NEW YORK: Brands should be wary of using characters and mascots to humanise online recommendation systems, especially if they draw on potentially "sensitive" information about consumers, a study has suggested.
Marina Puzakova (Oregon State University), Joseph F. Rocereto (Monmouth University) and Hyokjin Kwak (Drexel University) explored this area in a paper published by the International Journal of Advertising.
"Recommendation agents" – such as those built by Amazon, Netflix and Pandora – have become popular features of the web, they observed, particularly as they help users deal with an overload of information.
Some brands have attempted to add a human element to this process by "anthropomorphising" these systems –that is, by associating them with a character like Travelocity's "roaming gnome", a move that boosted its brand.
However, Puzakova, Rocereto and Kwak identified possible problems with this approach if it used "personally sensitive information" – for example, about an individual's health or sexual habits – to customise messages.
One of their experiments asked undergraduate students to imagine they were a consumer who might be suffering from a fungal infection. They then viewed a fictitious health website to learn about several available remedies.
Upon visiting this site for a second time, they were delivered a pop-up ad where a 2D human silhouette acted as an "assistant" for a made-up fungal care brand, and effectively offered to help them find the right medication for the infection.
The resultant analysis confirmed the "negative effect of the anthropomorphisation of the recommendation agent" when the information that was used to customise this message was personally sensitive in nature.
An adverse impact was recorded both with regard to the attitude towards the pop-up, and also in terms of the likelihood to click on the ad itself. (For more, including details of a similar experiment related to financial services, read: Ads are watching me: A view from the interplay between anthropomorphism and customisation
"A theoretical explanation of these results is that people perceive anthropomorphised entities as possessing their own mind and, therefore, as being intentional about their actions," the three academics noted.
The authors concluded that – as shown by the case of Travelocity – anthropomorphism can, in the appropriate situation and setting, build positive consumer connections with a recommendation assistant.
"However, delving deeper into the effects of anthropomorphism, our research reveals that it may, in contrast, be a damaging marketing tactic in certain circumstances," they wrote.
"This might especially be the case for firms who anthropomorphise recommendation agents for personally sensitive products."
Data sourced from Warc