• October 4, 2022
  • By Erik J. Martin, freelance writer and public relations expert

How Reliable Is Your Contact Center's AI Guidance?

Article Featured Image

Artificial intelligence has infiltrated many departments, processes, and operations in companies large and small over the past few years. But one area where it has proved to be particularly helpful—and exasperating when it doesn’t work as intended—is the contact center, often staffed by human agents who provide support to customers over chat, email, phone, and other platforms.

Accelerating customer expectations for instant gratification (fueled largely by the COVID-19 pandemic) helped drive the digital transformation of the contact center and the rapid adoption of AI solutions, according to Andy Traba, head of product marketing at NICE.

“Today, AI-powered smart self-service and attended automation is increasingly enabling organizations to improve customer service and strengthen loyalty—freeing agents to focus on more complex, higher-value tasks,” Traba says. “In doing so, AI unlocks a wide range of benefits, from reduced costs and improved employee retention to higher revenue and increased customer satisfaction. Self-service solutions that actually work are also equipping many contact centers to cope with spikes in volume during the pandemic and peak demand periods.”

Erik Ashby, head of product at Helpshift, says that only recently have we observed broad successful use of artificial intelligence in contact centers.

“Early AI attempts tried to mimic and replace the contact center agent. This approach often would lead to frustration, as the AI would become confused and/or the consumers knew that they were not being handled by a human but rather an imitation,” Ashby says. “Recently, however, brands began to take a more pragmatic approach to AI. They began to leverage AI as an assistant that works in conjunction with the agent to help guide the conversation or to provide the consumer with self-help guides that can work at any time of day or night.”

Considering how businesses increasingly require intelligent solutions designed to aid customer service agents and serve clientele faster, it was inevitable that AI would become integrated into contact centers as a crucial real-time assistance tool.

Problem is, this technology isn’t foolproof. Unreliability triggered by a myriad of factors can result in many negative outcomes, including customer dissatisfaction, agent irritation, and more.


The good news is that in 2022 AI can often be relied on to provide relatively trustworthy real-time guidance within the contact center, including scripts for handling calls, analysis of current interactions, upselling and cross-selling recommendations, and customer issue resolution suggestions.

“AI provides real-time guidance through interaction monitoring that instantaneously surfaces insights from every customer interaction and delivers personalized agent coaching in the moment,” Traba explains. “In addition, AI powers self-service tools, such as intelligent chatbots that can address common customer concerns before they reach the agent. When agents are on a call or digital interaction, AI tools help predict and flag knowledge resources and next best actions to streamline issue resolution and eliminate friction.”

Traba points out that AI also helps agents use their time efficiently through the elimination of redundant tasks that can now be automated and reduces agent stress through self-service tools that empower customers to address easier problems on their own, reducing the number of customers waiting in the queue.

“As a result, agents can complete more work, and consumers overall are more satisfied,” Ashby says. “Supportive AI can leverage rich consumer context with real-time solution information from the brand to create a unique solution and then guide the consumer.”

Case in point: Say a consumer enters a brick-and-mortar store but opens the store’s app on his phone to get help. The AI should be able to quickly suggest common actions for in-store assistance. If the patron types in a question, AI should be able to quickly comprehend the intent of the problem and either respond with some information, such as a location to find an item; route the customer to a central call center agent for human help; or contact someone in the local store to assist if needed.

“In this case, AI uses contextual information along with natural language processing to determine how to best solve the problem; then, it takes the appropriate action,” Ashby says.

Indeed, real-time guidance is ubiquitous nowadays, according to Trey Norman, chief operating officer of Mindbreeze, a provider of appliances and cloud services for enterprise search, applied artificial intelligence, and knowledge management.

“We see guided navigation and real-time guidance with chatbots based on natural language on about a quarter of customer service companies’ websites, and that requires no call center agent or human interaction,” he says. “The level of real-time guidance happening behind the scenes when you call your most used brands has skyrocketed in recent years with advancements in natural language processing and the ability to interpret the meaning of data. NLP can generate a response directly to the call center agent based on what the customer says on the other side of the call.”

For outbound communication from the contact center, AI can help in a variety of ways as well. It can, for example, provide sales teams with the right time to reach out to a particular lead based on a wide array of variables and the history of past interactions with leads of similar profiles and industries, notes Ram Ramamoorthy, global head of Zoho Labs at Zoho.

“The AI model can also offer contextual insight to push toward conversion based on past interactions and what that particular lead is interested in most. This helps sales teams tailor conversations to address the customers’ priorities first,” Ramamoorthy says.

For inbound communication, meanwhile, AI can offer a snapshot of the customer’s history with the organization so that the contact center agent has some background on the customer, helping to narrow down the possible reason for the call. This can reduce customer wait time and improve satisfaction. AI can also examine incoming information and assign the ticket to the appropriate agent depending on past interactions, customer categorization, and the nature of the problem.

What makes real-time AI recommendations and assistance in contact centers achievable and reliable? First, it’s helpful to understand what goes into making the technology.

Three of the key ingredients here are predictive modeling technology, which creates models that can anticipate future events; analytics, the process of extracting information from text or spoken data to identify keywords linked with customer satisfaction or to better understand customer sentiment; and NLP, which is used to comprehend customer queries or generate responses to them.

“Natural language processing and natural language understanding analyze customer interactions and provide recommendations. AI is also being used for intent detection in contact centers to understand the customer better,” says Anthony Chavez, founder and CEO of codelab303, a digital experience design agency. “This is done by analyzing the customer’s interactions and understanding their needs based on what they’ve said combined with the information they’ve provided historically.”

The most sophisticated AI solutions automatically turn large swaths of data—even trillions of interactions—into actionable insights.

“This includes every customer touchpoint, whether phone, chat, email, social media, app, chatbot, web inquiry, or Google search,” Traba points out. “Smart omnichannel routing gives organizations a rich understanding of customer preferences and matches them with the most suitable agent on the customer’s preferred touchpoint, using AI to assess emotion and intent.”

Furthermore, AI enables organizations to proactively reach out to customers with a highly customized experience, more quickly and effectively improving the customer experience by providing real-time feedback to agents.

“By helping them understand the behaviors that affect customer satisfaction—for example, showing empathy or building rapport—and giving them in-the-moment guidance, brands can give employees the support they need to drive meaningful conversations and continuous improvement,” Traba adds.


All of this raises a crucial question: Just how dependable is the real-time guidance AI provides to contact center agents? After all, technology designed to make their jobs easier can backfire if the data being processed by the AI is imprecise or erroneous.

“There is a concern over the reliability of the information AI provides. The reason is that artificial intelligence relies on data that may be invalid, inaccurate, or outdated,” Chavez warns. “Another concern is that AI has the potential to automate decision making. This could lead to call center agents following a script rather than using their judgment.”

Others echo those trepidations.

“Ineffective or inaccurate AI guidance can result in creating, rather than alleviating, friction, thus lengthening interaction times, frustrating customers, and lowering agent productivity. Separately, both customers and agents will rightly become irritated when simple issues like forgotten passwords or reservation changes need agent support because self-service tools were unhelpful or unavailable,” Traba cautions. “On the agent side, if AI cannot accurately surface the right information at the right time, agents are not equipped to resolve customer issues quickly and efficiently. This increases average handle time and agent frustration while simultaneously decreasing customer satisfaction.”

Also, ponder the possibility that AI could even cause problems with data privacy if personal information is mishandled.

Ashby believes real-time guidance via AI can be more correct and consistent if two best practices are followed.

“You must have a structured approach to conversational AI so that it has a clear path to a solution, and the AI must have a good understanding of the most common intents,” he says. “If either of these components is missing, it can generate frustration, as the AI can become confused and repeat itself or the AI may send a consumer down the wrong path.”

Ask Traba and he’ll tell you that AI-provided real-time guidance is currently highly reliable.

“That’s because AI and analytics help organizations learn from every interaction in real time to continuously improve the customer experience,” Traba says. He references a recent NICE study that found that 95 percent of companies reported growth in customer self-service requests in 2021, “meaning that more consumers trust the tools provided to help them resolve their issues without an agent’s help, saving customers time and the contact center resources.”

Ramamoorthy, on the other hand, rates the accuracy of AI real-time assistance at about 85 percent.

“This is why a human in the loop is always recommended to act on AI-enabled decisions,” he suggests.

Others strongly agree with that advice.

“AI should work in tandem with the contact center agent to provide real-time guidance, thereby enabling a frictionless and personalized customer experience,” says Michael Ringman, chief information officer of TELUS International, a provider of contact center and IT help desk solutions and services. “Additionally, organizations and their executives need to understand the nature of bias in AI and the risks and measures that need to be taken to mitigate it.”

Ultimately, the guidance AI provides to contact center agents can be extremely valuable and largely accurate, provided the organization doesn’t ask too much of the technology.

“With AI, what you put in is what you will get out. It needs to be given access to quality data above all else,” says Pieter de Villiers, CEO and cofounder of Clickatell.


Businesses can safeguard their call center operations and ensure better outcomes for agents and customers by striving to bolster the dependability, veracity, and consistency of AI technology and the information fed into it.

“To improve the reliability of AI, more data should be used to train artificial intelligence models. Also, data should be regularly updated to ensure that artificial intelligence models are using the most up-to-date information. And AI systems should be monitored regularly to identify any issues that may arise,” Chavez advises.

What’s more, companies should use a practical intent-driven workflow approach instead of unstructured AI chatbots.

“This allows the AI to capture the intent, then directs the conversation to a known workflow,” Ashby says. “For example, if someone asks, ‘How can I get a refund,’ the AI should automatically detect that the consumer is looking for a refund, then start a predefined flow that will help the consumer get a refund.”

Mindbreeze’s Norman includes proper data connectivity and research into solutions that use the most diverse AI techniques in his recipe for AI success.

“First identify a use case by asking which specific problems in your company need to be solved and why,” he recommends. “Next, define your success criteria, such as business needs, data quality, and defining key performance indicators. To ensure the most reliability, it is essential to test the solution with real company data and the actual people who will be using the solution. The users will be your best bet at getting feedback on the reliability because nobody knows the day-to-day processes better than them.”

Traba believes it’s essential to ensure a closed-loop feedback process is in place between AI-provided real-time guidance and the corresponding action-oriented systems.

“Connected intelligence, where real-time recommendations can trace back to actions and outcomes, affords AI a richer dataset to be trained on, thus increasing its reliability,” he notes. “With this feedback loop, and through increased exposure, AI will become more accurate at detecting and analyzing customer sentiment and intents, ultimately becoming more inclusive and effective.”

AI might not yet be capable of completely autonomously operating a contact center. But its current ability to augment human agent efforts and assist them in delivering quality service and information speaks to the promise and potential of technology to completely transform customer assistance.

“Artificial intelligence has already had an incredible impact, helping humans provide more informed service, be more productive, and experience less stress,” Ramamoorthy says. “This makes me very bullish on how AI can augment contact center agents in the years to come.” 

Erik J. Martin is a Chicago area-based freelance writer and public relations expert whose articles have been featured in AARP The MagazineReader’s DigestThe Costco Connection, and other publications. He often writes on topics related to real estate, business, technology, healthcare, insurance, and entertainment. He also publishes several blogs, including martinspiration.com and cineversegroup.com.

CRM Covers
for qualified subscribers
Subscribe Now Current Issue Past Issues

Related Articles

Interaction Analytics Helps Improve Coaching/Training

Data-driven guidance provides a better agent and customer experience.

As Contact Centers Become More Complex, Testing Grows in Importance

Tools to optimize customer service operations have grown in scale and depth of functionality.

The Top Customer Service Trends for 2022: New Service Channels and Challenges

Investing in digital, social, and virtual will remain a post-COVID priority.

Tools Can Now Uncover Real-Time Customer Behavior

Companies can respond in real time when they know what customers will do in real time.

Can AI Really Be Trusted?

How reliable are the marketing recommendations made by artificial intelligence?

Buyer's Guide Companies Mentioned