videos in your niche and check whether it stands out and remains readable on a phone.
What a thumbnail has to do
A thumbnail has to earn a click in under a second, so it should answer one question: what will I get from this video? One strong word, a clear face or object and high contrast usually beat a crowded design.
Building one step by step
- Write the video’s single promise in three words or fewer.
- Generate those words in the Brat Text Generator at 1280 by 720.
- Combine them with a simple image in your editor, placing text where the image is calm.
- Shrink the result to about 160 pixels wide and check that it still reads.
- Upload and compare it against competing videos in the same search.
Avoiding common mistakes
- Text that repeats the title exactly.
- Tiny text that disappears on phones.
- Misleading images that do not match the video, which can lower viewer trust.
Improving over time
Check click-through rate in your analytics, and test one change at a time, such as the background colour or the word. Keep records so you learn which changes work for your audience.
Reading your analytics honestly
Thumbnail tests only mean something with enough impressions. Wait until each version has been shown many times before comparing, and change one thing at a time. Note the date, the version and the result in a simple sheet. After a few tests you will see which kinds of words and colours your audience responds to, and you can stop guessing.
Frequently asked questions
How many variants should I test?
Two or three; more splits the data too thin.
How long should a test run?
Until each variant has enough impressions to compare; very short tests are noisy.
Does the thumbnail affect search ranking?
It affects click-through rate, which can influence how widely a video is shown.
Should every video have text?
Not necessarily, but a short phrase often helps. See the design guide.
What size should a thumbnail be?
1280 by 720 pixels is the commonly recommended size; check YouTube’s current guidance.
Brat Generator is an independent tool and is not affiliated with, endorsed by, or sponsored by Charli XCX, her record label, or any related brand. Any album or trademark names are mentioned only to describe a visual style.