Contact Center

Close to half of consumers — 46% — say AI-powered customer service either “rarely” or “never” leads to successful outcomes, according to a survey conducted by YouGov on behalf of Pegasystems Inc. released in February.

Nearly two-thirds of consumers say they are "not very confident" or "not at all confident" with how companies use generative AI to interact with customers, according to the survey of more than 4,700 consumers in North America and the U.K. Just over half aren’t confident that businesses can use the technology responsibly.

Over three-quarters of consumers say only dealing with a human “always” or “often” leads to better outcomes in customer service. Only 2% say they want to interact exclusively with generative AI chatbots.

Business leaders who have invested in AI often tout the benefits of the technology in customer service. But those reported successes are at odds with consumer sentiment toward AI support.

After Airbnb rolled out its AI agent ahead of schedule last year, CEO Brian Chesky said that it “can do an incredible job of customer service.” Last fall, Williams-Sonoma introduced AI agents across its brand portfolio to increase efficiency, reduce service costs and offer 24/7 support.

However, most consumers aren’t as excited by the technology as business leaders. Consumers are wary of fully automated customer support, though they are more open to human-led service in which AI exists in a supporting role.

Two-thirds of respondents say they prefer human-led support, while just under half — 48% — of consumers say they don’t trust companies when they use AI to completely handle customer service queries, according to Pegasystems. 

The real power of AI is in helping live agents follow a measured plan, according to Julie Geller, principal research director at Info-Tech Research Group. The technology can help teams act in a unified manner by helping them find the right next-best action without sacrificing the human touch.

“By removing the inconsistency where a customer’s experience depends on which agent they happen to get, we take the guesswork out of the interaction,” Geller said in an email. “The result is a reliable, polished journey where the customer feels the brand’s intent in every touchpoint.”

One company taking a people-led approach is Lowe’s, which equipped its associates with the Mylow AI assistant for use in customer interactions. This resulted in “dramatic improvements in customer service,” CEO Marvin Ellison said on a February earnings call.

Consumers are wary of their own use of generative AI as well, though they recognize its growing presence. Nearly half of consumers — 48% — say they don’t actively choose to use it in their everyday tasks, though one-quarter suspect they use it daily without noticing.

Consumers are more comfortable using the technology for research than making a purchase, according to Riskified research from last fall. Top concerns regarding autonomous AI purchases included payment security, potential mistakes and loss of control.

Companies are investing in AI to cut customer service costs and improve self-service options, but a prediction from Gartner throws into question the assumption that the technology is cheaper than humans.

The consultancy predicts that costs per resolution for generative AI will exceed $3 by 2030 — a rate that’s higher than many offshore agents.

Gartner points to rising data center costs, AI vendors pivoting from subsidized growth to seeking profitability, and complex use cases as reasons for AI costs to soar. 

Return on AI investments “is far from guaranteed,” according to Patrick Quinlan, senior director analyst in Gartner’s Customer Service and Support practice. “Full automation will be prohibitively expensive for most organizations; instead, leading organizations will use AI to drive customer engagement rather than to cut costs,” he said in a prepared statement. 

Though consumers’ use of AI tools has grown, many continue to be wary of the technology. Still, some brands are plowing ahead, betting that the money saved from AI customer service will be worth any resistance.

But AI isn’t the cheap technology it’s often made out to be. AI tools come with unanticipated ancillary costs, such as managing access credentials for autonomous agents or acquiring new data sets. 

“Historically, we have invested in less expensive technology to replace more expensive talent, and currently what we are looking at is investing in a more expensive technology solution to replace a less expensive talent source,” Emily Potosky, senior director of research at Gartner, told CX Dive in February. “And I do not know economically if that is going to pay off.”

Beyond costs, customers want the right to talk to a human, especially for complex or sensitive issues. 

Kinsta, a company that specializes in WordPress hosting, is hiring more customer service representatives for that reason, according to Roger Williams, community manager at Kinsta.

“There's so many different ways things can go wrong, especially with WordPress where you can add plug-ins and themes, and you can easily customize the code without knowing anything about code,” Williams said.

The ability to reach out to a human representative and say, “I did something. I don't know what I did. Help,” and have someone figure it out quickly is a valuable service, Williams said. “Your day is better in minutes.” 

That’s not to say AI isn’t a useful tool. AI can be a differentiator, but it will cost money. 

Businesses that lead AI adoption will “look beyond cost optimization to other benefits, such as increasing customer lifetime value, repurchase rate and brand loyalty,” Quinlan said.

Sponsored content

By CallMiner

Most organizations are deploying AI voice agents with a narrow objective: reduce contact center costs by automating inbound calls.

That approach captures some value, but it misses the strategic opportunity.

AI voice agents, combined with conversation intelligence, do far more than automate existing conversations. Properly deployed, they change which conversations happen at all, when they happen and where they occur. They fundamentally shift customer engagement upstream, often eliminating the need for inbound contact altogether.

For executives, this is not a technology question. It’s a demand, risk and experience design question.

To understand the real shift underway, it helps to look at two closely related ideas:

1. AI voice agents enable conversations that were previously impossible or impractical
2. AI allows customer engagement to move outside the traditional contact center

The limits of the traditional contact center model

Contact centers are inherently reactive. Conversations only occur if:

• A customer encounters friction, confusion or failure
• The customer decides to initiate contact
• The expected value of the interaction justifies the cost

Inbound volume, therefore, is not a neutral workload metric—it’s feedback. It is a lagging indicator of upstream breakdowns across product design, communication, operations and policy. Additionally, countless potentially valuable conversations never happen.

Historically, companies have accepted this model because the alternatives—proactive, continuous engagement—were too expensive and too complex to scale with human labor.

That constraint no longer exists.

AI voice agents enable conversations that were never economical before

AI fundamentally changes the economics of customer interaction:

• Near-zero marginal cost per conversation
• Always-on availability
• Consistent execution at enterprise scale

This makes it viable to engage customers in ways that were previously impractical or unjustifiable, including:

• Proactive outreach when anomalies or risks are detected (before customers experience a problem)
• Preemptive explanations before customers become confused (micro-engagements such as confirmations, reminders, or explanations)
• Timely confirmations and clarifications that reduce anxiety (through behavior or event driven outreaches)
• Engagement in any language, independent of agent staffing constraints

These are not automated versions of existing calls. They are net-new conversations that materially reduce downstream demand, churn risk and brand damage. This approach will fundamentally change the customer experience.

From an executive perspective, the true value extends beyond cost savings. It lies in preventing risk, preserving trust and creating frictionless experiences that strengthen customer relationships.

Shifting engagement outside the contact center

Instead of waiting for customers to reach out, organizations can engage earlier in the journey, embedding support where friction actually occurs. This shifts engagement from:

• Centralized to distributed
• Reactive to preventive
• Episodic to continuous

When AI voice agents are deployed proactively and contextually, support is no longer something customers “reach out for.” It becomes something embedded into:

• Moments of change
• Points of uncertainty
• High-risk transitions
• Behavioral signals indicating friction

As a result, many traditional inbound calls simply never occur. This is not deflection. It is demand elimination.

Three mechanisms drive this shift:

1. Proactive communication that removes uncertainty before it becomes a problem
2. True self-service that resolves issues end-to-end, not just reroutes them
3. In-journey assistance that appears precisely where friction emerges

The cumulative effect is a material reduction in inbound demand, not because customers are blocked, but because they no longer need to ask.

From reactive to preventative: The intelligent CX automation journey

Most AI voice agent deployments, as part of larger automation programs, focus on speed and volume, reducing handle time and increasing containment. Those gains matter, but they are fundamentally reactive. They optimize the response to customer problems rather than getting ahead of or eliminating the problems themselves.

Preventative engagement requires a different operating mindset. That’s where the concept of intelligent automation comes in. By continuously learning from the signals customers provide across every channel of interaction, organizations can not only better understand how, where and when to automate interactions – such as through AI voice agents – but they can also prioritize precision, prevention and trust.

This shift depends on a deep understanding of why customers contact you. AI-driven classification and journey analytics reveal the real drivers of demand, not just the surface-level topics, making it possible to identify the friction points, misunderstandings and failure modes that repeatedly push customers into the contact center.

These customer signals shouldn’t be contained within the contact center operational silo. Intelligence must move horizontally, informing product, marketing, billing and compliance. Prevention becomes possible when upstream teams anticipate and fix the conditions that generate downstream support demand.

Organizations typically progress through four stages as they move from reactive automation to preventative engagement:

Stage 1 — Reactive Automating what is already happening

Self-service resolves simple, well-defined issues after customers seek help.

Example: A billing dispute is handled through automated self-service.

Stage 2 — Intent-aware Understanding the “why” behind inbound contact

Systems detect intent, emotion, frustration, churn risk or compliance exposure and route accordingly.

Example: High-friction interactions are escalated to specialized agents.

Stage 3 — Predictive Anticipating needs from patterns, scoring, and historical behavior

The organization identifies which customers are likely to escalate and intervenes before the call.

Example: Proactive outreach offers resolution options before disputes arise.

Stage 4 — Preventative Eliminating the need for inbound contact entirely

Product, communication and policy changes address root causes so that customers never encounter the issue.

Example: Billing accuracy and clarity improve to the point that disputes disappear.

A practical industry example

Consider a utility or telecom provider.

Inbound call spikes are often driven by usage anomalies, billing changes or service disruptions; situations where customers are uncertain rather than broken.

With intelligent automation and AI voice agents, the organization can:

• Proactively notify customers of abnormal usage and explain why
• Communicate service disruptions before customers seek updates
• Resolve billing questions immediately through conversational self-service

The economic impact is not limited to call avoidance. It includes reduced churn, fewer escalations and improved customer trust during moments of stress.

What this means for executive leaders

When engagement moves upstream, the role of the contact center changes.

Human agents become the exception layer, focused on complex, emotional or high-stakes interactions. AI handles prevention, scale and timing.

This shift requires new success metrics:

• From cost per call to cost of preventable demand
• From handle time to issues avoided
• From efficiency to journey integrity

It also expands ownership beyond the contact center. Product, digital, operations and risk leaders all influence how and if customers need to engage at all.

The future of customer engagement

The next generation of customer engagement must focus on this concept: “The highest-value conversation is the one the customer never needs to have.”

The leaders and organizations that treat AI voice agents as contact center automation and engagement infrastructure will not just optimize yesterday’s operating model – they'll redesign how customers experience the business, reducing demand, mitigating risk and building trust at scale.

The people who know best about customer pain points aren’t in the boardroom, according to Shuvankar Roy. They’re on your front line. 

Roy, who joined Optimum as its first-ever chief customer experience officer in 2022, said the company has improved its net promoter scores over 40 points in the past three years. But competition is as stiff as ever.

“That being said, we have much more competition in the market than what the cable industry used to do,” Roy said at Reuters Customer Service and Experience East in November. “The only distinguishing factor is, from my point of view, with customer experience. If you do not make customer experience right for the customer, they have other choices, and they will make it.”

Finding customer pain points and fixing them is key, according to Roy. Optimum goes about that in two ways: listening to the frontline and analyzing what detractors have to say about its brand.

“I learn most when I'm in a call center in the Philippines or in India or in Brooklyn than what I hear through a bunch of PowerPoints in the boardroom,” Roy said. “And I think that's the starting point. Talk to the front lines. They know most what's not working.”

Roy makes a point to visit these contact centers in person. Generally, business process outsourcing companies look to “roll out the red carpet,” but Roy looks to “reverse completely the culture.”

Roy looks for feedback on his first day visiting a BPO, and it can be “very uncomfortable for my team who travels with me,” he said. “The first day is, ‘Tell me what my team haven't done,’ and the feedback that we get is paramount.”

Perhaps the call center has been asking for a process change or a tool deployment for months. He takes that information and assigns it to a scrum team in New York with a deadline. When they deliver, the frontline understands Optimum will follow through on what it promises, Roy says.

“The next time you visit the same BPO, it’s a different level of energy in the frontline,” he said. “This cultural transformation is much more transformative than adopting a new technology, and pushing it down.”

Optimum also examines its NPS scores. The company takes the time every three months to celebrate the progress, but it primarily analyzes the detractors and uses identify patterns among detractor comments. 

Roy shared an example: a common pattern among detractors was technicians running late.

“What we figured out is this running late is also dependent on where the technician is, because with some footprints of the U.S., you have to travel for an hour to get your technician while if you are in Brooklyn, you might be traveling the 10, 15 minutes,” Roy said.

Optimum has started providing customized messaging to the customer, given the scenario to help manage expectations.

Transformation begins, he says, with front line feedback and listening to the customer through NPS. “Take the package, and build on it the pillars of opportunities.”

One in 5 customer service and support leaders have reduced agent staffing due to AI, according to a Gartner survey of over 320 leaders.

While businesses are handling more customer inquiries, just over half of leaders — 55% — report maintaining staffing levels. 

Rather than simply replacing human staff, most businesses are focusing on using AI to aid representatives and boost their efficiency, according to Gartner. 

Businesses, from Atlassian to Salesforce, have laid off customer service workers, pointing to the efficiencies gained by AI.

Salesforce, for example, reduced its customer support workforce from about 9,000 people to about 5,000, CEO Marc Benioff said in 2025. He credited AI agents for handling “millions of conversations” on an earnings call. 

Atlassian laid off 350 customer service and support workers in 2025 and its CEO told employees that customer service issues would be largely addressed with AI.

But Gartner’s research suggests more businesses are using AI to address more customer service inquiries — and are keeping headcount stable or increasing AI expert roles. Two in five leaders are hiring specialized roles to better implement AI in customer service.

“Some leaders are seeing headcount reduction, but it's often indirect,” Melissa Fletcher, senior principal of research in the Gartner Customer Service & Support practice, said in an email. “Some leaders report that AI has made it possible for them to pause the backfilling of agent roles — allowing them to reduce headcount over time.”

And some businesses are pursuing multiple strategies at the same time; for example, hiring specialized staff while pausing the backfilling of some agent roles.

Once a business decides to invest in a contact center, the question becomes where to put that operation. 

Offshore locations in India and the Philippines have long been front-runners, but countries in Africa rising in popularity. Legislative momentum to compel onshoring, though currently insignificant, ebbs and flows in Congress. 

To Maureen Burns, a partner in Bain & Company’s customer practice, choosing one place over another depends on what aligns strategically. 

“An offshore contact center can be just as good as an onshore contact center. It's about how much time, effort and investment you put into it,” Burns said. 

Samsung Electronics America chose to open a contact center in Greenville, South Carolina, in October in partnership with Harte Hanks, a global customer experience company. The new Customer Care center supports more than 150 jobs and is now the hub for its customer service arm, Samsung Care. The facility is 14,000 square feet and includes a training lab to carry out ongoing upskilling initiatives and professional development for employees. 

Samsung has support teams offshore, too, and they drive key performance indicators, said Marcia Ellis-Green, executive director of customer support at Samsung. Agents in Greenville focus on more complex customer inquiries in tier 2 and tier 3. It’s a strategy backed by a strong business case, Ellis-Green said.

“Greenville is our circle of excellence,” Ellis-Green said. 

Proximity breeds synergy

Typically, cost is thought of as a significant factor in deciding whether to look overseas for contact center operations. A COPC survey from 2022 found “cost reduction” is the top reason companies choose to outsource contact center operations. 

While labor is still cheaper in many offshore locations today, costs are rising, meaning it plays less of a role in decision-making than it used to, Burns said. Factors such as the types of calls anticipated, call volume, ownership of data from those calls, and training required to handle those calls may now be prioritized. 

“When I put that all together, what's the right cost-benefit analysis?” Burns said. 

The answer depends on the company. Important to Samsung in its Greenville decision was synergy. Ellis-Green’s team, which handles customer support operations, was already in Greenville. With the new facility, agents and operations support can now meet with each other by walking across the parking lot, she said. In the case of new technologies or other solutions, the support team can bring them directly to the call center. Additionally, the teams can participate in workshops together. 

Though unfamiliar with Samsung’s strategy, Burns noted proximity isn’t just beneficial for customers. A “unified experience for employees,” as she called it, is its own advantage. 

Indeed, Ellis-Green said it’s easier for both customer service and operations colleagues to walk to meet with each other instead of taking a flight across the country, or even a flight across the world. That’s important for a call center responsible for more complex and technical cases, which Ellis-Green said are moments “that change customer loyalty.”

This concept becomes especially true as an increasing number of general inquiries are handled through digital channels without a human agent. A 2025 report from AI-powered agent platform Forethought found companies deflected one-third of their tickets with a digital solution, with or without AI. That’s an increase over 2024’s finding of about one-quarter. 

Though the trend is not currently ubiquitous, it has CX leaders thinking about where human interactions can differentiate them from their peers, Burns said.

“I think it … raises the stakes for those fewer calls,” she added. 

A balancing act

Ellis-Green has worked in the contact center space for more than two decades. In her experience, underpinning any strategy for consumer-facing industries is the pursuit of balance. And “balance” looks different for different companies, she said. 

Samsung has agents supporting millions of customer calls every day, as well as teams working in the same time zone close to support staff. While outsourcing some aspects of the contact center may make sense, Ellis-Green said she has an entire team in-house, working on escalations, processes and procedures. 

And balance can also pertain to the offerings a customer service team provides to its customers. Ellis-Green said her husband prefers to fix things himself, and so prefers self-service resources such as ordering a part online. But customers who would rather have a technician come to their house also have access to that option. 

The strategy requires looking ahead to what customers want from customer service and what the company can provide to meet those needs over the next few years, she said. 

“Where do we see that there is a gap or a new technology or a new special team that would need to be created to support them?” Ellis-Green said. “And we'll always continue to look at, where is that best located? Does it make sense to be here or does it make sense to be outside of the U.S.?”

Voice AI can help contact centers unlock their potential, Nicole Kyle, managing director and co-founder of CMP Research, said during a Krisp webinar. Interest in automated customer support may lag behind live support, but it’s growing every year among contact center leaders. 

While consumers are interested in automated voice interactions, they often fall short of expectations, driving customers back to traditional self-service portals, according to Kyle.

“Given that voice is the most popular channel, it follows that if you're going to win big on automation, it's by offering an excellent voice AI experience,” Kyle said.

Contact centers need to shape their automated voice AI to match what customers want from the technology.

“Voice AI tech is getting better, and there's an opportunity and there's demand to be captured, but it really depends on the quality of that experience to get consumers to change their behavior,” Kyle said.

Only about 3% of consumers cite interacting with a conversational voice AI agent as their top preference for self-service, but that number could be higher if voice AI met consumer expectations, according to Kyle. Self-service portals are the most popular option, followed by IVR — a preference born of habit.

A good first step for rolling out voice AI technology is to find the subset of customers who prefer call-based customer service and are also open to automated self-service, according to Kyle. Companies can connect these customers with their voice AI option, rather than try to move everyone onto a new path.

Leaders also need to make sure they’re investing in the right technology, too. Conversational voice AI doesn’t necessarily mean generative AI, and leaders should invest in a solution that makes sense for their needs.

“Depending on the maturity of your contact center, and depending on the appetite of your own customers, sometimes a really, really great conversational AI tool is all you need for now,” Kyle said.

Some companies are embracing voice AI outside of the contact center to improve their overall customer experience. Red Lobster rolled out voice automation systemwide to handle its phone takeout orders, freeing up workers while improving front-of-house service.