
The voice answering your call may not belong to a person. That has not stopped some customers from saying please, thank you and even sorry.
According to figures publicised by voice-AI company fonio.ai in November, seven in 10 calls to its assistants included a “please” or “thank you”. About 12% of callers apologised. Without more information about the calls analysed, it’s difficult to know how common this behaviour is among British callers.
The manners of the machine on the other end matter too.
A study published in the Journal of Service Research on 29 October examined how AI voice assistants used commanding or more deferential language. Across six experiments, researchers found that deferential language increased users’ sense of rapport and comfort, making them more willing to use the assistants.
However, the research focused on communication style rather than whether assistants could resolve customer problems. The advantage of deferential language also disappeared when participants were under time pressure, suggesting that politeness may matter less when customers need an immediate answer.
Politeness may make AI assistants easier to interact with, but it does not necessarily make them better at customer service. For businesses investing in automated support, the more important question is whether these systems can solve problems without frustrating the people using them.
When callers want a person
Customers may be willing to talk to machines, but many still prefer human help.
A November survey commissioned by contact-centre provider TCN found that 84% of 1,000 UK adults preferred speaking to a person when contacting a business. Only 8% preferred AI, although 37% considered it helpful.
The findings suggest that customers can see a role for AI in customer service without necessarily wanting it to become their primary point of contact.
The reluctance can turn into impatience. In the same survey, 33% said they would immediately try to reach a person when faced with a chatbot or AI system while resolving a customer-service issue.
Ofcom found similarly mixed attitudes in September. Some telecoms customers welcomed AI for straightforward requests, but others worried they might not be able to reach a person when problems became complicated or sensitive.
Some businesses are using conversational AI to offer customers another way to complete routine tasks, rather than replacing human support altogether. Travel-booking company Omio, for example, allows users to describe their intended journeys in natural language and complete bookings through a chat-based tool.
Catriona Meehan, director of public affairs at travel-booking company Omio, describes its chat-based booking tool as an attempt to “meet them where they want to be met”. Travellers can describe the journey they want in their own words and, if logged in, book it through the conversation.
More than a pleasant voice
Paul O’Sullivan, Salesforce’s chief technology officer for the UK and Ireland, sees a more practical measure of AI’s value. He asks businesses whether the technology could help employees “get through your customer contact backlog quicker”.
O’Sullivan describes a progression from automating routine tasks to helping employees work more effectively and, eventually, allowing AI systems to act with greater independence.
Even then, he sees a role for people. “I still believe that you need human oversight,” he says.
He aims to “lower your cost to serve and increase your value output for your customers”. Those goals depend on more than how pleasant an assistant sounds.
Businesses need to know whether callers get their problems resolved, how often they have to call again and whether they can reach an employee when the automated conversation stalls.
Someone calling for the third time about the same problem will be less interested in the assistant’s manners than in getting it fixed.
The real test of AI customer service comes when something goes wrong. A friendly voice may make the experience more pleasant, but trust ultimately depends on whether the business can deliver an answer, resolve the issue and provide human help when needed.
The voice answering your call may not belong to a person. That has not stopped some customers from saying please, thank you and even sorry.
According to figures publicised by voice-AI company fonio.ai in November, seven in 10 calls to its assistants included a “please” or “thank you”. About 12% of callers apologised. Without more information about the calls analysed, it's difficult to know how common this behaviour is among British callers.
The manners of the machine on the other end matter too.