the Mandatory of Marketing – Part Two: Measurement

First, a quick recap of Part One that focused on behavior. Emotion in marketing communication is pretty much universally regarded as a good thing.  There are two academically supported origins of that belief.  The first relates to decision making.  Humans require an emotional detonator to decide.  The second relates to memory.  Laying down a memory (such as a rational reason-to-believe) is markedly aided by disruption whilst simultaneously eliciting the target discrete emotion.

Now, onto Part Two.  Before we knew better, moderators would prepare to conduct focus groups for concept testing or advertising pre-testing, and creatives and brand owners would discuss testing the “emotional territory.”  Typically, moderators would enquire as to which specific emotion the creatives were seeking to elicit?  The assumption was that the named emotion was linked to buyer behavior.  Back then, the science did not yet exist to definitively answer those questions.

Today, the science is in, and it has become clear that using qualitative research for pre-testing emotional elicitation is a fool’s errand.  It is estimated that only a few percent of emotions progress to become a conscious feeling.  We can individually recall the times that emotions have become consciously felt feelings.  The death of a loved one, a nasty fright, a mighty frustrating bank interaction.  However, the average consumer’s interaction with a brand does not cause an emotion to become conscious. 

It is pointless to ask a focus group participant to say how they feel in response to a stimulus such as a story board or animatic.  Indeed, the more the moderator attempts to drill down to the emotion, the greater the likelihood that the participant will be providing a rational response rather than tapping into an emotion.

Even if the participant is in touch with a feeling made conscious, that conscious feeling is private.  Sharing feelings is subject to social and cultural norms.  Compounding this, is the observation that when participants are aware of their feelings and are prepared to share them, they systematically understate their negative emotions.

If the objective of the focus group is to develop a hypothesis or provide direction such as concept testing, then that is perfectly fine.  However, too often focus group moderators reside in the murky mire of asking focus group participants to “say how you feel.”  You wouldn’t ask a tennis player to tell you how they non-consciously applied the laws of geometry to instantly calculate the trajectory of the ball they are positioning themselves to return. 

Neuroscience has established that there is no direct neural pathway between the amygdala, the emotional processor, and verbal expression.  It is pointless to ask a focus group participant to say how they feel in response to stimulus like a story board or animatics, especially in the artificial environment of a focus group venue.

Forever and a day, the creative fraternity has quietly accused the marketing researchers of haphazardly meddling in the creative process; having a voice without accountability; a voice that often yields not much more than yet another subjective opinion.

“Having a voice without accountability; a voice that often yields not much more than yet another subjective opinion.”

Product or market driven innovation for measuring emotion?

A decade ago, it was common to see multiple conference papers being presented at marketing and insights conferences on the measurement of emotion.  The topic today is as important as ever however, the high attrition rate of those who claimed to have breakthrough insight has left a mere couple of firms globally remaining in the discussion.  Many of the firms that were the pioneers between 2008 and 2015 have either moved on to something else or have disappeared without trace.

Indeed, for many, from an economics and investment returns perspective, attempting to measure the presence of emotion has been for clients and vendors alike, economically fruitless.  Such “neuromarketing” innovators could best be described as well-intentioned explorers tinkering with medical imaging equipment that detects emotional arousal and valence (a broad categorization of emotional response as either positive or negative, pleasurable or painful, agreeable or disagreeable), and mistakenly claiming breakthrough marketing insight.

The measurement of emotions for the application of marketing has been the perfect illustration of product-led versus needs-based led innovation.

Those product-led “innovators,” focused on the physiological measurement of brain activity and in essence, proposed a new purpose (ad testing) for what was largely, an existing technology (medical imaging).  Back then, debate at conferences raged between which physiological based measure was best for valence.  The techniques had one thing in common; they could not measure discrete emotions but rather were limited to measuring emotional arousal and valence.

“The techniques had one thing in common; they could not measure discrete emotions.”

The presence of emotional arousal might have been a marker of potential memory encoding however, there was no evidence to suggest a linkage between general emotional arousal and specific behavior.  Indeed, based on the performance of the emotional anthems discussed in Part One, there is ample evidence to the contrary.

At the time, some of the untested assumptions such as positive valence was good and negative valence was bad, was plainly incorrect.  For more than a decade, American insurance company Allstate has successfully run a campaign (Mayhem) that intentionally elicits negative valence.  Whilst the campaign brings humor to the situation, the residual emotion arising from the spots is negative (specifically, implicitly measured it is anxiety).  This topic is dealt with in greater detail in Part Three.

The product-led approach to measurement – physiological measures

Fifteen years ago, vendors were “chomping at the bit” with excitement, urging marketers to incorporate emotion into marketing measurement.  Output from various techniques falling into the category of physiological measurements were on offer.  Medical science tools such as magnetic resonance imaging (MRI), electroencephalography (EEG), and magnetoencephalography (MEG) revealed a physiological moment by moment, precognitive presence of arousal to advertising.  These approaches were expensive and based the findings on exceedingly small samples.

The proponents claimed that they could detect and graphically represent the presence of emotion and better still, characterize the neural processing, moment-by-moment, as either approach versus avoidance.  At least for a time, the proposition of an imageable emotional response being used to optimize marketing communications was seductive to marketers.

Two problems – first, the proponents had no objective means for determining which discrete emotion was being elicited.  The second problem related to the blanket assertion that avoidance signaled a negative outcome for marketing communication.

“The blanket assertion that avoidance signaled a negative outcome for marketing communication.”

There was another issue.  Moment-by-moment measurement was analogous to measuring a paragraph, word-by-word.  The desired behavioral affect occurs not at the end of a word but at the end of the paragraph.  Moment-by-moment had limited diagnostic value.  Perhaps the one exception was the measurement of an emotional peak-end but, not knowing which emotion had been elicited remained a major limitation.

Perhaps facial muscles were the panacea?

Jump forward to facial coding.  As other physiological methods were discarded, facial coding began to gain traction.  To the extent that this method purported to measure some discrete emotions, facial coding appeared more promising however, on closer inspection, it was shown that the technique was shallow and lacked construct validity.

Shallow, in that facial coding only measured discrete emotions where there was a facial muscle tell, and invalid for emotions such as happiness, because it required the respondent to activate their zygomaticus major (the smile muscles).  In addition, eudemonic happiness could not be captured since it might not involve smiling and therefore, would go undetected by facial coding.

On top of that, facial coding does not measure emotions such as pride.  If the technique you are employing to measure emotion in consumer choice does not include the emotion pride, then you have omitted the most common emotional detonator of consumer behavior.  In Forethought studies, pride has been found to be a top three emotion in 73% of cases.

Using an implicit scale, Forethought replicated a facial coding study undertaken by one of the major Hollywood studios.  Tests were conducted to measure the happiness elicited by watching excerpts of television shows across two genres: comedy and drama.  It was found that for both genres, facial coding markedly understated the level of happiness elicited from the programs.

The replicated study found that whilst there was a higher proportion of people who experienced an increased level of happiness elicited from comedy, the strength of happiness was more shallow and not as strong in driving behavioral change as the happiness elicited from a drama segment.  It is simply nonsense to assume that people are only happy when they are smiling – Exhibit Two.

Exhibit Two: Comparison of Elicited Happiness via Facial Coding versus Implicit measurement

The market-led approach to measurement – implicit measures

Alternatively, a market-led approach determines which discrete emotion is most associated with consumption/usage behavior for the specific product category.  It also determines what the resting state of a brand was on that emotion and then, how the marketing communications performed on eliciting that discrete emotion.  Finally, the market-led approach determines if the emotion elicited in the marketing communication, attached to the brand?

Even if you could intuit which emotion was most associated with the category, objective measurement of discrete emotions is required to predict how effective the creative will perform.

“A market-led approach determines which discrete emotion is most associated with consumption/usage behavior.”

The contemporary approach for the measurement of emotion is implicit measurement.  Implicit scales assess a respondent’s non-conscious, spontaneous response to the stimulus thereby, circumventing the potential bias brought about by conscious information processing (such as would occur in a focus group).  There is no precise time stamp for what is considered implicit, although it is generally accepted that responses around 700 milliseconds (0.7 of one second) are uncontaminated by conscious thought. 

The average response time for the Forethought Feelings scale is 750 milliseconds (Exhibit Three – illustrating happiness).  The scale is gamified, animated avatars that are metaphor based.  Above all, it is non-verbal.  Metaphors are considered a window into the non-conscious.  Respondents simply move their finger to reveal how they feel.

If we were being religious about the 700 milliseconds demarcation point, responses which are too slow to be considered implicit would be culled.  In the commercial world of insights, this is not the practice.  Therefore, the application of the implicit scale is not considered strictly pre-cognitive but rather, subject to minimum cognitive interference.

Exhibit Three – Metaphor based implicit scale and Response Time (milliseconds)

Identifying the emotion most associated with the consumption/use in your category is just the first step.  Achieving mastery in eliciting that emotion through marketing communications requires a deep understanding of how the emotion manifests in the consumption context, in order to most effectively resonate with consumers.

Exhibit Four illustrates the emotional resting state of three consumer US brands.  The faint red line at zero is the standardized calibration of how respondents were feeling prior to being shown any stimulus.  The Hershey brand elicits positive emotions whereas, the McDonald’s brand elicits negative emotions.  Subway is the emotional leader in the Quick Service Restaurant category.

Exhibit Four – Emotional Resting states of brands

All in All

Presently, the only means for measuring a hierarchy of emotions driving consumption behavior is via an implicit scale.  No other physiological or biometric method captures the range of emotions associated with buying behavior.  Physiological measures do not measure discrete behavior and techniques such as facial coding, which does measure discrete emotions, misses the most common emotion in buyer behavior, pride.  Furthermore, facial coding identifies smiling as happiness whereas, it is more like to be the fleeting emotion of joy.  Incidentally, you do not need to be smiling to be happy.

For all this century, Forethought has been seeking to understand emotion in marketing communications.  It has been more than ten years since we invented the implicit scale to add emotion to our causal marketing models.  These models (Prophecy Thoughts & Feelings®) are highly correlated with third-party validated, in-market behavior.  We are not aware of any other organization in the world able to substantiate such a claim.  We have certainly learnt a lot but would be the first to say, there is still a great deal more to learn.

References

The Greatly Maligned Focus Group. https://www.greenbook.org/mr/gain-and-retain/the-greatly-maligned-
focus-group/
2 When Will There Be Peace Between Creatives and Researchers? https://www.greenbook.org/mr/gain-and-retain/when-will-there-be-peace-between-creatives-and-researchers/
3 https://www.greenbook.org/mr/gain-and-retain/emotion-the-mandatory-of-marketing-part-one-behavior/
4 For example, https://www.youtube.com/watch?v=m2vzNgEzrH8
5 Please see ‘Two Faces Have I – One to Show, One to Hide’ https://www.greenbook.org/mr/market-research-
methodology/two-faces-have-i-one-to-show-one-to-hide/)
6 Roberts, Ken., Roberts, John, H., Danaher, P.J., Raghavan, R. 2015, Incorporating Emotions into Evaluation
and Choice Models: Application to Kmart, Marketing Science [P], Vol 34, Issue 6, Institute for Operations
Research and the Management Sciences (INFORMS), Hanover USA, pp. 815-824. (Nov-Dec 2015)
7 United States Patent 8,939,903, Measurement of emotional response to sensory stimuli, Inventors: Roberts;
Kenneth George, Wong; Elaine PhD, January 27, 2015