Digital-assistant voices should be female, says science

Many of the voice assistants currently available to us have something noticeable in common—their female preset voices. Many of them have also been given distinctively feminine names.

Why have so many tech giants decided to assign a gender to their voice assistants? Is their femininity a reflection of an industry still struggling with diversity? Could it be considered regressive stereotyping built into their design? Or is it simply because consumers have an inherent preference for female voices, as argued by the head of Amazon’s Smart Home division? With 70% of people claiming they don’t care about the gender of their voice assistant, it certainly raises the question of whether there is any value for brands in challenging the status quo.

At Neuro-Insight and Mindshare, we wanted to investigate this further. By measuring subconscious responses, we sought to uncover how people really react to the female digitised voice. In a world where conversational commerce is set to proliferate, it’s imperative for brands to understand the true impact elicited by different voices.

To understand this, we conducted a study using a brain imaging technology called steady-state topography to explore the neurological response to the gender of voices. We built two Alexa Skills, one with a male voice and the other with a female one, from the Alexa application programming interface Polly. Each Skill contained identical responses to a set of queries. Half our sample (105 people aged 18 to 65) carried out the preselected interactions with the male Skill and the other half with the female.

The study measured the level of positive brain response (approach) elicited by each gender, as well as how compelling each gender was from the brain’s perspective, measured by combining three metrics that together indicate how strongly information is being processed: emotional intensity, personal relevance and long-term memory encoding.

Our research confirmed the long-held belief that people find digital assistants with female voices more approachable. But, interestingly, it revealed a considerable variation in the degree to which the information delivered was processed and stored depending on the age and gender of the consumer.

What we discovered was a significant bias towards female voices among people under the age of 35; respondents of both genders in this group found a female voice both more approachable and more compelling than its male counterpart. In contrast, while those over 35 also found the female voice more approachable, they actually found the male voice to be more compelling. 

The generational difference in response could well reflect traditional attitudes towards authority figures and gender roles, suggesting a residual impact from a past in which male voices may have been accepted as having more authority. It may be that these responses reflect an unconscious bias among older people of both genders (but especially male) towards trusting a male voice in providing information and recommendation.

In comparison, younger people have grown up with much greater representation of female figures in authority positions and so respond more positively to the female voice. We saw the strongest response to the female voice among young women—this may in part be driven by a stronger affiliation to female representation as a result of the recent momentum in gender diversity.

Additionally, young people are more used to voice interactions in general and, because these are often female, they may be more familiar and comfortable with female digitised voices.

Either way, the findings point to a clear psychological rationale for using more female voices—particularly when communicating with future generations.

The research clearly suggests that it’s necessary to adapt gender and tone to drive the best engagement among different consumer groups, depending on whether you hope for a warm, emotional response or simply look to convey information. To create brand affinity with almost anyone, a female voice may well work best but, for example, when designing a voice navigation system for an older male driver, a male one would potentially be more appropriate.

As brands start to create more of their own voice experiences, it’s essential to understand these nuances and variations in response. The future does indeed sound female for some consumer groups, but brands should certainly not assume it does for everyone.  

Heather Andrew is chief executive of Neuro-Insight and Jeremy Pounder is futures director at Mindshare.