NEW ORLEANS — Advertisements on a voice assistant like Alexa or Siri may be more effective if the assistant is viewed as the medium for the ad, not the source. However, user motives play a role in how these persuasive messages are received, according to a new study in the proceedings of the premier ACM Conference on Human-Computer Interaction (CHI 2022).
In 2017, Google faced backlash from Google Home users who felt that ad messages from the voice assistant (VA) were inappropriate. It seemed generating ad revenue on VAs — like that on search engines — was not feasible.
The study presented today (May 2) says there still may be an opportunity to run effective ads on VAs. It found that ads are better received if the VA is a medium — like radio— and not the source of the ads.
Reactions to ads depended on whether the user was informationally motivated (one who asks about the weather or news) or socially motivated (one who uses VAs to cope with loneliness or just to chat).
Socially motivated users see VAs as “companions in their midst,” according to S. Shyam Sundar, James P. Jimirro Professor of Media Effects in the Donald P. Bellisario College of Communications at Penn State and co-director of the Media Effects Research Laboratory.
“There are many skills that have been added to VAs that are very social in nature,” Sundar said. “You can have a casual chat, ask them to tell a joke, thank them when they wish you ‘Good Morning,’ and so on.”
If the VA promotes a brand in the middle of such social exchanges, it is not so off-putting, according to Eugene Cho, assistant professor at the College of New Jersey, and lead author of the study, who worked with Sundar.
The researchers recruited 264 participants who were familiar with VAs for their experiment. They were provided two different scenarios in which a specific question was posed to Siri. They listened to Siri’s response to the question, which was followed by an ad related to the query.
Example: When a participant asked, “Siri, how do I make pumpkin spiced latte?” Siri responded with the recipe, followed by an advertisement for Starbucks and how to get a pumpkin spiced latte using the company’s app. Chosen randomly, some users heard the ad as a human spokesperson and others heard the voice of Siri directly relaying the ad.
Those with high informational motives responded negatively to ads when Siri was the source — versus the human spokesperson. On the other hand, social motives led to “higher social presence” when Siri delivered the ads.
“Motivations are linked to usage,” Cho said. “[For Google Home], maybe it wasn't the ad that was so bad. Maybe it was the content and the context that was the problem.”
These findings are particularly relevant to companies like Google whose main source of revenue comes from advertising.
“We understand that ads are inevitable to their business model,” Sundar said. “We want the way ads are delivered to be more human-centered and contextually relevant. Convert ads into a service. That’s how you can be both commercially viable and socially responsible.”
Sundar said advertising shouldn’t be a “sneak attack.” Through their research, the researchers are advocating a user-responsive approach that is more helpful and less deceiving.
According to Cho, companies that make VAs have the capability to learn about their users and provide an experience suitable to them. “Instead of running ads indiscriminately, they can personalize ad delivery based on user motivations,” she added.