Skip, skip, skip to my video

We are constantly working to find the right ads for the right content on YouTube, so as to create an experience that works for our users, partners, and advertisers. When we first tested in-stream ads in 2007, we learned that abandonment rates (especially for pre-rolls) were as high as 70%, and that users were far more likely to watch and engage with overlays. But over time we found that different kinds of content provide different experiences for viewers, and that in-stream ads work pretty well on certain videos, like clips from TV shows or full-length movies. We first launched in-stream ads last year, and we've continued to innovate and test different in-stream formats on YouTube (like user-choice pre-rolls).

Today, we've started a small test of skippable pre-rolls, which will allow users to choose whether or not they want to watch the ad that appears at the beginning of a video. (If they skip the ad, they'll go straight to the video.)


We know what you're thinking: who would choose to watch an ad when they can skip it? Well, that's what we're trying to find out. In our previous research, we've actually seen that lots of users will watch pre-rolls. Abandonment rates are affected by several factors, notably length and creative. When a pre-roll is only 15 seconds, we see completion rates as high as 85%. Also, creative matters a lot: the quality and relevance of the ad itself seems to have 3x the influence on abandonment online as it does on TV. Viewers online tend to be much more active in making choices about what they watch.

Skippable pre-rolls have the potential to solve this problem and create a win-win-win for everyone on YouTube. For users, this format gives them more control over their experience. For advertisers, we're working toward a solution where they pay for ads that users actually watch and engage with. And for partners, skippable ads attempt to minimize abandonment rates, helping them protect the audiences they've worked hard to build. We've learned from Promoted Videos that advertisers are often willing to pay more money for an engaged opt-in view, as opposed to a forced view like an in-stream ad, so this also has the potential to increase CPMs.

There's a lot about in-stream ads that we just don't know, which is why we're excited about this test. We hope to discover more about what works so we can help lead the industry towards a solution that benefits everyone. We'll let you know what we find out.


Posted by Jamie Kerns (Software Engineer), Lane Shackleton (Product Specialist), and Dan Zigmond (Technical Lead)