The Search vs. Recommendation Conundrum

"A company here in the UK spent 450,000 quid on creating a video and putting it online" said one speaker on a panel called "Measuring Success" at the Online Video Strategies event in London last week, "but they couldn't tell who watched it."

"Worse yet, they couldn't tell whether any action was taken as a result of watching the video," he continued.

Panelists in this session spent almost the entire hour answering questions from content creators and companies thinking about starting their own online video platforms.

Jonathan Milne, Ooyala's senior director of sales and operations for EMEA, demonstrated how video metrics can be broken down by continent, region, and even city, much like a Google Analytics breakdown for basic web pages.

"Measurement is always an afterthought, whether it's web or video analytic," he said, "but too often videos are posted without any clear sense of what success looks like."

At the same time, Milne pointed out that search is only one of the forms of discovery. The other, which may be growing quite a bit faster than search is discovery via recommendation.

Panelists in a separate session at Streaming Media Europe titled "Video Search and Recommendation: Finding Content in a Thousand-Channel Universe" agreed with Milne's premise that recommendation is a growing force in the consumption of content.

Thomas Dvorak, chief marketing officer with APRICO Solutions, a spin-off of Philips, said that search was a feature highly requested during trial runs of the company's set-top box interface and back-end solution.

"Search was the number one request within the first few days of our testing," said Dvorak, "yet only 5 per cent of the total households used search to find content. We think this is partly due to the innovative way we approach recommended content but also due to the fact that search is less popular in a lean-back scenario like television viewing."

APRICO offers on-demand, live television and over-the-top content, so the company is well aware that its users face a thousand-channel universe, but Dvorak says that the key is personalization rather than really big search engines.

"We let viewers personalize content in to channels," said Dvorak, "such as an action channel or a football channel, and we find that each person in the household will choose to create no more than ten personalized channels each."

Michael Woodley, Doovle's chief technical officer, said his Cambridge University spin-off search company sees granular search as key to video viewership rates.

"It's not enough to just tag a video with a few key words," said Woodley, "when only a portion of the video may be applicable. We see that granular search-being able to go to a direct location within a video-will be key to keeping search relevant.

While agreeing with Doovle, Justin Hayward, MD of Make It Rain, said that recommendation is key to his client base of luxury product and service providers.

"We have content that markets some of the world's finest products," said Hayward, "such as yachts, watches, and sports leisure companies, such as the David Lloyd Leisure Group. These companies rely on word of mouth for sales, so we see recommendations and the social graph as a large part of their online strategy."

By social graph, Hayward is referring to tracking Facebook "likes" or tweets (dispatches from Twitter) or other social sites.

This ability to track what others are saying about you shouldn't just stop at Facebook social tracking, said Stephen Tallamy, technical lead at LocateTV, a division of NDS.

"Just like competition is good in the search engine world, between Google and Bing," said Tallamy, "you also wouldn't want to just rely on Facebook, since there aren't key analytics yet in place-for anyone other than Facebook-to measure success rates. Don't forget that simple comment sections can also help provide engagement with customers."

Nedstat GM Ade Adeosun, of the "Measuring Success" panel, agrees. When asked by a representative from Nintendo about how best to gauge feedback on the video trailers for Nintendo games, Adeoson mentioned social chatter analysis but said it goes well beyond that.

"Don't simply put a Twitter or Facebook link," said Adeosun, "as everyone is doing that. Use comment areas to gauge interest, and consider optional methods that move the user beyond the binary of choosing 'like' or taking no action."

"We measure Flash games," he added, "to see if a level is too difficult, and then pass that information on to the client. They often find that forums and comment sections will reveal some of these frustrations, too."

Milne wrapped up the thoughts on this topic by noting that, during recent UK election results telethon, ITV rapidly gained viewers throughout the night.

"We saw during the elections that one channel lost a number of live viewers to ITV," said Milne. "The reasoning-from what we can tell-was that ITV launched a companion online application that allowed for commenting. This means that tracking the social graph and recommendations are key."

Want to learn more about search, recommendation, and social media? Join us at the Online Video Platform Summit for the session "If You Publish It, Will They Come?," moderated by ReelSEO's Mark Robertson and including speakers from Funny or Die, KickApps, Taboola, and MeFeedia.