This audio is auto-generated. Please let us know if you have feedback.
YouTube is testing a way to help people avoid being misled by video thumbnails. The platform is looking at providing short preview snippets of recommended video clips, which will better highlight the actual content before users click.
So now, rather than relying on random images of YouTubers with the most exaggerated, ridiculous expressions overlaid over some artificial intelligence generated thumbnail image, users will be able to actually get a glimpse of what’s in the clip before they tap through.
As explained by YouTube in a March 19 support update: “We’re experimenting with a new ‘Discover videos with Previews’ feature. After clicking the entry card on your homepage, you’ll see 5-10 previews featuring short, engaging moments from videos that are already recommended for you. You can then watch the full video, add to your ‘watch later,’ and more.”
The system will essentially choose a highlights package to show to potential viewers, in order to provide more context of what they can expect. The change could also help viewers avoid being sucked in by exaggerated thumbnails.
Which could be bad news for creators, who rely on thumbnails to maximize their view counts. YouTube star MrBeast, for example, claimed to have a team that works on optimizing his thumbnail images, creating up to 20 images to choose from for each clip, and running a range of experiments to decide on the best option.
Indeed, MrBeast’s Jimmy Donaldson estimated that he pays around $10k per thumbnail image through this process.
Given that some creators are putting so much emphasis on this, a move to re-align YouTube viewers around a different selection process could be costly for creators, and could force a big change in their approach.
But for viewers, it may provide a better pathway into finding relevant content, and for catching key moments as highlighted by YouTube’s system.
YouTube said that this is a small-scale experiment for now, with the highlights currently available only to a small percentage of users on the Android mobile app.