• Sarah Macpherson

The Science Behind Personality-driven Marketing

Advances in machine learning are beginning to make their way into the field of quantitative psychology, and have started to reveal new opportunities in the same field of personality-driven marketing that we are pursuing here at Pinpoint. As academics in diverse fields collaborate to explore the potential for these innovations, a variety of academic studies are emerging with the common perspective that consumers’ digital footprints provide a strong, predictive signal regarding actionable psychological traits.

Most recently, distinguished Computational Social Scientist, Psychometrician (and Pinpoint Advisor), Dr. Sandra Matz published a number of studies investigating consumer emotional response to images—scrutinizing the appeal and success of images targeted to the personality of the individual. Foundational to the work is the notion that consumers are individuals and messaging that resonates with one person won’t necessarily resonate with the next. Dr. Matz and her associates assessed this theory in a number of experiments determining the reliability of online behavior to predict consumer personality, the success of personality-driven ads, and predictability of image features to appeal to distinct personality types.

Digital footprints predict personality

The torrent of information consumers provide through their online behavior can actually be used to predict their personality dimensions. Dr. Matz and her colleagues investigated this using Facebook ‘Likes’. In the study, the investigators administered personality tests to participants and analyzed their Likes to find relationships between personal traits and their web-professed affinities. The study found that Facebook Likes could predict subjects’ Big Five personality traits with significant accuracy, together with race (95% accuracy), gender (93%), and political orientation (85%). Beyond Facebook Likes, the subjects’ words, phrases, and punctuation also had a strong, predictive relationship with the Big Five Factors. Put simply, this element of the study found that determining customer personality through online behavior was not only possible, but achievable at scale. While Pinpoint explicitly avoids the use of social media data for privacy reasons, our findings from anonymized and broad-based behavioral data (offline, web & mobile) demonstrate very similar predictive power metrics.

Personality matched messaging

With personality assessment based on online behavior established, Dr. Matz and her team investigated the role of personality in persuasive appeals, namely via advertising. Partnering with a UK beauty company, they created ads to appeal to either introverts (‘Beauty doesn’t have to shout’) or extroverts (‘Dance like no one’s watching—but they totally are’). In order to correctly target introverts or extroverts with the designated messaging, the team relied upon their previous finding of Facebook Likes that predicted high and low extroversion. The introverted audience was defined as individuals that liked the 10 most predictive Likes of introversion, while extroverted audience was defined as individuals that liked the 10 most predictive Likes of extroversion.

When subjects received ads that matched their personalities, they were 1.5x times more likely to convert than when the ad served did not match their personality. Matched ads resulted in 40% more clicks and 50% more purchases than mismatched ads, indicating that ads tailored to the personality of the audience materially outperform.

Personality tailored images

Dr. Matz and her team also investigated whether known personality features could reliably predict which images would appeal to specific consumers. That is to say, are there features of images that appeal more to different personalities, and can personality targeted images increase engagement?

To investigate, Dr. Matz and her colleagues analyzed images with respect to 89 features; looking at characteristics such as hue, color diversity, number of people in the image, etc using automated algorithms. Subjects had their Big Five Factor personality evaluated then rated the images based on how much they liked them, allowing the investigators to look for a relationship between personality and image features. The investigators found specific instances of such preferences. While extroverted people tended to prefer images that were simple in composition and contained people, open-minded people tended to prefer cool colored images without people.

The takeaway

Taken together, Dr. Matz and her colleagues have demonstrated that personality-based audience assessment and creative optimization present an emerging frontier across the fields of digital advertising and customer relationship marketing. Pinpoint’s services are focused precisely in these areas and we remain very excited about the potential to contribute to further academic evaluation of each ethical and impactful use case as it emerges. For those readers in the academic community, please reach-out to explore collaboration opportunities.