Want to learn to grow your YouTube views and subscribers? Book a 15 min call with one of our specialist's today

How to beat the YouTube algorithm according to the YouTube product team and MrBeast [2022]

How the YouTube algorithm works, according to the YouTube product team and MrBeast [2022]

Share This Post

MrBeast shared how his team optimised each video element for the YouTube algorithm to grow to 70 million subscribers and just how much time he spends on testing and optimising. 

Todd Beaupré, who leads the Discovery product team at YouTube pulled back the curtain on how YouTube works behind the scenes with his talk “YouTube Explains Its Recommendations Algorithms”.


Speaking at Vidsummit, MrBeast AKA Jimmy Donaldson is one of YouTube’s top creators with over 70 million subscribers.

How to make better thumbnail designs for YouTube that get you views

“If a video doesn’t get X amount of views in the first minute, then that’s a title problem.  If it doesn’t get a certain number within the first 15 minutes, usually that’s a thumbnail problem” – Financial YouTube personality Graham Stephan (3.7m Subs)

Throughout the Vidsummit conference, leading YouTubers spoke again and again about how vital the video thumbnail is for growing views. 

The video thumbnail, the little image that gives you a peek into what the video is about – is a vital and powerful visual cue. It is critical that creators spend time choosing the right thumbnail for their videos or viewers will simply skip past them and on to the next one that stands out. 

The way the YouTube algorithm works with Thumbnails is pretty straightforward.  When a user searches for a query or keyword. The more people that click on a video thumbnail (and it is usually the images that catch a viewer’s attention before the title) tell YouTube that the video is the best match or answer to the query. So it moves up the page results.

The urge of course is to begin using clickbait-type thumbnails and titles – deceptive imagery that doesn’t really represent the video – to get their audience to click on it. But this only ever generates short-term wins and sacrifices long-term growth. Once the YouTube algorithm starts seeing people abandon the video, it will push it down the rankings again.

There are plenty of websites available to help you to create an engaging, professional-looking thumbnail to get people interested, so make sure that you use these brilliant free resources. When creating the thumbnail itself, making sure that it’s a great match for your brand and what you’re trying to achieve is critical to getting it right. Pick an engaging photo (it doesn’t even have to be a still from the video itself) and one of two short, sharp, engaging quotes or descriptive sentences that will reel the viewers in.

As Casey Neistat often says, ‘Intrigue, Don’t Deceive.’

Mrbeast backed this up pointing out that “the thumbnail needs to accurately describe the content within the video”, he said. “If a video delivers what it promised in the title and image, then it will likely have a higher watch time.”

MrBeast probably needs no introduction as one of YouTubes top creators. He now has 6 YouTube channels with some individual videos achieving over 100 million views each.

But he pointed out that achieving viewership like that has almost as much to do with the video thumbnails as it does with the content of the video itself. 

“Obviously, no one will watch your videos unless they click on it,” said Jimmy “So, how do you increase the percentage of people who actually click on your video?”

He and his team spend an inordinate amount of time getting the thumbnail right and then testing it. They will often recreate a key scene from the video and re-photograph it, with improvements. “Simple, visible, colourful.”

They will then create multiple versions of it to review and test. 

https://lh6.googleusercontent.com/D8GcI89GtU2nlBOBFLhhpAs30XWSdSJ6JzrrT29Wv2l-TYMUlF0etsGQia1KqXWOcz8X2J6NDYZ5Tnm_7deNXQTBw20n_Q6zwfIbtIpNEJeAv1WExT-1nkbX5HoHFZHB7gs9xGp7MrBeast on stage at VidSummit. Image c: @U_Do_It

Testing first involves placing the thumbnail in a YouTube page mockup in Photoshop to see if it stands out.  Then the team typically chooses two to three versions of a thumbnail so that if a video does not do well, they swap it out.

“Majority of your views are probably coming from mobile, so it’s important that you see exactly what the thumbnail will look like for people when they watch on mobile.”

The thumbnail needs to reflect what is happening in the video. If a YouTube video delivers on what the thumbnail promises then you will likely have a higher watch time. (How much watch time does YouTube look for? Todd Beaupré from YouTube shares that data below.)

Then when they put the video live, if it does not receive the expected views within the first few minutes they will swap out the thumbnail image.

“I don’t care if we spend up to $10,000 making a thumbnail, I just want the best thumbnail possible,” said MrBeast.  “The thumbnail literally decides whether or not they watch your video,” he added. “For us, it does make that big of a difference. It makes a difference of tens of millions of views.”

So to recap:


MrBeast’s strategy for YouTube thumbnails

*Thumbnails need to accurately describe the content within the video

*Ensure the focus of your thumbnail is right. (Jimmy typically uses himself as the focus).

*Recreate the image if you need, to make the background simple, colours pop, and to ensure it’s easy to understand from the thumbnail what the video is about. “Simple, visible, colourful.”

*If the video is not getting the expected views and ranking, switch the thumbnail.

*Test different thumbnails.

*Preview how the thumbnail will look like among other video results (In Photoshop or similar). 

*Don’t forget to preview on mobile as that is where the majority of viewers come from.

YouTube Explains Its Recommendations Algorithm

Todd Beaupré, who leads the Discovery product team at YouTube, is responsible for the home feed and recommendations. His talk, “YouTube Explains Its Recommendations Algorithms,” covered the principles and strategies behind how the platform recommends videos and explained critical factors that influence these recommendations.

Todd brilliantly and succinctly explained what an excellent average view watch duration is, helping his audience better understand the concept. 

What is interesting about this YouTube data is the flattening of the curve as video lengths get longer. This is probably due to podcast and longer video viewers going in with the expectation of them being long. So retention stays high. It indicates that Videos do not need to be short. But just as short as they need to be. And you set the viewer’s expectations.

YouTuber Zack Freedman backed this up from his own experience:  “Seems right. I average 50%-ish no matter the video length”. 

As did Vidsummit founder Derral Eves in his Keynote case study. He noted that  @drstenekberg had a high average video view of 57%. Even though Dr. Ekberg’s videos were regularly 1.5 hrs in length.

Longform videos can be just as effective at keeping audiences listening and interested when done right. Creators must make sure to not shy away from long-form videos as they can still pull in significant amounts of ad revenue, simply due to the number of ad reels that you can put into a longer video, as compared to the shorter ones.

An inside look at the YouTube ranking algorithm

Todd took us behind the curtain at YouTube and opened up about what his discovery product team and the algorithm look for when deciding to rank your video.

What factors influence how videos are recommended?

#1. Personalisation:

Everyday, What the YouTube algorithm learns from 808+ viewers signals

  • What they watch
  • What they don’t watch
  • How long do they watch
  • Where they watch
  • When they watch
  • Likes and dislikes
  • Shares
  • ‘Not interested’ feedback
  • Pre and post watching surveys
  • and more

https://lh4.googleusercontent.com/Dhtda2me3jVmI0pppB6NvrOIkYc7EJE_dUPeEK5E-Hefn260usPh_OFYqUTDo4uHobcNG1uOcKDlvWDJsjfyRaf3KdcRdOLYSyeQlMtUgl66Djizc1ZE42AOv52-AY3RMA5yZu01Images c: @JayJayGhatt

#2. Performance:

How does the YouTube algorithm look at your video performance?

“Example of Google trends, which you should be looking at to determine habits. This is something I used for TekThing and Tekzilla, and I continue to use every week” – ????Shadow???? Morse @Snubs Content Creator ➳ Morse Code.

#3. External factors:

Todd explained how they look at external factors like Google search trends to influence what the YouTube algorithm should be prioritising. We show you how to use Google Trends in this video here.

https://lh4.googleusercontent.com/cD1-P0pf8N_3cwHaal4fTw0fiR-Frw3K70Lt9TyNHReYmLPFt5-lbrpMPdj-u67nZrA62A_TS6zJHhqUpX9spI5z0RSMfKt1WEDfe8uXeWEG99TIBxHFKvkmyr_WBoY1b5eUN7UfTodd Beaupré of the YouTube discovery product team. Image c: @Snubs

Seasonality: How many viewers are online watching YouTube changes during different times of the year?

  • Recently, we’ve seen fluctuations in traffic related to COVID, which dramatically changed people’s viewing habits.
  • Many creators also experienced drops in views in mid-August when their viewers go back to school. Often accompanied by spikes at the weekend.

Remember your casual audience and the value of evergreen content.

https://lh4.googleusercontent.com/VjQNwtESzFXmej-ELSt15mpZtooblaIQ-R3cJ1daGXqiStBouJfPf1CkCecoQK2wQNKfc5MHwvOHSvOjIdwOyrQRSPQKBrxED7CiH3yCCVEPBPfXw_kG6hLkqD1204TVgwQUtcwrImages c: @JayJayGhatt

External factors:

https://lh4.googleusercontent.com/cinOpvUPga5QGdh8yNVGSpNbWE71UpLCS9xRmw_ciVtbXdsib4YpHf9JtNJCf5HxFGi84IVJ8dCY83mImQ7Z92XGTCCwBv1VQlErC9Qy8WJHpMLFosbYXs2HzYvNnpjDk_D6irrgImages c: @JayJayGhatt

How often should you publish to YouTube?

Upload at a cadence that works for you and your audience.

  • Compare your “Videos published” stats to the “average views per viewer” stat in YouTube analytics over the last 28 days.
  • Find the right page for yourself.
  • Balance quality and quantity.

https://lh5.googleusercontent.com/5aMbCoxb2n_DA_7UkMQjDOC9cJgS2DZ_z431Q5Yu-Drpg-D98mtmjHjwfIgsL6R6n0qdkLab_6B2CvPjNqyIJU9pKxqeKro-UievHgydiAn9DirXabFPpD2T14hKXBk8Q0XT5Me7Images c: @JayJayGhatt

“Todd Beaupre responding to a common YTer complaint YT isn’t showing their videos, usually bc a sub said so, analysed thousands of channels & determined for a typical channel, if 100 viewers saw your video last week, 700 potentially will see it this week”

Todd brought up common YouTubers complaining about YouTube not showing their videos, often because a subscriber said so. 

So Todd and the YouTube discovery product team analysed thousands of channels and determined for a typical channel, if 100 viewers watched a video of yours last week, then 700 people will see one of the thumbnails of your video this week, which is a fascinating statistic and demonstrates the potential growth that is possible when you optimize your video for clicks and engaged views.

Summary:

Growth opportunities abound!

The algorithm follows the audience. look through the viewer’s eyes.

Upload at a cadence that works for you and your audience.

Consider how you might convert more potential and casual viewers.

Consider the long-term value of your content.

More To Explore