The Internet of Things has ushered in a revolutionary transformation: Things immediately learn how to communicate with a natural, human interface, whether sending or receiving information. And crafting this communication is a major design and strategic task involving all types of different questions: Where is the conversation taking place? At home, in the car, or in public? What is the purpose of the discussion? Conveying information? Providing entertainment? Arranging insurance policies? Or parking the car? And does the machine conversationalist have a discernable personality?

Of course, every brand has its own voice. Our everyday work is devoted to shaping this voice, making it distinctive and clearly recognizable. That’s nothing new. But conversational interfaces literally give an entirely new dimension to the idea of the tone of voice: its audible sound. This has always been relevant for selecting the right speaker for commercial or podcast voiceovers, but they can still be somewhat interchangeable – like a singer in a choir (no offense). But conversational interfaces allow for solo performers – the representative sound of the brand.

Understandable, pleasant, matter-of-fact, nice, calming, trusting, emotional, and, yes – sexy. These are just some of the criteria that Statista asked users to evaluate conversational interfaces. But what would the assistant system in a Ferrari sound like? Loud, fast-talking, a little arrogant, and with an Italian accent? This is reminiscent of Domitzlaff and his theories about branding from the 1920s: Brands are personalities that can be described with personality attributes. 

This approach is the obvious one for designing human-computer interactions that are supposed to seem as natural as possible. When designing robot assistants, however, achieving the closest possible similarity to real people did not necessarily prove to be the most successful strategy. It’s important to remember the “uncanny valley”: How anthropomorphic can a machine be before it becomes completely unsettling?

Alexa has a friendly woman’s voice and very human word choice and syntax, but is still reassuringly machine-like. And her self-deprecating robot humor is almost legendary (Alexa, self-destruct!). People use Alexa in their homes, where they speak relatively normal and hands-free with her – just like talking to another family member. Siri, with her awful bureaucratic hotline speak, is the total opposite. When it comes to robot humor and »Natural Language Processing« (NLP), however, they are relatively equal.

Fun fact: NLP used to be primarily associated with the controversial self-improvement technique »Neuro-linguistic programming«, but still invites critical examination. Why, for example, do both of these digital assistant systems have women’s voices? Does this fuel and confirm gender stereotypes, or the opposite?

Another aspect involves how information is structured and filtered. How well do Siri and Alexa actually know their users? What happens when services are actually perfectly customized – by the filter bubble in the echo chamber? When we use search engines and websites, we’re used to looking at the information architecture and making the choices that match what we’re looking for. In contrast, conversational interfaces serve us information in a strictly hierarchical and linear way. Of course, this can also be very pleasant, especially when you can have something that resembles a pleasant conversation instead of the annoyance of filling out forms. Good examples include Maya, the »charming artificial intelligence bot« used by Lemonade, an innovative renters and home insurance company in New York. Or Vee by Vhi, the smart personal assistant that motivates users to run in a charming way.

And even if it’s mostly used as a design gimmick: GE’s landing page welcomes visitors with an oversized search field that asks questions, like »What can we help you find?«. Using »We« is interesting here, because it already seems like a hybrid. The site addresses the user personally. Instead of an ethereal personified assistant asking the question, however, it’s the entire company.

Well-designed conversational interfaces are extremely powerful touch points for every brand. They can help build and strengthen emotional relationships with customers, as long as they make the lives of their users nicer, better and easier – not like Clippy.

Interesting links on the topic:
www.convercon.co.uk 
www.robothalloffame.org
www.ibm.com/watson/index.html
www.lemonade.com
https://de.statista.com/infografik/8846/beurteilung-der-stimmen-von-digitalen-sprachassistenten/
https://botlist.co/bots/2570-catbot

by Agnes Bachmaier, Copywriter at KMS TEAM

Press contact

Annette Koch
T +49 (0)89 490 411-368
a.koch@kms-team.com

Close details