Back to Research Overview

Canvs’s Emotional Reactions more accurate and more fluent in social vernacular than traditional sentiment

Millennials have a unique perspective, and they’re not shy to share it. In fact, 56% of Millennials that use Twitter say it is intended for sharing information, and they use it to chronicle their lives in real time. Most Millennials that use Twitter also agree that Tweeting about events makes them more enjoyable, and they are more inclined to follow or contribute to conversation if event-related hashtags are present (1).

Forbes states that Millennials could account for $1.4 trillion in spending, or 30% of total retail sales, by 2020 (2). And according to YuMe’s 2014 Mix & Measure Report, 92% of Millennials access a connected device while watching TV (3). Quickly changing the buying and TV-viewing landscape in America, it is no surprise that studying the social media behaviors of Millennials now is extremely valuable.

Accessing Twitter TV data from Nielsen, Canvs is the only qualitative solution to provide accurate measurement of Emotional Reactions at scale. At our core, we believe that understanding how Millennials emotionally relate to content is the first step in building a meaningful relationship with them. We also know that entertainment content and the unique way in which Millennials emote are much more complex than the positive-negative-neutral paradigm adopted by most sentiment analysis tools. Our emotional lexicon, therefore, is composed of countless permutations of real ways Millennials speak.



To that end, we used MTV’s Teen Wolf, a series popular among Millennials (4), to conduct an experiment. We compared emotional conversation to the Season 5 finale, as captured by Canvs, against an industry-standard implementation of the SentiWordnet Sentiment Analysis built on top of the WordNet project from Princeton University.

We found that sentiment analysis was unable to understand 67% of Millennial’s Emotional Reactions to the Teen Wolf finale. Unlike Canvs, sentiment analysis offers a simplicity of analysis and is not able to keep up with the ever-evolving pace of Social TV vernacular. Our results concluded that emotional analysis, as measured by Canvs, understood more accurately how Millennials actually feel than traditional sentiment analysis, validating Canvs as an authority in emotional analysis.


Key Findings


Canvs was 200% more accurate than sentiment analysis at parsing through how Millennials felt about the Season 5 finale of Teen Wolf.


If you take 100 Emotional Reactions as defined by Canvs, sentiment analysis missed 55, miscategorized 12, and correctly identified 33.


“OMG” and “WTF” accounted for more than 13% of all Reactions to the S5 finale of Teen Wolf, which sentiment analysis altogether ignored as indicative for analysis.


From misspellings to slang, not understanding the complex nuances of the English language used by Millennials to express their emotions means not truly understanding Millennials. The unique vernacular that Millennials use — along with their penchant for social media expression — is one of the distinguishing attributes of this generation. Because of this, Canvs is uniquely positioned to not only comprehend Millennials, but their emotional investment to content, as well.

(1) Moy, Aaron. “Four insights about millennials on Twitter.” Twitter. 09 July 2014.
(2) Shin, Laura. “How The Millennial Generation Could Affect The Economy Over The Next Five Years.” Forbes. 30 April 2015.
(3) Marshall, Carla. “92% of U.S. Millennials Turn to Second Screens While Watching TV.” reelseo. 05 June 2014.
(4) “MTV’s ‘Scream’ Premiere Grows to 6 Million Cumulative Total Viewers in Live +3 Ratings.” TV by the Numbers. 06 July 2015.


Reach out below for additional information.

Name *