How To Help Mad Men Avoid Making Terrible Ads

TV ratings company Nielsen promises that EEG, biometrics, and facial scans will help make ads that appeal to short-attention-span viewers.


When was the last time you saw an amazing commercial–one that makes you smile every time you think about it? If you don’t have an answer, you’re not alone. The challenge of attracting eyeballs and the hearts and minds of viewers has bedeviled advertisers since long before the era of Don Draper.


“If you’re making a funny ad, and nobody smiles, that’s a problem,” says Carl Marci, chief neuroscientist at Nielsen–the company most people know for its TV show ratings. Provoking a smile, a laugh, or other emotions is ever more important to making a commercial relevant in a media-saturated world that goes well beyond the living room screen, says Marci. “You can remember lots of ads that had emotion,” he says. “What I guarantee you don’t remember is all the ads [the creators] thought had emotion, and they didn’t. And you just kinda blew by them because they weren’t engaging.”

Ad writers–even with a panel of viewers providing feedback–have lousy track records of picking winners, says Marci. Nielsen’s proposition to frustrated Madison Avenue execs: Put some scientific insight into it by hooking panels of viewers up to electrodes and pointing cameras at them. Nielsen is making a big bet on consumer neuroscience as a way to stay relevant in a fast-changing media world where it’s facing more competition from rivals like comScore, the online audience-measurement company that recently acquired TV- and movie-audience tracking firm Rentrak. Nielsen has snapped up two neuroscience marketing companies in recent years: Neurofocus in 2012 and Innerscope in 2015. It’s rolled their technology together and just published the first results of its approach: a study for CBS of about 60 commercials that were currently running on TV, viewed by 900 participants from November 2015 to April 2016.

Nielsen isn’t the only company providing these so-called neuromarketing services, and it hasn’t purchased all of them, either. A research firm called Neuro-Insight, for instance, uses electroencephalogram (EEG) measurements–employing electrodes to measure brain activity. Neuro-Insight collaborated with Samsung, for instance, to evaluate ads for its smartphones aimed at potential iPhone switchers.

A firm called MediaScience combines biometrics (such as heart rate), facial coding that identifies expressions, and eye tracking to determine what parts of the screen draw the most attention. It recently used the technology to evaluate more than 120 ads for candymaker Mars. Traditional research companies like Ipsos and Millward Brown (part of advertising behemoth WPP) are also adding neurotech to their offerings.

Combining Data

An EEG provides the most data, says Marci, determining whether and to what extent the viewer is paying attention, forming or recalling memories, and having an emotional response. What it can’t tell for sure is if this activity is for good or bad reasons. “If I see a spike in the EEG, I don’t know, without looking at the creative or other channels, necessarily if it’s something that people are smiling at or frowning at,” he says.

If EEG spikes correspond to the punch line of a joke, that’s a good indication that the viewer found it funny. But Nielsen adds other measures to help answer the question. The most obvious is simply to look at someone’s face–in this case, with a computer vision system that does facial coding. That gets to another part of Nielsen’s sales pitch: It isn’t offering just one or a few measurement types, such as EEG for brain activity or an electrocardiogram (ECG) for heart rate, but a suite that combines all of them.


Facial coding by itself is pretty weak. Nielsen reports that it has only about 9% “explanatory power” as Marci describes it–because people aren’t typically animated when they are vegging in front of the screen. But it boosts the accuracy you would get from just an EEG (62%). Biometrics, with 27% explanatory power measure body responses like heart rate and increases in the skin’s electrical conductivity–the latter indicating some type of arousal. Combining all three yields 77% explanatory power, which Marci calls huge.

“We found that no one approach alone is totally predictive of the return on investment that is going to be generated by the campaign,” says David Poltrack, the chief research officer at CBS. “But the combination of multiple neuro techniques with the traditional survey techniques generates substantially enhanced findings.”

Are They Overthinking It?

It’s hard to imagine Jimmy Fallon wiring up subjects to test his monologues, or PewDiePie pouring through data on whether people are into his videos. But unlike a TV or YouTube show, a good ad doesn’t just entertain people; it also motivates them to buy something. “Where people get confused, and where scientists got confused for many years, is that the motivation system in the brain and the liking system in the brain are two different things,” says Marci. You might like an ad for M&Ms (one of the commercials Nielsen has tested), but if you are on a diet, you’re not motivated to actually eat them. Nielsen promises that it can see not only if people enjoy an ad, but if they feel pulled in by the brand, by monitoring their vital signs second by second.

That hasn’t been an easy sell. “Most of our clients have no background in the science and have to learn from scratch,” says Marci. “So I gotta now talk to some creative director at an agency about alpha [brainwave] shifts in the prefrontal cortex, so I’ve already lost them.” Recognition is growing, however. In 2010, the Advertising Research Foundation began digging into neurotech methods, evaluating 50 previous studies and conducting two of its own, published in 2011 and 2013.

The latter included an evaluation of 26 commercials, which had already run, so data on their success, such as sales figures, was available. ARF used traditional methods as well as EEG, biometrics, eye tracking, and functional MRI (fMRI, which measures blood flow in the brain). “These methods can predict advertising impact better than traditional (typically pencil and paper) methods,” says Horst Stipp, who oversaw the ARF research, in an email to Fast Company. ARF doesn’t advocate dumping traditional research like online questionnaires but rather adding neurotech into the mix. “I’m not someone who says we can do it all with neurotech without asking, did they like the ads and what did they think of the product,” says CBS’s Poltrack, who is also the ARF chairman.

Neuromarketing was a hot topic at ARF’s Re!Think conference this past March. It coincided with the release of an ARF study titled “How Advertising Works,” which includes a section on neuromarketing. The technology has received enthusiastic support from Poltrack. Back in March, he was already praising the study CBS did with Nielsen before its formal presentation at ARF’s Audience Measurement conference on June 13.


The Nielsen-CBS study also evaluated ads for Mars, which has gone all-in for comedy with its goofy M&Ms ads. In one live-action/animated combo, a husband comes home early to find his wife in bed with a figure huddled under the covers. He expresses shock and disappointment–not because she cheated on him, but because she didn’t share. A giant animated red M&M pops out from under the covers, saying, “So that’s what you meant by dessert!” Nielsen’s sensor data says this joke is funny, after a tense setup. The ad features a second joke when the yellow M&M pops out of the closet, saying, “Pssst, Scott’s home early.” This got a weaker response, says Marci, so perhaps it could be cut if the ad were shortened for online.

Does it take dozens of electrodes to figure that out? The second gag is less of a surprise and doesn’t add to the plot. And of course it would be the one to go if the commercial had to be shortened, say for online. One of the findings from the Neuro-Insight study with Samsung was that iPhone owners don’t like to be made fun of in ads. That may not be surprising, for any audience. This new research method might be neuroscience, but are the results rocket science?

Maybe not always, but advocates tout some specific numbers that they claim show that the technology can boost sales. Mars, for instance, reported at Re!Think that biometric data from MediaScience predicted sales results, which had actually occurred, of TV ads 78% of the time, versus a 58% figure for ads tested by surveys. Plus, advertisers are hungering for some approach to evaluating ads that promises better results. “We’ve had access to surveys for three decades,” says Marci, “and we’re still not getting it right.”

About the author

Sean Captain is a Bay Area technology, science, and policy journalist. Follow him on Twitter @seancaptain.