If you feel like TV commercials have gotten louder, faster, and busier—in a word, more energetic—you’re right. A Kellogg study of more than 27,000 ads shows that ad power has increased over time.
And in general, more active ads get viewers to watch longer—especially for certain types of programming.
This is the conclusion of a marketing professor’s research Lakshman Krishnamurthyalong with a Ph.D Yingkang Xie at Kellogg, a former Kellogg PhD and postdoctoral researcher at Stanford Joonhyuk Yang (now assistant professor at the University of Notre Dame) and Purush Papatla, professor of marketing at the University of Wisconsin-Milwaukee. They were able to determine this by cleverly mimicking a system Spotify uses to rate the energy level of songs.
“Ad content is a hyperdimensional monster,” Yang says. “You could measure thousands of different elements in an ad. But our approach focused on one feature—energy—and we made a careful effort to validate what energy is and whether we could find interesting patterns.
While advertisers have less say over how people watch ads in broadcast or taped programs, this study shows that, at least for live programming, advertisers can influence viewers’ attention. “They have a way of keeping viewers tuned in,” says Krishnamurthi.
Understanding commercial viewing habits
The idea that ads have gotten louder and busier isn’t new. In the early 2000s, many advertisers turned up the volume of their TV commercials to get viewers’ attention. This led to the CALM (Commercial Advertisement Loudness Mitigation) Act in 2010. Now, advertisements must not be louder than the average volume of the programs they appear on.
So advertisers found other ways to increase the energy of their ads—often through bold graphics and loud music. Previous research showed that more than 80 percent of ads on Hulu were rated as active.
However, previous studies of active ads tend to rely on small sample sizes or viewers in a laboratory setting, making it difficult to determine the overall effect of these new active ads on viewers.
To get the bigger picture, the Kellogg team used a large-scale data set of television ads shown in people’s homes.
This dataset, from iSpot.tv, watched 9 million television sets from September 2015 to August 2018. The company relies on automatic content recognition technology to detect whether an ad is playing on screen, what brand is being advertised and what appears during the ad. Every time a participating TV is turned on, this information is tracked, down to the second. This allowed the researchers to see the exact moment an ad was interrupted, either by changing the channel or turning off the TV. (The data set only includes viewers who watched live TV, and the researchers had no data on who was watching — just that the TV was on.)
This gave the researchers a data set of 27,000 commercials aired on the five major broadcast networks (ABC, CBS, FOX, NBC and CW) in the United States. They also collected and analyzed 3,077 ads that aired during the Super Bowl between 1969 and 2020, in a dataset made available by adland.tv.
Defining active ads
The researchers then had to measure the energy level of the 30,000 ads. But how could they do that without seeing and rating each one?
It turns out that such a measure of energy already existed through Spotify, the music streaming service. Spotify uses algorithms to quantify song characteristics for features such as danceability, acoustics, and tempo. Energy is one of those features, which Spotify defines as “a perceptual measure of the intensity and powerful activity unleashed throughout the track. Typical energetic tracks are fast, loud and noisy.”
Spotify’s energy measurement algorithm is proprietary, but the researchers were able to revise it with a dataset of more than 13,000 songs from the Free Music Archive and help from the music analysis tool Librosa. They used machine learning techniques to find the right set of acoustic features to construct the energy levels of the sound and compared it to Spotify’s measure.
Once they verified their algorithm, they extracted the audio from the 27,000 TV commercials to analyze the energy levels of the ads, not only for music, but also for speech and non-musical sounds.
But this reverse engineering algorithm only measured the energy level of the ad audio. The researchers had no large-scale way to measure the energy level of optical elements. So they designed an Amazon MTurk study in which 2,432 participants watched randomly selected commercials and rated their energy levels. Some participants heard only the audio of the ads, some watched only the visuals, while others watched the ads normally. They then rated the energy levels of those they had watched or listened to.
The researchers found that acoustic energy levels are highly correlated with visual energy levels, giving them more confidence that their large-scale acoustic analysis translates well into an overall measure of energy.
“We thought there might be some variation in the energy level of the audio and video content, but it turns out it aligned very well,” says Yang.
High-energy ads keep viewers’ attention—most of the time
Overall, the researchers found that the energy level of ads increased by 33 percent between 2015 and 2018. Super Bowl ads, an energy bundle to begin with, had a 20 percent increase in energy level between 1969 and 2020.
They also found that ad energy levels vary throughout the day, dropping in the early morning and peaking at 11pm on weekdays. Ads that run on weekend afternoons are more active than those that run on weekdays.
The product category of the ad often determined its energy level. They found that ads for trucks, sports vehicles, entertainment and retailers tended to be fast, loud and noisy, while ads for politics, government and education had the lowest energy levels. Ads shown during sports programs had the highest energy on average, while ads in news programs had the lowest energy levels.
So, what was the impact of advertising on consumer behavior?
For each ad, the researchers collected data on the percentage of viewers who stayed tuned for at least 25, 50 and 75 percent of the ad’s duration.
They found that, overall, higher energy levels resulted in fewer people tuning in, after controlling for the network, day of the week and time within the program that the ad aired. This extra airtime for active ads translated to about 5,079 more TVs tuning in for at least 75 percent of an ad’s duration.
“We thought that at some point, viewers might get annoyed if the ads were too energetic,” says Xie. “We were surprised that it didn’t seem to be the case across the board.” In fact, by and large, it was the opposite.
But context turned out to be important, the researchers found.
Active ads for food and beverage products were likely to be viewed for longer when placed in entertainment and news programs but not in sports programs. Active health and beauty ads were also likely to be viewed for a shorter time when placed in sports programming. This could mean that advertisers need to be aware of the context when placing active ads in certain programming.
Overall, the researchers say their findings give advertisers another tool to grab viewers’ attention without alienating them.
“Advertisers now know they can’t increase the volume of their ads, so they’re looking for other ways to get viewers’ attention,” says Krishnamurthi. “Our research shows that increasing ad energy can do this, and now advertisers can continue to tailor their ads to find the right fit.”
