Bots in CX: A Love-Hate RelationshipBots in CX: A Love-Hate Relationship

On Thursday, October 17, 2024, Sinch published its Future of Customer Communications survey which covered a wide range of topics, including consumer awareness and perception of chatbots, technology and device use, use of Rich Communication Services (RCS), and their perspectives on various types of interactions (sales, service, etc.) with businesses. The survey was conducted by Propeller Insights in September 2024, of 1,000 adults, gender-balanced and distributed across age groups from 18 to 65+ in the United States. Propeller Insights is a market research firm based in Los Angeles.

Yer A Bot, ‘Arry

The following chart shows that 63% of respondents said they could detect a chatbot when it was being used in customer service interaction. Sinch did not specify any parameters around the definition of chatbot but rather used it as a broad term to refer to automated customer service support agents. Forty-seven percent of respondents said they figured out it was a chatbot by the language it used, how quickly it responded (26%) and because the answers given didn’t fit the questions they asked (27%).

Gen Z (18-27) and Millennials (28-43) were, by a few percentage points, the best at detecting chatbot usage, followed closely by Gen X (44-59). Boomers (60+) were 50-50 at knowing when they were interacting with a chatbot.

In her presentation at Enterprise Connect AI 2024, Robin Gareiss, CEO & Principal Analyst with Metrigy, discussed similar findings. As the following leftmost chart shows, about 29% of respondents to Metrigy’s survey said they had interacted with a voice bot, 45% were unsure and 26% said ‘no.’ The rightmost chart shows that when the respondents were transferred, 56% were sometimes transferred to the right place. (This question was only asked of those who said they knew they talked to a voice bot.)

Consumers Don’t Like Bots

The following chart illustrates one of the business-consumer disconnects Gareiss spoke about in her presentation: 61% of respondents expect to speak with a human but 47% of consumers start their interactions in self service, which is often a voice or chat bot. Eleven percent of companies said that call containment – meaning the call (or chat) is ‘contained’ within that system – is the main reason why they are using bots.

“So even though consumers are saying, ‘I expect to talk to somebody,’ the businesses are basically saying, ‘We don't want to talk to you,’” Gareiss said. “And, by the way, adding voice bots doesn't mean that a customer won’t talk to somebody. The bot could be doing triage – what do you need, what language, etc. – and then they get transferred to a person.”

Gareiss said consumers aversion to bots may stem from businesses seeing the potential of early AI and more recently, generative AI, and then improperly training and/or misapplying their first-gen AI bots. “A lot of companies didn't really worry about their underlying data or their knowledge bases. As a result, the bots were basically looking at all these different data sources and not knowing which data was right,” Gareiss said.

This led to wrong answers and/or invented responses which soured the customer’s experience with the company. And as discussed here, Gareiss’s research (corroborated by a report from Qualtrics) also found that many customers simply stopped doing business with the company in question.

“I think one of the most straightforward ways to get AI out there, in a good way, is for businesses to showcase their successes. Say I'm a customer and I'm on a chat bot that just solved my problem, and [the system asks] ‘Did we solve everything?’. I say, ‘Yeah.’ The system then responds with: Wonderful! We solved your problem today in two minutes and seven seconds. Had you called into our contact center today, based on current volumes, it would have taken 14 minutes and 10 seconds,” Gareiss said. “If you do that every time that someone uses a chat bot, people will start to internalize that the chatbot just saved them 12 minutes. I think we’ll start seeing that approach and as customers have more good experiences that will help change their perspective on chat bots.”

With that in mind, the Sinch survey found that 50% of respondents expect that AI chatbots will eventually replace human agents. Millennials (28-43) were the most pessimistic – or optimistic, depending on your viewpoint – regarding this potential effect of AI. But Gen X (44-59) and Gen Z (18-27) were only a few percentage points behind them.

This perspective changes a little when trust enters the equation. Forty-three percent of respondents said they wouldn’t trust AI responses within their customer service conversations. Boomers (60+), Gen Z (18-27) and Gen X (44-59), respectively, were most likely to distrust AI responses. Millennials (28-43) were the most likely to trust AI responses.

The following chart shows that 80% of Sinch survey respondents were ‘somewhat’ (39.8%) or ‘not at all’ (40.3%) likely to trust a business that uses chatbots to handle all customer service interactions. To some extent every business has automated their customer service function, but it is unlikely – today, at least – that any company has automated every aspect of their service/support much less standardized on chatbots.

Gen Z (18-27) and Millennials (28-43) showed the highest propensity (very likely or somewhat likely) to trust a business that uses chatbots for all customer service interactions. Gen X (44-59) and Boomers (60+) exhibited lower comparative propensity to trust such a business.

Metrigy’s research below (right-side chart) suggests that consumers are conflicted in their use of chatbots: 47% use them, 41% avoid them and 13% prefer them. The left-hand side charts shows why chatbots are avoided or used. The top two reasons why consumers avoid chatbots are because the bots don’t understand what the person is asking, and they don’t have the answers the person needs.

Gareiss said those two issues likely result from improperly or insufficiently training the AI model and/or applying the model to the problem. “But getting stuck and not being able to escalate? That's bot design. No one should get stuck in a loop these days,” she said.

Consumers use chatbots because they can save time but note that one of the reasons why consumers avoid bots is because they take too much time. Convenience, answering questions and solving problems are all essentially tied for the secondary reasons. Gareiss suggested that this likely comes down to consumers being savvy enough to know when a chatbot can handle their issue versus needing to speak with an agent.

“But the last one – they're more polite than humans. Bots will never get angry. But humans? Say you work for an airline, there's a weather issue, and you have customers yelling at you. [We] have emotions. Are you going to stay polite after your fifth straight call? You might be a little bit frazzled at that point,” Gareiss said.

But Businesses Like Bots

The business perspective is different, with 80% of companies are either using or planning to use bots at this point – and 45% of companies already view bots as part of their workforce. “As they schedule and do capacity planning, [many] companies look at bots and humans together,” Gareiss said.

Sinch’s survey found that 47% of consumers believe AI/chatbots will dominate customer service in the next 10 years. Interestingly, that perspective dominates all generations (Z through Boomers). Gen Z (18-27) indicated that chatbots/AI-driven communications. All generations also indicated that they think SMS/texting will be the “#2” communications method – a view that increases by several percentage point grows if RCS is added to the total. (RCS is basically a better version of SMS/MMS.)

Note, further, that AI can of course be used across all these communication methods, as well as more mundane automations (e.g., order/shipping confirmations, etc.). Gareiss noted that while AI-powered agent assist is not customer-facing (like a bot is), those solutions directly help the agent and indirectly assist the customer by resolving their issue more quickly and/or accurately.

Metrigy found that virtual assistants/chatbots/voicebots were the most planned CX transformation project, followed by analytics and self-service capabilities. Analytics can help companies figure out whether what they’re doing is working, while self-service capabilities could include customer-facing knowledge bases or other solutions that are not bot-based. Metrigy defines a CX transformation as the application of a new or existing technology to improve the agent and/or customer experience that ultimately results in measurable business improvement.

“Voice of the Customer programs are important. You can speak with your customers and [maybe ask]: ‘Should we be using bots?’,” Gareiss said. “If you get a lot of complaints, maybe you should rethink what you’re doing.”

Improving CX with Bots

Metrigy also asked consumers to identify what business could do to improve CX, as illustrated below. Hiring more knowledgeable people (60%) and reducing hold times (57%) were by far the top two results.

Gareiss noted that AI-powered solutions can be used to reduce (or perhaps eliminate) hold times via self-service. Agent Assist can be used to continuously train agents and even make those people appear more knowledgeable than they might be (via summarization of answers, Gen AI-generated scripts, etc.).

Gareiss said that Metrigy has developed an algorithm that allows them to correlate in their surveys what successful firms are doing compared to others. Metrigy’s ‘success group’ shows higher than average measurable improvements in their business metrics by certain technologies, namely AI and more successful CX transformations.

The following chart illustrates that while 78% of consumers say people are more important than technology, among Metrigy’s success group those companies investing in technology are seeing those ‘above average’ improvements in their business metrics. “When companies deploy technology in the right way, it can become even more important than people, because it can empower people and enable people – and empower your agents to just do a better job,” Gareiss said.

But with respect to voice and chat bots, Gareiss advised companies to spend time building them, training them, designing them and applying them to the right problems.

“We can't really move forward as an industry with AI when only about 60% of your customers are willing to use bots,” Gareiss said. “I'm not saying not to use bots. I'm saying you need to use them properly. You need to educate your customers on why they should use them.”