peshkova - Fotolia

How Netflix uses emotional analytics to improve CX

Emotional analytics enable businesses to obtain customer feedback without words. Netflix employs this tactic as a means to engage and retain customers.

Over the last several years researchers have developed emotional analytics tools that use AI to correlate facial expressions, voices and text with emotions; and while it's still early, some organizations are starting to see value in using it to improve CX.

One such organization -- Netflix -- uses emotional analytics to craft a better customer experience by efficiently shortening movie trailers for consumers to evaluate movies more quickly, with a goal of retaining customers by putting movies in front of them that they enjoyed. Previously, the company used long-form trailers provided by movie studios, but this did not translate well to online attention spans, said Thales Teixeira, co-founder of, a user experience consulting firm, who worked with Netflix on the project.

"The challenge was that as Netflix put more trailers online, consumers would get overwhelmed and did not know what to watch," he said.

Crafting a better preview

Traditionally, Netflix relied on word of mouth or brief descriptions to help people decide whether they would like content. However, this only provides limited value, in the sense that reading a description of a food dish will help you predict if you like it or not, Teixeira said. There is nothing like a brief sample to help you make up your mind.

Thales TeixeiraThales Teixeira

Netflix also created a recommendation engine to help prioritize the order in which suggestions are shown to consumers. But it's still up to the consumer to predict whether they will like the movie from a limited bit of information.

When Netflix started experimenting with smaller clips to make it easier for customers to decide if they would enjoy a movie, they initially took just the first couple of seconds from the movie studio trailer, but this led to suboptimal results, Teixeira said.

Filtering the good stuff

So Netflix used emotional analytics to try to help them shrink clips in a better way.

"Our original idea was that people watch movies to feel certain emotions," Teixeira said.

Netflix uses emotional analytics to craft a better customer experience by efficiently shortening movie trailers for consumers to evaluate movies more quickly.

At a high level, the plot is a delivery mechanism for experiencing happiness, surprise, fear, sadness and even anger. Netflix hit upon the idea of using emotional analytics to identify when people felt more of a particular emotion such as happiness in a comedy, sadness in a tragedy or fear in a horror movie and recruited customers to watch the trailers and record their facial expressions.

This worked out well for comedies, as people spontaneously laugh or smile in response to a humorous scene. Researchers were able to identify the moments in the comedies that elicited the greatest laughter, which could be filtered down into a shorter trailer.

Context is everything

The idea did not pan out as well for other kinds of movies. People don't reflect intense fear, sadness or anger in their faces when they watch ads and shows on the screen, Teixeira said.

"It's not because the technology is bad -- it is a matter of context," he said.

Netflix found that these other types of emotions take more intense stimuli or more time for people to spontaneously express them on their faces.

At the beginning of every technology there is a tendency to overhype its value, Teixeira said. With emotional analytics, early proponents suggested the technology would make it possible to film virtually anyone to know how they are feeling and then influence or persuade them. In Teixeira's experience, emotional analytics is useful within a limited context, such as understanding the virality of an advertisement or understanding how people feel about an internet experience.

The core idea for emotional analytics builds upon research by psychologist Paul Ekman that identified common patterns of facial expressions associated with given emotions. Software tools can then help translate and record these emotional expressions as people are exposed to a particular experience.

This used to be done by hand, but now researchers can record changes up to 10 times per second, Teixeira said. With emotional tracking, researchers can get more fine-tuned data about a customer's experience during an ongoing ad, movie trailer or internet experience.

Next Steps

Lawmaker digs into emotional recognition technology

Dig Deeper on Customer experience management

Content Management
Unified Communications
Data Management
Enterprise AI