Search, Recommendation, and Curation: Attract and Keep Viewers

“Closed captions are gold. If you have that, you can basically find any word the speaker said and you can navigate to that instant in the video with one click,” Burns says. “Our accuracy is roughly 70 percent and it can be higher if your audio quality is very good and your speaker is very clear.”

Of course, there’s the age-old problem of garbage in, garbage out. Panopto and Ramp’s base products use machine transcription, and both companies also have an integrated transcription service that includes human transcription at closed-caption level accuracy, for an additional fee.

Recommendation

Recommendation, also called discovery, is a term that covers how users find content that they might not have known even existed. Users want to see content relative to the search terms they enter, and publishers want to make sure that more of their content gets seen. So rather than requiring users to do another manual search for their next video, most publishers will present recommendations for additional content along with every video they play, based on the video just watched and what’s known about the viewer’s preferences.

Based on your initial choice, the search engine or video publisher knows you’re interested in cat videos. “Everything after that [first video], we’re programming, given that first choice that you made,” says Robert Bardunias, chief revenue officer of Iris.TV. “What we focus on is getting the right content to find the right users. Once that video gets started, we’re taking the rest of the discovery out of the hands of the user and making it more of a lean-back approach through a programmatic algorithm to program for that user based off that initial video cue.”

Iris.TV’s technology lives inside the video player. “We partnered with people like Brightcove and Kaltura so that we’re keeping you in that video experience,” he says. “We’re optimizing experience for the publisher and giving them the opportunity to serve up their best content. For the user, we’re giving them the ability to get content in the path of least resistance.”

Prior to this type of service, publishers would have manually created a list of similar videos, but there’s no way to do that for every single article or page where videos are embedded.

In addition to recommending popular titles, Amazon Instant Video makes recommendations based on viewing and browsing history. Someone in this house is clearly a big SpongeBob fan.

“We’re essentially replacing that methodology in a scalable way where now you can personalize each user, or each device in every category in every instance where there’s a video,” Bardunias says. “Obviously the natural benefit of that is eventually publishers are going to be able to better monetize, but at the end of the day I think it still benefits the user.”

Amazon Instant Video has hundreds of thousands of titles, which makes finding what you want very important, and so the company has created its own recommendation engine, which assesses what viewers want based on their viewing history and other factors.

“Search is a great tool when customers know exactly what they want to watch,” says Michael Paull, vice president of digital video at Amazon. “But ultimately we want customers to come to the Amazon Instant Video homepage and not even need to search—they won’t have to because they’ll immediately see movies and TV shows they’re interested in.”

That’s all based on the data Amazon collects every time you make a purchase or search the site. “We’re also able to zero in and provide recommendations based on customers’ individualized experiences on Amazon,” says Paull. “For example, we have a ‘Related to Titles You’ve Watched’ category, which is specific to each customer and their personal tastes. We have a team dedicated to curation and recommendations. Titles are curated by a variety of factors including Amazon Original Series, Amazon exclusives, new releases, seasonal ties, past viewing behavior, and much, much more.”

In addition to supplying video from multiple sources that content publishers can then hand-curate before publishing, Waywire also has its own collection of curated channels, such as this one for heavy metal.

If you are choosing mainstream content, the Amazon approach will work very well for you, but what if you want something very unique, long tail. Then you might choose to go to Vimeo.

Curation

Amazon isn’t the only video site putting an emphasis on curation. In fact, there are plenty of services that emphasize curation to a much higher degree.

“Our basic thesis is consumers don’t want more video, they want less,” says Steve Rosenbaum, CEO of Waywire Networks, which calls itself a “platform of quality, human-curated, tech-enabled channels.”

“What our users came to understand was there was no shortage of content, but our users want the most recent, most relevant videos around their topics,” Rosenbaum says. “I found myself trying to explain to people there was a different way to organize video, other than simply making more of it. The word I came up with is ‘curation.’ It means the human effort of finding relative related content and putting it together in a contextual way.”

Waywire powers enterprise and consumer sites, allowing clients to publish using its tools inside of a branded wrapper. The curator/ administrator can set up automatic video discovery from any number of sources, including AOL, Yahoo, MSNBC, Condé Nast, YouTube, etc. Waywire runs a search based on publishers’ criteria, but instead of automatically publishing live links, publishers are sent an email inviting them to check out the videos and decide whether or not to make them live. This is where the human element comes into play. The Waywire-powered search is automated up to the last decision point, when a real person can look at the search results and pick the most appropriate content for his or her site.

“I think discoverability is really much more about being where your customers are when they need information, not waiting for them to ask a question. If you’re the editor of a site, you don’t want to go to 15 sites and type in the keyword and find the MPEG code and add it onto your WordPress blog,” Rosenbaum says. “Our technology searches all these open APIs in real time.

“I’m hugely against the idea of automating content,” he says. “When you talk to people that are algorithm people, they tell you as soon as they get the signals right you’re going to be able to produce a relevant piece of content without any human editors at all. I think humans are not irrelevant in the web going forward. I think they are more relevant.”

High-end, user-generated content site Vimeo is an open platform where anyone can upload and share content that also uses human creation to surface the best content.

“So from a user perspective, if you’re looking for the highest, handpicked quality content on the platform, it sits with Staff Picks,” says Mike Weissman, general manager of the creator platform at Vimeo. “When a user signs up to Vimeo, they automatically begin following our ‘Staff Picks’ channel and will see our curated content on their homepage.

“We have an internal curation team of filmmaking experts who know what’s good and where to find it,” Weissman continues. “Each day they select videos to be featured in Vimeo Staff Picks.”

User-generated content site Vimeo has its own selection of curated videos, called Vimeo Staff Picks, which is updated daily.

However, much like any other content hub, you can browse by genre, search by keyword, or follow what others like, which brings us to a different type of curation. Vimeo and any other site that has a social feature will show you what others viewers think is good.

“If you’re looking to follow others with similar tastes and see what they are watching, then that’s a great way too (to find content),” Weissman says.

In the end, search, recommendation, and curation results vary widely. Each will bring viewers to a video destination much more efficiently than if they were left to fend for themselves in an ever-expanding world of video. All these tools and processes are essential in a world where viewers are trying to find the right cat video among 400 million choices.

This article appears in the May/June 2015 issue of Streaming Media as "Field of Screens.”