Digital mass persuasion: psychological profiles make marketing more effective

The digital footprints left in our wake as we traipse around the Internet can be used to influence the behaviour of large groups of people, a new study has shown.

Advertisers have known for a long time that tailoring the message to the target audience is critical when it comes to making a sale. Pitch it right and it resonates with the customer, who obligingly puts his or her hand in their pocket.

But when you cannot see or interact with the audience, which is increasingly the case with digital media, this opportunity may be lost. Nevertheless, it might be possible to dramatically increase the effectiveness of advertising - and not just for new cars and sunny holidays but also for serious health or educational messages - and influence massive groups by exploiting information people give away about themselves online.

In a proof of concept paper published this week in PNAS, Columbia University researcher Sandra Matz and her colleagues show that it's possible, by appraising a user and then returning an advert tailored to their personality type, to increase the uptake of different products offered online by at least 150%.

The researchers considered two main behavioural traits in their study: extraversion, and "openness". Extraverts tend to be energetic, talkative, sociable and outgoing, while individuals labelled as "open" tend to be intellectually curious, imaginative and unconventional.

In their first experiment, one of two different advertisements for the same beauty product were freely presented to 3 million Facebook users. One of the ads was more flambuoyant and designed to appeal to extraverts. The other was a more reserved version intended to strike a chord with introverts. The team logged how many clicks each of the two adverts received and by whom.

When a extravert-favouring advert was presented to a user appraised as extraverted, the click-through rate on the advert was up to 1.5 times greater. The same was true for an introverted advert presented to an introverted user. And similar findings emerged when the researchers looked at users levels of "openness" using a different product.

The researchers also test-marketed a computer game that involved shooting down bubbles. This time they established that people who bought similar games tended to be highly introverted, so two promotional "strap" lines were written. One was brash and favouring extraverts, the other a more soothing invitation to play the game and unwind. Introverts presented with the introvert-friendly avertising slogan clicked on the game 1.3 times as often.

The team made their assessment of users' personality types just by using the list of things that these people had "Liked" previously on Facebook. These "Likes" were cross-matched with data from a publicly-available database that contains the Facebook "Likes" of millions of other people as well as the results of a personality questionnaire completed by the same people. In this way the team were able to draw up a list of the top ten things an extravert, an introvert, an open or a more conservative individual tends to "Like".

Critically, this was all achieved without requiring any of the millions of people who voluntarily participated in the study to surrender additional information about themselves beyond that which was already readily accessible within the public domain. This the team emphasise in their paper "carries both opportunities and ethical challenges."

On the one hand, psychologically-tailored health communication could be used to better target vulnerable groups, such as those at risk of mental illness. The flip side of the coin is that the same approach could be used by nefarious operators to target individuals susceptible to fake news stories, dubious financial offers, or online gambling.

What's concerning, Matz and her colleagues point out, is that the same technique can also be used by advertisers to deduce information that end-users had not intended to give away. For instance, by merely clicking on an advert that strongly targets a specific personality type, a user will reveal that trait to the advertiser.

More alarming still is that the methods described in this paper fall clean through the cracks in existing online legislation. In recent years, the EU in particular have placed considerable emphasis on privacy and data security. But, as Matz points out, "none of the measures currently in place or in discussion address the techniques described in this paper."

Maybe Sting should start singing "every step you take, every click you make, someone will be watching you..."