Two years ago, the launch of YouTube’s video fingerprinting scheme was viewed as an overdue attempt to appease copyright holders like Viacom, which had sued the site for more than $1 billion earlier in 2007. Content owners had to upload their entire libraries to YouTube, a company they weren’t sure if they trusted, in order to police its pages for infringing videos. But today, YouTube’s Content ID is a market standard, with every major U.S. network broadcaster, movie studio, and record label using it, including Viacom. More than 1,000 content owners have uploaded more than 1 million reference files to the system, and the majority of partners elect to leave infringing content up and try to monetize it by linking to official content or overlaying it with ads.
YouTube today is announcing it has tied together Content ID with its analytics tool, YouTube Insight. Now, content owners who choose to leave up YouTube user uploads of their content will be able to view stats about each video’s demographics, referrals and engagement. They’ll also be able to view stats about all their videos, official and unofficial, in one place. So for instance, Sony Music could see that the JK Wedding Entrance Dance is its eighth most popular music video on YouTube. No word yet on whether viewers of unofficial videos are demographically different than those who find the official ones — but perhaps the data will be instructive for Sony and UMG’s YouTube-powered Vevo music video site, set to launch in December
How else has Content ID evolved in the last two years? We visited YouTube HQ in San Bruno, Calif. last week, and spoke to Senior Product Manager David King to find out.
First of all, when the Content ID system launched, rights holders weren’t happy that it scanned videos shortly after they were posted to YouTube, not before. The delay was built in to avoid slowing down the upload process more than than necessary — users just want their videos to go live — and also, possibly, to protect YouTube from claims that it pre-screened videos, which could potentially make it more liable if they contained copyrighted content.
But content owners said they wanted all content fingerprinted before it ever got posted. “This was a religious point for some studios,” said King. And so, about nine months after Content ID launched, the entire YouTube infrastructure was migrated into the Google cloud “at great cost,” he said. Now, videos are checked for copyrighted content before they go live.
However, over the last two years, Content ID has been naturally migrating from an anti-piracy tool to a marketing and monetization tool. From the beginning, rights holders had the ability to leave unauthorized uploads up and monetize them with ads. The majority of partners now do so, said King. Those who leave user videos up have seen their overall views more than double.
On the flip side, Content ID is now being used to control videos on a more granular level. Content is increasingly geoblocked, said King, so for instance, something that was uploaded in France could potentially be unavailable there because of local rights issues, but viewable in the rest of world. Gets a little awkward when you were just hoping to show a video to your pals!
The other big infrastructure change YouTube has made is a fast-track reference system. As YouTube has grown, King’s team has actually decreased the amount of time it takes creates to create a match between uploaded content and a reference file. The site also has to constantly do legacy scans of its entire library as it intakes new reference files from rights holders. Where the underlying database used to be rebuilt three times a day, now it’s a RAM-based system instead of hard-disk based, and can be rebuilt entirely in under 30 minutes. That upgrade was done in advance of last year’s Beijing Olympics, when NBC took a particular interest in taking down content that had just aired live so it could drive the U.S. audience to its own site. “We’re continuing to focus on closing that live gap, and we will continue to have more announcements [about it],” said King.
So now that YouTube has put all this work into Content ID, would it want to license it out to other sites? “We’d be interested in opening it up to other people,” said King, but he didn’t have anything to announce yet.