Emotional data analytics at scale

The problem with measuring emotions

Given its importance, you would imagine that detecting, measuring, and understanding emotions as they occur in others is as straightforward as ABC. However, capturing and quantifying authentic emotions has always been a challenge and has posed one of the most vexing problems in behavioral science.

Conventional surveys and Biometric measures

People hate surveys. We are ALL over-surveyed. Digital Natives—the key to future growth—don’t like to answer questions. Their world is fast-paced, and conventional consumer surveys don’t work for them. The New York Times described it as “poll fatigue,” often leaving respondents feeling inconvenienced and distrusting that pollsters’ intentions are disguised sales attempts. In addition, ask respondents how they feel and you trigger a series of defense mechanisms that can lead to inauthentic and inaccurate responses.

Conventional surveys and focus groups that ask panelists to describe their emotional experiences are hampered by challenges of scalability and are violating a central precept: feelings are gut-rational responses that are authentic only when they require little deliberation and reasoning and are experienced in the right context.

In addition, biometric and AI-based technologies (such as facial coding), violate people’s right to privacy and are often grounded in flawed or incomplete theories of emotional communication.

Conventionally, emotional classification uses one of two methods: Sentiment or Discrete. In the Sentiment method, emotions are classified as either “positive”, “neutral” or “negative.” An issue with this method is that it is hard to achieve universal agreement as to what a “positive” vs “neutral” emotion might be, and what feelings fit into each bucket.

Sentiments are an expression of opinion or a mental attitude toward a subject matter, be it a brand, product, service, or concept. They provide an overall impression on where an individual stands, with ranges from positive to neutral to negative. A consumer might, for example, state that they “like” a product (positive sentiment) or “dislike” an incumbent (negative sentiment), or “have no opinion either way” (neutral sentiment). Sentiments may be informed by emotions, but do not offer the necessary granularity to help explain with any degree of depth what elicited that sentiment.

Discrete Emotions are constructed from a consumer’s cultural, experiential, and body management systems. They are complex psychological states that involve subjective experiences accompanied by physiological and behavioral changes. They are multidimensional in nature, capturing both valence (i.e., whether the experience feels positive or negative) and degree of arousal (i.e., how intense the experience feels). Annoyance may thus be considered to have a negative valance with a relatively low degree of arousal, whereas rage, also a negative valence, has a high degree of arousal. Both express negative sentiments, but both permit insights into the nature of the sentiment and its possible origins. The discrete method allows emotions to be perceived in defined, and more readily objective categories (anger, fear, disgust, sadness, and enjoyment).

Only in shifting from Sentiment to the Discrete emotional analysis can brands discover a new layer of actionable insights that are critical for a marketing and business advantage.

Emotions are more actionable than sentiments. They provide the building blocks for optimizing trust and empathy among consumers. They give you a richer, more dimensional picture of how consumers are truly feeling about your brand and communications. Sentiments represent low-hanging fruit. Knowing whether your brand community is experiencing calm, happiness, or excitement when engaging with you, versus a mere positive sentiment, enables insights with far greater depth and clarity to inform action. Problem is, most CMOs and brand managers don’t know the Emotional North Star bonding their most loyal customers to their brand. Why? Because emotions are notoriously difficult to measure authentically. If you don’t know how people really feel about your brand, how can you strengthen the emotional bonds that generate loyalty, engagement, and buy-in?

Emoji are “The world’s fastest growing form of communication”. While women use emoji slightly more frequently than men, emoji use is use spread almost equally across age demographics.

Researchers at Harvard Business School, following research involving two studies and 1325 people (average age 33) reasoned that emoji can serve as a proxy for emotional expression in digital communication, due to their frequent usage among diverse age groups and populations and around the world. Nearly 3-in-4 Gen-Z respondents agreed that they find it easier to express their emotions using emoji.

In a survey conducted by Harris Poll and commissioned by GIF platform Tenor, 68% of millennials stated that they were “more comfortable” expressing emotions through emojis than through the written word.

"We suspect and suggest that emoji can serve as a novel and ubiquitous means for investigating the psychology of emotional expression at scale.” (Harvard Business School, 2018)

Users' understanding of the emotions communicated through various commonly-used emoji is also remarkably consistent. In these Harvard studies, 13 unique scenarios achieved at least 93 % emotional agreement between participants. And in 2017, a project spanning 1000 consumers in the USA and China across 5 surveys and 33 different emoji, found that when facial emoji replaced words in a questionnaire (product survey), the emotional data profiles on product preference were valid and repeatable.