When ads miss targets

Professional magicians have long been aware of one quirk of human behaviour: People are less interested in a trick if they know how it’s performed. Behavioural scientists are now discovering that a corollary of that principle extends to targeted online advertising.

A new study from three researchers at Harvard Business School came to the conclusion that consumers are less likely to respond positively to targeted advertising if they know how exactly they were targeted. The study (available at https://hbs.me/2Ga48Xu) examined the impact of transparency on ad-consumption. 

The study took a sample of 449 persons (all recruited online) who were asked to browse a website, which presented them all with versions of the same advertisements. These ads were divided into three sets, differing only in disclosures about why the individual subject was being targeted to see that specific set of ads.

Some ads had no disclosures at all. A second set claimed surfers were targeted with ads based on their history and page views at that specific website. A third set claimed that viewers were targeted on the basis of their surfing history across the entire internet. The third disclosure always resulted in significant loss of interest. Surfers presented with the third disclosure were 24 per cent less likely to visit the advertiser, or to follow-through, in other ways.

The subjects participated in another experiment, where some were told nothing about why they were being targeted with specific ads; others were told they were targeted on the basis of information they had voluntarily provided; a third set were told that they were targeted on the basis of information inferred about them. The third set of subjects were 17 per cent less interested in purchase, or follow-ups. 

Assuming these results are rigorous, this insight presents online marketers and advertisers, and regulators, with a conundrum. In the interests of transparency, an ethical company “should” tell consumers why they are being targeted. But what if the company knows that telling a consumer will, in itself, very likely reduce her interest? 

Targeted advertising is based on both consensual and non-consensual surveillance. Artificial Intelligence is extensively used to analyse IAO (interests, activities and opinions) and infer OCEAN (the personality traits of Openness, Conscientiousness, Extraversion, Agreeableness, Neuroticism). 

Mobile apps routinely track and monitor activity on handsets (with permission granted by users). Social media platforms like Facebook monitor user-activity across multiple websites. Google ties together vast amounts of data and information based on email content and search histories. Amazon presents ads that tell customers that others who bought “X” also bought “Y”. Netflix and YouTube guess what sort of programming users will be interested in, based on their usage history. Tinder presents profiles that fit previous matchups of users.

This study suggests that consumers are more or less okay with targeted ads that use voluntarily disclosed information, or information that users are aware they have revealed to that specific website. For example, Amazon’s suggestions and YouTube’s suggestions are popular with consumers. But consumers tend to be unhappy about being targeted by data gathered from multiple different websites. The FB-user doesn’t want FB to know what she is buying from Amazon. 

The study says, “Ad transparency reduced ad effectiveness when it revealed cross-website tracking — an information flow that consumers deem unacceptable, as identified by our inductive study.” This is not hard to understand if you draw an analogy from real life.

We are comfortable discussing details of our private lives with friends and family because we have already consensually disclosed a lot to them. We are comfortable talking about our medical history with a doctor, and our financial affairs with a chartered accountant. But we would be uncomfortable if the doctor gleaned our financial history from the CA, without our knowledge. 

But this still leaves the conundrum unsolved. What should the ethical corporate disclose when it targets you? Or more cogently, what disclosure should regulators demand?

Twitter: @devangshudatta