🧑‍🎨 The New Thumbnail A/B Test Feature

How to YouTube #16

Hey YouTubers,

If you haven’t heard yet, at some point this year YouTube are releasing a new thumbnail A/B test tool.

In YouTube studio, you’ll be able to upload 3 thumbnails with a video and see which one performs best, based on watch time per thumbnail.

We’ve actually had access to a beta version of this tool for the last few months, so today I wanted to share a few things:

• What does it do?

• Is it useful?

• Will it replace the current A/B test tools?

• What does this all mean?

What does it do?

It’s very simple. When you upload a thumbnail to a video, instead of being able to only upload 1 thumbnail, you can upload 3.

It looks like this.

You can also go back and run tests on all your old videos too.

When you start the test, it then shows all 3 thumbnails to a range of viewers, all at the same time.

For example, at 11am viewer 1 sees thumbnail A, viewer 2 thumbnail B and viewer 3 thumbnail C. It is also possible though that viewer 1 sees all 3 thumbnails. If they don’t click on thumbnail A, then YouTube might show them thumbnail B later.

This is different to other external thumbnail A/B test tools, which are only able to switch out thumbnails every hour or every day.

After you start the test on a video, it takes a couple hours for some data to appear. This is one we’re currently running:

And then after a certain amount of time, you get the results.

The tests are surprisingly short most of the time, often running less than 24 hours. But sometimes, if the thumbnails perform similarly, it can take a bit longer for the test to finish.

In some instances, YouTube say the test was inconclusive because they performed so similarly. But most of the time you get a result that looks like this:

The percentage on the right is based on how much of the total video watch time the thumbnail gained during the select period.

E.g. If during a certain time period there were 100 total hours of watch time on this video from all the viewers who clicked on it, thumbnail 1 would have contributed ~31 hours of that, thumbnail 2 around ~35 hours, and thumbnail 3 around ~34 hours.

Therefore, thumbnail 2 was the best performing thumbnail during this period.

They use watch time because ultimately watch time is the metric that YouTube care about most. They just want people to stay on YouTube consuming content they like. So we’re encouraged to make thumbnails that both get the viewer to click but also accurately represent the video.

And if they used click-through rate, then we’d all be encouraged to make more clickbaity thumbnails.

Is it useful?

At face value, yes this tool is very useful.

For all I’ve learned about thumbnails, and all the advice from the army of thumbnail gurus on Twitter, it’s still hard to predict what will perform best.

We now upload 3 thumbnails for almost every video we post to make sure that we are in fact using the highest performing one.

This takes the guessing out of it.

But there are a few caveats to this tool.

There’s no such thing as a perfect thumbnail. Something that performs one way today, might not perform the same way tomorrow.

Similarly, an A/B test run today might have different results to one run tomorrow, because the viewers will be different. Therefore this data is not perfect, and we should factor that into our decision making.

Will it replace the current A/B test tools?

The two best options over the past few years for A/B tests have been TubeBuddy and Thumbnail Test. The latter is by far the better of those options now though (I wrote a long post about running A/B tests and using this tool here if you’re interested).

Using these tools for thumbnail tests will potentially become pointless because YouTube will be doing it better in house. However, you can’t yet run title A/B tests in YouTube so you’ll likely still need one of these if you want to do that.

What does this all mean?

When every YouTuber has access to this tool, YouTube is going to become an absolute sea of thumbnail tests.

Here are some high level thoughts for what this might mean for you and YouTube in general:

1. The overall standard of thumbnails on YouTube will increase. Naturally, testing will improve the average quality of thumbnails across the board. This will make it harder to stand out, but potentially level the playing field between bigger and smaller creators.

2. It would be silly not to take advantage of this new feature if you can, even if you have 50 subscribers. Famously, you can trap great content behind a terrible thumbnail, so this could hopefully save some videos and therefore save some YouTubers.

3. Some YouTubers will become paralysed by the thought of having to make 3 thumbnails instead of 1, and others will see it as a relief.

4. Thumbnail designers will suddenly have a lot more demand, but individual thumbnails may go down in price, as everyone’s reminded that a) it’s hard to predict what will perform beforehand and b) making great content is ultimately what matters.

5. Everyone will want to test thumbnails on every video, but only some will be able to afford outsourcing it.

6. When everyone realises how wrong they can be about a thumbnail, thumbnails will be seen less as something that have perfect guidelines and rules, but more as an ever changing art form.

For now, there’s not much you can do but try to improve your thumbnails and consider working with a designer if you can afford to.

I hope that was helpful. Have an epic week making videos!

Tintin 🧑‍💻

The YouTube Tin

🎙️ Podcast: I’ve been loving The Billion Dollar Creator podcast. It’s full of incredible information and insights.

🎥 Channel: The Creator Insider channel often has a lot of useful content on it that helps YouTubers understand YouTube. It feels like an underutilised resource.

🚀 My own channel: I have a low lift YouTube channel so I can share more interesting things that I’m learning about YouTube to help you guys out. This latest video is about 5 common mistakes that beginner YouTubers make.