How YouTube Is Reprogramming Video and Increasing Watch Time
Engineers Tackle the Paradox of Choice for Viewers and Marketers
YouTube was always about video that was wider than it was tall, the kind you watch in movie theaters and on wide-screen TVs. So once apps like Snapchat and Periscope proved that vertical video could rival or even usurp horizontal formats, YouTube suddenly found itself behind. Its horizontal player created huge dead zones on either margin of vertical videos and shrunk the image to fit the available height.
Then, at a meeting earlier this year where someone was complaining about the problem, an engineer piped up. "'Oh yeah, that's horrible,'" the engineer said, recalled Matt Glotzbach, VP-product management, YouTube. "'I fixed that. I've implemented full-screen support for vertical video. Here, turn on this experiment and you can test it.' We were like, 'That's great! We should launch that!'" YouTube adopted full-screen vertical playback in July.
Meet a crucial new breed of entertainment programmer: YouTube engineers.
Products, not a portal
YouTube has always denied ambitions to become a programmer akin to CBS or Netflix, and it remains an atypical sort today. It is not, for example, arranging a marquee section on its home page to push a flagship series the way a broadcast network flogs its new lineup. But under CEO Susan Wojcicki, its engineers are steadily rolling out products, updating existing platforms and making it easier to connect videos with audiences -- and the advertisers that follow.
Amid the rise of video competitors including Facebook, Snapchat, Twitter, Vessel and Twitch, YouTube engineers have added support not just for vertical but also 360-degree video, created apps dedicated to popular categories like gaming and overhauled the company's primary mobile app to make it easier for people to find videos they want to watch. To the potential chagrin of advertisers, they are also developing an ad-free service for people who would rather pay than contend with pre-roll and overlays.
"That whole idea -- we're going to reallocate our product resources to really getting better at content discovery -- that to me is an example of saying, 'Hey, we need to be better programmers,'" said Chad Gutstein, CEO of digital video network Machinima.
YouTube is pursuing programming in the classical sense to some degree, planning to debut several feature films from DreamWorks Animation's AwesomenessTV and producing original series with YouTube stars such as the Smosh Brothers, who will star in a comedy that has them working at a theme restaurant.
But in a telling moment during an interview at YouTube headquarters, when Ms. Wojcicki was asked whether YouTube will use traditional programming tactics to promote its original content, she said: "That's where Cristos' team comes in." She was talking about Cristos Goodrow, YouTube's director of engineering.
Mr. Goodrow and the search-and-recommendations team he oversees are tasked with tackling a monumental challenge. With 300 hours' worth of videos uploaded to YouTube every minute, people face a paradox of choice. "This is the biggest problem everybody seems to have on YouTube: We have a billion videos, and you can't figure out what to watch," Mr. Goodrow said.
It has become one of YouTube's biggest vulnerabilities. Instead of digging around YouTube to find something to watch, people can pull up their Facebook feed and let their friends make the choice. Or they can opt for more category-specific video services like Netflix and Hulu for TV-style programming; Vessel for early access to some of the top YouTube stars; or Amazon's Twitch for gaming videos, one of YouTube's most popular categories.
"Much like how YouTube became so popular by becoming a destination where a lot of different types of content existed that didn't exist anywhere else, you start to see a lot of other places crop up, with gaming probably the biggest success story so far with Twitch," said Adam Shlachter, chief investment officer at DigitasLBi.
But now people can also tune in to YouTube Gaming, a mobile app introduced last month that's home to only gaming videos and live streams.
"We want YouTube to be an experience where you can go and find any kind of video for education or for entertainment," Ms. Wojcicki said. "But if you have a specific need or a specific interest like gaming, for those users you can have a full-featured gaming experience too." She cited the more prominent placement of livestreams within the app as one example of a category-specific feature.
YouTube has similarly rolled out a category-specific app for kids and is testing one for music content, though Ms. Wojcicki made a point of saying that YouTube isn't looking to spin off any or every category into its own app.
"I don't think it makes sense for us to have specific apps for small verticals or for all verticals," she said. "It really makes sense when there's a use case where there's a lot of usage and the app enables a different or better kind of experience." In the case of gaming, not coincidentally, it also gives video creators a reason to keep posting to YouTube instead of migrating their audiences to Amazon's Twitch, which is primarily home to gaming videos and live feeds of people playing games in particular.
Indeed, despite Twitch's rise and billion-dollar sale to Amazon last year, YouTube doesn't appear to have suffered an exodus of gamers. Instead, there are more new gaming channels on YouTube this year than any other category. From the start of 2015 through the end of August, 15,788 new ad-supported gaming channels -- including branded gaming channels -- popped up on YouTube to total 107,583 gaming channels, according to digital video ad and analytics firm Outrigger Media. By comparison music is the second biggest category of YouTube channels at 36,906 channels after adding 7,297 new channels since the start of the year.
"There's been an explosion of new ad-supported channels on YouTube in the past year," said Outrigger Media CEO Mike Henry.
Making brands safe
That explosion only adds to the pressure on YouTube to make sure those channels find audiences and advertisers. Efforts to that end may be Ms. Wojcicki's biggest legacy so far in her 19 months running YouTube.
In addition to the category-specific apps, YouTube, under Ms. Wojcicki, last year began an ongoing campaign code-named "Project Beacon" to promote its creators with ads on TV, in subway stations and in print magazines; created the Google Preferred ad sales program to pitch advertisers on the top 5% of YouTube channels across roughly a dozen categories; and given a handful of creators money to produce original programs it will premiere.
"When you see them creating YouTube Gaming and starting to say this pocket has different needs, in the last year and a half that she's been in her job, there's definitely been this shift to putting more resources into being a programmer and not just a platform," Mr. Gutstein said.
Google Preferred in particular has addressed a sore spot among advertisers that is the marketers' version of viewers' paradox of choice. Brands know there are billions of videos on YouTube that more than a billion people are watching each month, but they don't know which ones to advertise against.
Google Preferred introduced a level of transparency about where brands' ads would run that YouTube hadn't previously offered, said Mr. Shlachter. That complemented work done by YouTube networks like Maker Studios or Fullscreen as well as companies that bought, repackaged and resold YouTube ad inventory in ways brands appreciated, the said.
While YouTube was focused on targeting ads to specific audiences, brands' ads could just as easily wind up running alongside a clip of someone's cat as a YouTube star's polished video. The YouTube networks and third-party resellers helped by creating TV-like packages of channels that could ensure the type of content a brand might be associated with.
A year after rolling out Google Preferred -- which buckets the top 5% of YouTube channels into roughly a dozen categories -- YouTube has booked three times more commitments from advertisers with this year's iteration than a year ago, Ms. Wojcicki said. The number of brands running video ads on YouTube has increased by more than 40% year-over-year, she said, and YouTube's top 100 advertisers have increased in their average spend per brand by more than 60%.
Changing the measure
But the redesign of YouTube's mobile site and apps may be just as important as any of that other work.
In an effort to get people more quickly to the videos they want to watch, YouTube decluttered its mobile design in July by splitting the interface into three main tabs: a home tab with recommendations of videos to watch based on what you've previously watched; a subscriptions tab listing the most recent videos uploaded by the channels you subscribe to; and an account tab with the playlists you've created and videos you may have uploaded.
"As a user you may open the app knowing exactly what you want -- in which case you could either search for it or go to your subscriptions to look for that content -- or you can open the app and say 'Hey, I have 20 or 30 minutes, I just want to watch video content' and you look at your home page and see what we're recommending for you," said YouTube's product management director Manuel Bronstein.
The goal of the mobile redesign, and really of anything the company does these days, is getting people to spend more time watching videos on YouTube. For the first half of YouTube's existence, the service's currency of choice was the view count. But that changed soon after the company hired, in 2011, an engineer from Google named Jim McFadden, who built a machine-learning system that was able to ingest the types of videos people were watching and automatically find and recommend similar videos in order to maximize view numbers.
Mr. McFadden's machine-learning work was a great success in driving views. View counts increased roughly fivefold, Mr. Goodrow said. But then YouTube realized the figures were inflated. YouTube's recommendation algorithm was pushing videos to people that they were likely to play because the headlines or thumbnail images were appealing, but then the actual videos didn't live up to the hype. YouTube was pushing click bait.
"The system was causing that because we told it more views was better, so it kept pushing videos that people would view but not watch," Mr. Goodrow said.
So in 2012, in another engineering feat that would influence its programming, YouTube adopted a new yardstick based on the amount of time people spend watching videos on it, which the company refers to as "watch time."
YouTube now monitors how much time people spend watching videos on the service per visit as well as over a 24-hour period. It also looks at how long people spend going down the proverbial rabbit hole, which YouTube calls a "trail," in which they start watching one video and come across more videos to watch until "you've decided I've had enough videos about the VMAs and now I want to switch back to NBA highlights and do a search or return to your home page," Mr. Goodrow said. To extend these trails, YouTube includes suggested videos adjacent to the video currently being watched. "Those drive a large portion of the viewership on YouTube, and making them better has contributed a lot to the overall increase in watch time on YouTube," he said.
Watch time on YouTube has increased by 60% year-over-year, and the number of people watching videos on YouTube has increased by 40% year-over-year, according to Ms. Wojcicki. And in what may be a measure of Mr. Goodrow's team's success or the impact of Facebook's news feed favoring its own native videos over YouTube links, or both, the number of people visiting YouTube's home page to start a session has increased threefold year-over-year.
There is one problem with YouTube's emphasis on watch time. If YouTube prizes watch time so much more than view counts, why doesn't it display watch time statistics next to a video like it does view counts? Mr. Goodrow said that company has "more than just discussed" that idea. "We wrestle with this exact issue."
One hesitation is that view counts have become a popular metric for the ecosystem built around YouTube, if less so now for YouTube itself. Another is that YouTube would need to figure out how to display watch time in a way that immediately meant something to people. YouTube does report watch time, but only on the desktop site, hidden in a separate tab, three clicks deep.
Programming for ads
YouTube is also doing things to increase the amount of time people spend watching ads. Earlier this year, YouTube updated its skippable TrueView video ads to carry interactive cards that people can click on to find information like product details or store locations. YouTube and a gaming brand have even been discussing whether to run an interactive game preview within one of the TrueView cards. "We would have to make the cards a little more interactive, but it's something we're up for," said Diya Jolly, director of product management for ads at YouTube.
The hallowed living room is another area in which YouTube is looking to increase watch time. YouTube has thrice redesigned its connected-TV apps. The first version of YouTube's connected-TV apps "tried to bring all of YouTube content but in a way that was really hard to navigate," said Sarah Ali, senior product manager at YouTube. The second version "was basically like one page and then search," she said, which it easier to find specific videos but harder to browse and find something you didn't know you wanted to watch. The most recent iteration brought search front-and-center as well as recommendations -- but now "in really big tiles."
The work that's been done on the connected-TV and mobile apps appear to be primarily about setting a foundation for how YouTube wants people to use the video service. It doesn't want to only answer queries typed into its search box; it wants to answer the all-important question that people ask when settling down in front of the TV: What should I watch? Or what should we watch? "The ideal future would be to get recommendations for who's in the room," Ms. Ali said.
"The current TV experience is so easy," said Ms. Wojcicki. "You just press one button, and then you have something to watch. So we need to make it even easier to use, so you think about YouTube in the living room context."
Despite a threefold increase in watch time on YouTube's connected-TV apps in the past year, the living room is "a really big and untapped area for us going forward," Ms. Wojcicki said.
The work is far from over. Mr. Bronstein said the mobile redesign will enable YouTube to easily add new features to the app in the future. He wouldn't say which features he was referring to, though he's likely talking about the biggest product yet to be released by YouTube: an ad-free version of its service that's currently in development.
Ms. Wojcicki declined to go into details about the upcoming product, but it is yet another play to extend YouTube's reach as a service rather than a portal. "We have some users that want to have an ads-free experience, and we want to be able to give them a way to do that and for them to pay for a service," he said. "Also creators see value in having multiple monetization models. So having both subscription and ads makes sense for them."