In the world of beer advertising, emotional attachment is everything. Marketers are often trying to sell consumers feelings—game-day excitement, nightclub cachet, good times with friends—as much as they are the beverage itself. But if the two aren’t intrinsically connected, the advertiser risks producing emotional ambiguity.
That was the dilemma that startup Spark Neuro’s researchers had in mind as they monitored the brain activity, heart rate and palm sweat levels of beer commercial test-viewers recently on behalf of an unnamed major brewery conglomerate. The advertiser had strong emotional creative, but it didn’t seem to generate a thirst for beer. The team’s solution: adding exaggerated fizzing sounds of pouring a beer, which tests showed snap viewers’ attention back to the product.
“If you ever pour beer from a tap into a glass, it’s essentially inaudible,” said Spark Neuro CEO Spencer Gerrol, adding people’s brains lit up with massive emotional responses when those sounds were exaggerated.
Advanced forms of sentiment reading and neuroscientific data collection from entities like Spark Neuro, Affectiva and MIT’s Media Lab have made it possible to build a nuanced and granular picture of the feelings that various pieces of content elicit. Some companies are trying to turn that capability into a creative toolkit of emotional hacks for advertisers, agencies and media producers. Proponents claim that these methods can improve on the types of focus groups the industry has used for ages by picking up on subconscious cues, eliminating survey biases and supplying vastly more data for less money.
Small visceral tweaks like adding fizz to a beer spot are a specialty of Spark Neuro’s. The startup uses a science-fiction-esque headset and wrist attachment to determine emotional reactions to video advertisements and media with biometric data and proprietary algorithms. It’s worked with big brands like Fidelity, State Farm and Walmart as well as major Hollywood studios and streaming services. The company raised $13.5 million in August in a round led by Peter Thiel’s personal venture fund with additional backing from Will Smith and former Disney CEO Michael Eisner.
Gerrol claims that there are all sorts of subconscious quirks in the way people respond to things as seemingly inconsequential as shifts in music, scene transitions and the speed of the action. He compares it to a gut reaction to a sneeze or yawn, which he said shows “how human beings are biologically programmed to notice certain types of sounds.”
“The emotional response to content remains a critical question for marketers to answer,” said Graham Page, evp, head of global research at Kantar Millward Brown (KMB). “They’re putting out so much content now they need really good tools to understand how good it is and as early as possible.”
KMB is one of the biggest clients of an emotional artificial intelligence startup called Affectiva, which aggregates facial data and speech analytics from remote video interviews with test-viewers all over the world.
As it’s been fed tens of thousands of ads and video reactions to them, the startup’s system has learned to notice cultural nuances in the way its diverse audiences show emotion. For example, South Koreans tend to not be very expressive, whereas Brazilians are much more expressive, according to Page.
In one recent instance, KMB used Affectiva’s tech to study how people perceived female gender stereotypes in ads on behalf of Unilever. Based on a two-year survey of 800 ads, they found that the mere 5 percent of ads that cast women in authoritative roles significantly outperformed those that had more stereotypical parts.
FedEx evp, chief marketing and communications officer Raj Subramaniam said that when the brand shifted its marketing strategy “to capture the emotional side of package delivery,” it tapped Spark Neuro to measure the emotional impact and attention levels of a series of potential ads. The startup helped the delivery company arrange a multichannel campaign based on the reactions.
The MIT Media Lab, which spawned Affectiva, serves as a ground zero for this work. A research team published a study in partnership with McKinsey last December after they trained a neural network to trace the emotional arc of a movie by having it watch thousands of short clips of random films, TV shows and online features. They then checked their work against behavioral experiments with humans to improve the machine’s accuracy.
Eric Chu, the lead researcher on the so-called Story Learning Machine, said the report has since attracted interest from media and film companies.
Chu said the tool can surface more emotionally resonant stories or discover types of stories that aren’t being told, adding, “Some people were surprised that we were able to find that the emotional arcs had statistically significant effects—not because they doubted that emotional arcs matter—but because it mattered so much.”