In the course of just a few years, the sophistication of brands’ digital interactions with consumers has increased by leaps and bounds. This is thanks to companies’ growing use of artificial intelligence (AI) to guide those interactions. As they crunch ever more data and “learn”, with the help of machine learning and natural language processing, AI chatbots and other intelligent tools are becoming more responsive and taking on more human-like attributes. Digital interactions between consumers and brands are increasingly driven by such “machines”, with limited involvement of human sales and other agents.
How do consumers feel about this? Are they comfortable with the roles played by humans and machines in their online interactions with brands?
To find out, we surveyed 5,601 people in 16 countries. They told us that although they need human agents to remain accessible, they are increasingly comfortable with automated, digital-only interactions.
As that comfort increases, brands will need to find the right balance between the technology and human elements in interactions. Ensuring transparency in how they use AI—and the consumer data it uses to learn—will help to build trust in automated interactions and strengthen their customer relationships.
We surveyed 5,601 people in 16 countries.
Our key findings
Acceptance of machines is growing
There is widescale consumer acceptance of automated interactions with brands, and that acceptance is growing: 56% of respondents are comfortable with fully automated interactions, and just 16% express discomfort. Almost half of the respondents (47%) have grown more positive about such interactions in the past two years.
The option to interact with a human is essential
Today, AI is capable of managing entire customer journeys and providing a superior experience. But consumers want to be able to speak or chat with a human agent if they need to. In fact, 78% of respondents consider blended technology-human interactions a better experience than those that are human-only.
Trust in data use is integral to building trust in machines
Consumers are generally forgiving of technology malfunctions in interactions. Using personal data obtained from third parties, on the other hand, could have serious consequences for brands. Out of all the problems they might encounter in online interactions, this is by far the most likely reason for consumers to stop interacting with a company.
A lack of transparency will lead to lost revenue
Nearly two-thirds of respondents (65%) want companies to be honest about using AI-powered bots to guide interactions—a response that differs little by age group. Such transparency appears to be the norm, but 41% of respondents would reduce their involvement with a company if it were not forthcoming.
Perceptions of what makes a connection “human” are likely to evolve
Few consumers today consider a chatbot or other automated interaction to offer a “human connection”, but younger respondents are more likely to than older respondents: 48% of those between the ages of 18 and 34 (and 41% overall) say it is entirely possible to have a human connection in a fully automated interaction.
Part machine, part human
AI and machine learning (the branch of AI that crunches data and enables software to learn) are now ever present in our digital interactions with brands. These capabilities drive features of online experiences that people are familiar with, such as personalized ads and product offers and automated, end-to-end shopping and purchase transactions.
AI also supports virtual assistants such as Amazon’s Alexa and the interactive chatbots that “converse” with people online by offering responses based on the real-time analysis of personal and other data.
The growing prevalence of these interactive technologies that display human characteristics has elicited angst from some consumers, partly because of potential privacy issues, partly because of poor functionality, and partly because of personal unease with the idea of human-like bots.
In their dealings with brands, consumers prefer the involvement of humans to that of mainly or exclusively technology. In our survey, the preference for mostly human interactions is visible in five categories of goods and services, and it is particularly strong where healthcare services are involved.
“AI can deliver excellent customer experiences, but I don't think you can build a sustainable relationship based solely on AI today,” says Michał Szaniecki, Managing Director, SEAT & Cupra, at Volkswagen Group. “AI can deliver the message, the date, the facts and the figures. The human element is much more important when the interaction becomes complicated.”
“In our industry, most interactions with customers are very transactional and can be handled by technology,” says Amy Shore, Chief of Customer Experience at Nationwide, a US insurance and financial services provider. “But our products also provide protection often at times of a major life change or tragedy when a human touch is really needed. We have to be very smart about which customer journeys to make wholly automated, wholly digital or a blend of both.”
A preference for human interactions
Figure 1: Share of respondents choosing between 1 and 4 (“mostly non-human”) on a 10-point scale when asked the type of interaction they prefer to have with companies, and those choosing between 7 and 10 (“mostly human”). The scale ranged from 1 (“No human interaction”) to 10 (“Frequent or regular human interaction”).
Can a fully automated interaction offer what people could consider to be a human connection? Not according to most consumers in our survey. They define the human element in online interactions in the traditional way, as communication with a real person, whether in a real-time chat, an audio conversation or a video conversation.
Anna Noakes Schulze, Head of Business Development at Women in CX, an online membership community, agrees with them: “As good as they’ve become, AI chatbots are not yet at the point of fostering a real sense of human connection.”
What's a human connection?
Figure 3: Share of respondents who consider different elements of an online interaction with a company to be a “human connection”.
A closer look at the survey results suggests this could change, however. The share of younger consumers defining automated interactions as human is much higher than it is for older consumers.
For example, 25% of respondents in the 25–34 age group define a chatbot interaction as human, compared with just 10% of those in the 55–65 age group. And 41% overall (48% between the ages of 18 and 34) say that it is entirely possible to have a human connection in a fully automated interaction. Just 28% say that this is not possible.
The experts we interviewed say that chatbot technology will eventually be able to understand nuance in conversation and offer empathy. However, consumers must never be tricked into thinking they are interacting with a human (see “Technology transparency”).
Figure 4: Share of respondents who somewhat or strongly agree that it is entirely possible to have a human connection in a fully automated interaction.
Get the balance right
In many online interactions, technology takes consumers through their entire journey with a company, whether it involves just browsing or ends with the purchase of a product. And AI and other technologies often guide aspects of after-sales support interactions.
Technology-only interactions are common when consumers shop for relatively low-value household products or clothing, but AI is also prevalent when more expensive items are involved. Volkswagen Group’s Michał Szaniecki says that most of his firm’s customers go through the entire online car purchase journey using only AI.
“Whenever people are engaged in and knowledgeable about the category, they don’t need to have a human interaction,” he says. “The sales process typically lasts six weeks, after which we'll deliver the car to their door.”
say that blended interactions are as or more satisfying than those that are human-only.
Blended is better
However, our survey respondents’ comfort levels are highest when machine communication is combined with genuine human communication: 78% say that blended interactions are as or more satisfying than those that are human-only. As in many other aspects of our research, age seems to matter here: fewer older respondents than younger ones (although still a majority) are likely to hold this view.
In automated interactions, respondents want to be able to speak/chat to a human if they need to. This is top of their lists when we asked them the most important aspects of positive interactions with companies—it is even more important than speed of response.
“A machine is more likely to ensure speed of response to many customer issues,” says Standard Chartered’s Millie Gillon. “But it needs to be capable of actually resolving the problem.” When the machine cannot do that, the customer journey suffers.
“AI is inferior to a human agent if a consumer requires trust,” says Szaniecki. “Whenever AI asks for personal financial data need for car leasing, for example, 99% of consumers drop out of the journey. They do not trust anonymous machines to collect and send their financial data.”
A connection with a human is particularly desirable in customer service situations. Most survey respondents (62%) would prefer a direct interaction with a human—voice, real-time chat or in-person—to a fully or partly automated interaction (37%) in order to resolve a customer service issue. The oldest respondents in the survey, 55–65-year-olds, are most emphatic about this (75% of them agree), but a majority of 18–24-year-olds (54%) also state this preference.
Successful combinations of humans and technology
Figure 5: Share of respondents who say that the combined automated/human interactions they’ve had with companies in the past 12 months have been as satisfying as or more satisfying than fully human interactions.
would prefer a direct interaction with a human in order to resolve a customer service issue.
Human or machine in customer service?
Figure 6: Share of respondents specifying the type of interaction they would prefer to have with a company to resolve a customer service issue.
Transitions and trade-offs
Blended interactions may be better, but brands often struggle to get them right. “Companies in our industry are good at supporting human interactions with AI and other technologies but have difficulties making the transition from technology to human agents,” says Amy Shore of Nationwide. A key constraint at many firms, she says, is the legacy technology systems and resulting back-end complexity that hampers the integration of newer technologies. Her company has invested heavily in platform modernization across its business lines in the past five years, says Shore, which has facilitated easier interfaces with AI in its customer-facing operations.
Balancing bot and human elements involves allowing for a quick resolution of routine customer requests while letting customers to decide when human intervention is needed, says Anna Noakes Schulze. One of the keys to improving the transition, she says, is getting the bot to recognize when frustration is mounting.
Companies are putting sentiment analysis to work on this, efforts that are being facilitated by advances in natural language processing (NLP). “NLP is doing so much more to enhance AI,” says Shore. “We can do voice-guided chatbot now, and we have tools that listen to and analyze every call that comes into all of our call centers every day. The tools analyze the key issues our customers are facing and what’s driving their negative as well as their positive sentiments. It’s a game-changer to have these capabilities at our disposal.”
The balancing act between human and machine in online interactions has direct implications for the bottom line. “Bringing a human agent into an interaction with consumers is an expensive proposition compared with keeping it fully automated,” says Michelle Emerson-Russell, Director and Global Lead, Workplace and Customer Experience Sales at Verizon.
So there is an ROI decision to be made about the balance between the AI and human elements, says Szaniecki. “I cannot deploy both fully for every interaction—it makes no financial sense,” he says. “I need to incentivize customers to stay within the AI channel, and only transfer them to humans when absolutely needed. If more than 5% of an online interaction involves a human, it becomes uneconomic.”
Of course, there are also negative financial consequences if companies lose customers because of poor online interactions, including those in which accessing a human agent proves difficult. “It’s not just a cost equation,” says Gillon. “It’s about building and maintaining trust with customers.”
Acceptance of machines around the globe
Where are consumers most accepting of AI-led interactions with brands? The survey results suggest that acceptance is higher among US consumers than elsewhere. For example:
- Half of US consumers in the survey say they do not usually need a human agent to help them get customer support—considerably more than in Europe (43%) or Asia-Pacific (38%).
- More respondents in the US (47%) than in the other regions also say it is possible to have what they define as a human connection in a fully automated interaction.
- When asked about the brands they feel most positive about, 52% of US consumers say the interactions they have with those brands tend to be mostly or fully automated. For Asia-Pacific consumers it is 45%, and among European ones it is 42%.
Attitudes toward automated interactions appear to be changing faster, however, in Asia-Pacific, where 55% of respondents say this is a more positive experience than it was two years ago.
Consumers in Asia are generally comfortable interacting with bots, says Millie Gillon. But she says the picture is more complex in financial services. “When engaging in many types of banking, people in this part of the world need a relationship with a human in order to build trust,” she says. “That need for high-touch will diminish, though, as Millennials and Generation Z form a larger part of the banking market.”
Regional nuances in views about humans and machines?
Figure 7: Share of respondents who express agreement with a range of statements about automated interactions with companies.
Few online interactions between consumers and brands are problem-free—even if they have a positive outcome. When consumers have what they deem to be a negative interaction, there are two common sources of annoyance:
- A slow response in interactive conversations (29% of respondents)
- The inability to speak or chat with a human agent (28% of respondents)
Consumers are growing increasingly impatient with such problems in online interactions, observes Michelle Batt. “The digital quality of interactions may be improving, but consumers want brands’ responses to be extremely fast and relevant,” she says
Consumer expectations about the quality of online interactions in her industry are growing, says Nationwide’s Amy Shore, partly due to the superior experiences they are having with other types of firms (such as e-commerce companies). However, she adds, it usually takes a lot to lead a customer to switch brands.
That holds true for our survey sample. When the above types of issues occur, consumers are generally forgiving. Most of those in the survey will either continue with the interaction, in some cases all the way through to a purchase, or they will shop around before returning to the company’s website. A small percentage—20% in the case of delayed response time and 27% if they are unable to speak to an agent—would stop interacting with the company altogether.
Only 13% of respondents say that a company obtaining information about them from a third party is a factor in a negative interaction. But this is the top reason for consumers to cease interacting with a brand completely: 34% of respondents (and 52% of those aged between 55 and 65), say they have done this. This is consistent with a finding in our earlier study on consumer trust in data practices, in which 32% of consumers said they would cease interacting with a company if they learned it had shared their personal data without their consent.1
“When people believe that their data hasn't been cared for, the distrust engendered by it will often lead people to leave and not return,” says Verizon’s Michelle Emerson-Russell.
Companies’ use of third-party data is on the way out, according to Volkswagen Group’s Michał Szaniecki. “Now they need to focus the journey on their first-party or zero-party data,” he says, “which are more transparent and provide a better foundation for the customer relationship.”
When consumers cease their involvement with brands
Figure 8: Share of respondents saying they have ceased interacting with a company due to different problems encountered in online interactions.
There is another aspect of online interactions that will lead many consumers to rethink their association with a brand: lack of transparency about the role of AI. About two-thirds of respondents (65%) say it is important for companies to indicate, in an interaction, that they are communicating with an AI-powered bot and not a human being. This attitude does not vary much by age.
Fortunately, that kind of transparency is more the norm than the exception: 53% of respondents say that in the past 12 months they have been given the option (sometimes or often) to decline the use of AI. Should such an indication not be forthcoming, however, 41% would reduce their interaction with a company (17% ending the association entirely).
“The biggest mistake with early chatbot implementation was bots pretending they were real humans,” says Anna Noakes Schulze. “What was meant to seem friendly to people made them feel deceived.”
Instead, she says, brands should practice radical transparency: “Tell customers that the chatbot wants to help but sometimes makes mistakes. Enlist customers as allies to help the chatbot learn and do better. People can be very generous and forgiving when you tell the truth and ask for their help.”
Only 13% of respondents say that a company obtaining information about them from a third party is a factor in a negative interaction.
Building the human connection
Our survey shows that a continuing role for human sales, marketing or customer service staff in online interactions is assured for now. Consumers are happy for technology to handle most of the interactions they have with brands, but they want a human agent to be available if an interaction becomes complicated or problematic.
The business reality is that brands cannot afford for human agents to comprise a large part of interactions that are designed to be fully or mostly automated.
There is an obvious takeaway here for brands. They need to ensure that their chatbots and other intelligent tools continuously become better at handling not just the basic issues that users of their product typically have, but also the more complex ones. By doing this, they can reduce the need for human intervention.
The good news is that AI and machine learning are becoming capable of this. As the tools that AI and machine learning power become better at understanding nuance and sentiment, automated interactions will take on more human-like characteristics. Many of our youngest survey respondents already see these as human connections, and their number is likely to increase. If brands don’t try to pretend these are real human-to-human experiences, they will profit from the improved customer experiences they create.
Three features of good online interactions
- Clarity. Ensure that the option to click through to a human agent is visible to users throughout their journey. Difficulty locating the option is almost as bad as not having it at all.
- Transparency. Do not try to convince consumers that they are interacting with a human when they are not. They may not boycott you altogether, but they are unlikely to invest more in the relationship.
- Trust. Transparency is part of building trust, but it is not all of it. Trust also comes from consumers’ faith in the ethical sourcing and use of their personal data, as well as its protection.
Appendix: About the research
The analysis in this report is based on a survey of 5,601 consumers conducted in June 2021. The survey was carried out by Longitude, a Financial Times company, on behalf of Verizon.
The respondents live in 16 countries and territories, are aged between 18 and 65 and are evenly split between males and females.
Figure 9: Survey respondents by country.
Figure 10: Survey respondents by age group.
Figure 11: Survey respondents by gender.
Education (highest level reached)
Figure 12: Survey respondents by education (highest level reached).
Let's get started.
Choose your country to view contact details.
- Select Country...
- Costa Rica
- Hong Kong
- New Zealand
- United Kingdom
- United States
Call for Sales.
Or we'll call you.