One of the most important aspects of improving anything is actually making sure you're improving. It sounds obvious, but many people don’t properly track how well their thumbnails are doing, and as a result, they aren’t actually sure which changes are making a difference and which ones aren’t. Properly tracking your thumbnails' performance will help negate these issues and, as a result, make it easier for you to get more views.
CTR, or click-through rate, is just the percentage of people who saw your thumbnail and clicked. You can find this data in YouTube Studio (or whatever other platforms' analytics you use). This is easily the most important part of tracking your thumbnail performance, as it doesn’t matter how good or bad the video itself actually is; if your thumbnail is bad, the CTR will be low. Now, to clarify, the title of your video can also impact CTR, so if you want to really isolate for thumbnail performance, you can either try to keep titles as similar in style as possible or you can A/B test them. As a disclaimer, you need to be careful when optimising for CTR or any other metric, as making thumbnails ONLY thinking about CTR could cause you to start creating “clickbait” (basically just misleading thumbnails), which likely will hurt your reputation. So make sure you still keep up the quality.
Split testing, or A/B testing as mentioned above, is when you compare 2 versions of a thumbnail to see which one will perform better. This is the purest way to measure thumbnail performance, as if you don’t change the title during your A/B testing, the only factor that could impact your CTR is the 2 versions of the thumbnails. You can also A/B test things like the title; however, this article is about thumbnail performance specifically. To properly A/B test, only isolate one feature about your thumbnail so you can guarantee any difference to the thumbnail's performance is due to this one feature alone. For example, let's say you have a basic thumbnail with some black text. A good A/B test would be making another version of this thumbnail but with text coloured something bright, like red or yellow, as maybe you think that would get more people's attention. Now, if the brightly coloured text thumbnail outperforms the other, your hypothesis was right. If there is no difference, then you would know that the specific text colour you chose had no impact on thumbnail performance compared to regular black. Now if the brightly coloured text thumbnail performed worse, then your assumption would be completely wrong, and the black text thumbnail is actually superior. Make sure to rigorously A/B test any feature you think may or may not hurt thumbnail performance. Sidenote: If something will obviously help your thumbnail, for example, improving image quality, then just skip the A/B testing, as it wouldn’t be worth the time to prove something almost guaranteed.
Now one final thing you should consider is how your thumbnails are performing with your subscribers and how they are performing with a general audience. It’s important to understand that people subscribed to you are likely used to a certain style of thumbnail, so drastic changes could turn off your dedicated viewers. If you're a smaller channel, this isn’t an issue; however, for larger channels, this is something to keep in mind. If you want to grow your channel, then obviously you would have to target more towards a general audience, which could also hurt subscriber CTR, which is also something to keep in mind. With all that said, if a change helps subscriber and general audience CTR, it's probably a good change. To find general audience CTR, find both the homepage and the recommended feed's CTR. For subscriber CTR, use the subscription feed CTR. You might be wondering, “If a subscriber doesn’t open the video from their subscription feed but instead the homepage, would this be properly tracked? ”. The answer to that is no; however, this method of getting subscriber CTR still accurately tracks how many of your most dedicated followers react to your thumbnails. Once again, you can get all this data in YouTube Studio/equivalent analytics platforms.
To accurately track how well your thumbnails are doing, measure how many people click on your thumbnail after seeing it (CTR), compare different versions of a thumbnail to see which one increases CTR (A/B testing), and make sure your thumbnails still perform well with a general audience and your subscribers.