Keith Dawson's Analyst Perspectives

AI’s Value to Contact Centers: What Are the Use Cases

Posted by Keith Dawson on Oct 28, 2021 3:00:00 AM

In part one of this Analyst Perspective on the use of artificial intelligence within contact center applications, we focused on the evolution — and resulting benefits — of tools embedded with AI, including ease-of-use for non-data-scientists.

In this Analyst Perspective, we’ll explore the most beneficial applications of this technology in contact centers. To understand how contact centers typically enable AI, we’ve catalogued the most mature use cases available with current technology (besides chatbots).

Interaction routing. When contact centers were primarily vehicles for routing voice calls, there were universally accepted methods for forecasting volume and scheduling agents to handle that volume. Calls had well-understood characteristics: they started and stopped in predictable ways, lasted for certain knowable durations, and could be predicted with some certainty. Now that contact centers are multichannel, the comfortable predictability of interaction volume has disappeared. Emails, chats, messages and other customer communications can be of any duration, and are often asynchronous.

Multichannel interactions also create an enormous pile of data to analyze, making it one of the best available opportunities to put AI and machine language to work. AI is used to run historical interaction data through analysis, making complex cross-channel and multichannel interactions more predictable so that volumes can be forecast. Once forecasts are developed using realistic predictions, better schedules can be created. AI can also sift through an agent pool comprised of multiple skillsets to put the right agent in the right place at the right time.

Voice of the customer (and the agent). Voice of the customer and voice of the agent are among the best use cases for data analysis, with or without AI. Contact centers have a long history of using customer data to manage agent productivity, but the real value is derived when an organization uses it to listen for trends in customer sentiment, intention and buying behavior. This requires moving the data outside of the contact center, usually to marketers who have more experience using AI/ML.

Without AI, analyzing enormous stores of unstructured voice recordings is cumbersome at best, impossible at worst. That makes it a good candidate for cross-departmental deployments that foster the integration of contact centers into wider customer experience programs.

VR_2021_Voice_of_the_Customer_Assertion_2_Square (2)Ventana Research asserts that by 2022, one-third of organizations’ marketing departments will be larger users of VoC analytics than the contact center, using sentiment and behavioral insights to target customers. We expect having this kind of data will help foster a more organization-wide view of CX than is common today.

Knowledge management. Knowledge management is a way for AI to get into the contact center from the side. Contact centers exist to solve customer problems. To do that, the center needs information about products, customer histories, previously reported problems and solutions to those problems (both those that work and those that don’t). Systems that track cases and trouble-tickets provide most of the tools for managing that knowledge, which is often managed by IT teams with input from product developers, who also take input from customer communities, social posts and agent results.

Searching through that input can be time-consuming and expensive — so much so that many service organizations don’t actually make much effort to do it. That’s contributed to the perception of customer service as minimally helpful and frustrating. And it opens the door to using ML to detect trends in customers inquiries and match those questions to all of the possible solutions an organization has, even if they are separate and siloed. AI has breathed new life into KM, which was a moribund sector until data analysis raised its profile.

Agent assist and guidance. Imagine two scenarios: In one, a caller poses a question to an agent, and the agent types what the customer has asked to query a database and find the answer. In another scenario, an automated system is listening along with the agent and automatically displays a selection of the most likely relevant responses to the customer. Showing agents the next-best action or presenting detailed responses based on the query — in real time — is one of the most revolutionary contact center operations advances to come along in decades. It shortens interactions, makes responses more consistent from one agent to another, and reduces friction for the customer. Plus, it provides agents with a place to start that has a higher likelihood of being the correct answer than previous methods.

Agent evaluations. Contact centers record nearly 100% of interactions. When supervisors select calls to evaluate for quality control, they are often choosing at random, or based on a gut sense that a particular agent has trouble or needs help. Modern agent management applications use AI to scan the entire corpus of recordings and select those for evaluation that meet certain defined attributes. This makes the supervisor’s job easier, allowing him or her to focus support where it’s needed most. It selects for what needs to be fixed, rather than a random spot-check of a few calls out of thousands.

These five broad use cases illustrate how AI has become embedded within tools that contact centers already use, and provides extra capabilities that augment old-but-necessary processes that are difficult to navigate in the modern environment.

It’s important to note that the use cases for AI that make the most sense are those that put the technology’s analytic output into the hands of line-of-business workers. When AI is restricted to data scientists and IT teams, it is almost never effective in contact centers. Those groups have their own use cases for it. But contact centers have very strict operational constraints that ensure that, unless a system boosts productivity along some very narrow key performance indicators, it will not likely be deployed and will not get a chance to display the innovative and serendipitous outcomes it can deliver.

Perhaps the best way to think about using AI in the contact center is to identify where there are reservoirs of data that are being underused — voice recordings, as an example. What insights or problems can be uncovered by applying data science to those ignored piles of data? Answering that question leads to these use cases and opens the door to finding other rich veins of data to explore.

More information on advanced technology used in contact centers can be found in these related Analyst Perspectives:


Keith Dawson

Topics: Customer Experience, Voice of the Customer, Analytics, Contact Center, AI and Machine Learning, agent management, Field Service, customer service and support

Keith Dawson

Written by Keith Dawson

Keith leads the expertise in Customer Experience (CX), covering applications and technology that facilitate engagement to optimize customer-facing processes. His focus areas include: agent management, contact center and voice of the customer and technology in marketing, sales, field service and applications such as digital commerce and subscription management. Keith’s specialization is in natural language and speech tools with intelligent virtual assistants, multichannel routing and journey management, and the wide array of customer analytics. He is focused on how businesses can break down technology and operational silos to provide more efficient processes for two-way engagement with customers. Keith has been an industry analyst for more than a decade and prior was the editorial director of Call Center Magazine. There he pioneered coverage of cloud-based contact centers, speech recognition and processing, and the shift from voice to multichannel communications. He is a graduate of Amherst College.