Canvs Raises $5.6M for Platform to Map Emotions via Social Media
Claiming to know how people, especially millennials, feel based on what they say through social media sounds a bit like a myth marketers might whisper about.
It is easy to see when someone says they like a television show for instance, but what in particular did the audience react to? What was the funniest line in an episode?
New York-based Canvs says its platform can decipher such reactions. “We’ve created the first accurate measure of how people react to content, integration, and advertising at scale using social data,” says CEO and co-founder Jared Feldman.
The company announced Thursday it raised $5.6 million in a Series A round led by KEC Ventures, with Rubicon Ventures, BRaVe Ventures, Social Starts, Milestone Venture Partners, and Gary Vaynerchuck participating. Feldman says the new funding will go towards research, product development, and marketing to get the platform more in the spotlight.
Plenty of software exists to capture what people rant about online. There is a difference, by Feldman’s reckoning, between what Canvs does to understand the public’s emotions and “sentiment analysis,” a phrase tossed around by some companies that track social media comments.
Canvs takes Twitter data from Nielsen and analyzes tweets about first-run television shows to map out emotions, divided into 56 categories. These social media reactions are linked to specific moments in the broadcast, providing a view of how the content made viewers feel The company has some 30 customers, Feldman says, who are mainly television networks, content creators, and advertisers, which include NBCUniversal, Viacom, Starcom MediaVest, and Sony Pictures.
The process is not as simple as seeing if someone follows a show, Feldman says. For instance, a comment meant to be sarcastic could be misunderstood because of the way it was phrased. Further, the proclivity of some folks to misspell—as well as totally misuse—words online complicates the job even more.
“If you look at a show like Teen Wolf, and run it through state of the art sentiment analysis, about 65 percent of the emotions will be completely missed,” he says, due to misspellings, slang, emojis, and a general lack of standard sentence structure in social settings. “The millennial generation essentially doesn’t speak properly” especially online, he says. “We needed to design something that actually captures how people talk.”
To better understand the complexities of what people say via social media, Canvs grades conversations about TV shows in 56 categories for emotion, such as beautiful, annoying, and love, rather than shoehorn reactions into simplistic positive, negative, and neutral results.
Back when Canvs was called Mashwork, its original name as a social media insights business, the company tried taking the pulse of the TV audience for networks by using third-party tools, Feldman says. “For four years, we struggled with how to answer the ‘How did I do last night?’ question,” he says.
General buzz surrounding shows was easy to understand, he says, but getting a detailed picture of audience reactions could be costly with older methods. “The only way to do that was to spend $30,000 to get 20 people in a room, make them watch the show, and figure out what they were responding to,” Feldman says.
In an era where social media has democratized the ability to express oneself, he believed there could be a better way to figure it all out. That led to the development of the Canvs platform, first for internal use, and then in beta for a small group of customers.
Feldman claims Canvs can predict, with an 85 percent confidence rate, whether or not a show will be renewed based on the audience’s emotional response to the first three episodes of a new season. He sees the platform as a proxy for the general audience, and claims it can modernize market research through social media. “Every content creator, every advertiser ever has wanted to make sure that what they’re doing resonates emotionally with the person on the other side of the table,” Feldman says.