The Meandering Path from IVRs to IVAs

Cost is another roadblock. Conversational systems are quite expensive. While a company can get a small, simple chatbot up and running for free or a few thousand dollars, more sophisticated conversational AI systems usually start at hundreds of thousands of dollars and can quickly jump to the millions of dollars.

And then finding an executive willing to champion such substantial investments can be difficult. “Many companies do not have anyone in charge of the customer experience,” notes Nuance’s Arakelian. Such individuals would need a wide enough view of the company’s business to be able to identify the benefits from improved customer service.

But even with someone in the role of customer experience officer or some similar title, most companies are not organized in a way to deliver comprehensive, cohesive customer service. Business units function in silos. To improve efficiency, corporations break departments up into distinct functions, which often have little to no understanding or collaborative interaction with other groups. These delimiters create friction, if not utterly painful transition points, as tasks get handed from one group to another.

Customers, meanwhile, see one company, not a group of separate departments. They expect their vendors to understand who they are and why they are calling. Convincing managers to visualize the company like customers do can be challenging. “One of best ways to convince managers that they have a customer service problem is to have them use the organization’s service,” Arakelian advises. “Pull up their web page and let them walk through the process of ordering a new phone. Then the silo problems hit home, and they feel their customers’ pain.”

Some companies have made great progress in breaking down traditional barriers and using new technology to improve voice interactions. Healthcare companies are beginning to deploy medical virtual assistants (MVAs), which are capable of collecting information about patients’ insurance, demographics, prescriptions, and medical histories.

Other businesses have integrated IVAs into their omnichannel initiatives. “A hotel chain used an IVA to monitor exchanges, identify customers who were getting angry, and move them to agent chat for immediate service,” [24]7.ai’s Rubin recalls.

There’s no getting around interactive voice solutions’ checkered past. The technology enabled companies to reduce personnel expenses, but sometimes at the expense of quality customer care. With advances in artificial intelligence and machine learning, new solutions are emerging with the potential to balance those two business drivers more effectively. If they are successful, the technology might be able to one day shed its notoriety. 

Paul Korzeniowski is a freelance writer who specializes in technology issues. He has been covering CRM issues for more than two decades, is based in Sudbury, Mass., and can be reached at paulkorzen@aol.com or on Twitter #PaulKorzeniowski

The Differences Between Automation, Orchestration, Artificial Intelligence, Machine Learning, and Deep Learning

Through the years, software has taken on more of the work that humans once performed. At one time, individuals had to punch cards that computers would then read to carry out a specific task. Today, machines don’t always need to be told what to do; some systems can make decisions on their own and carry out certain tasks based on what they’ve learned from previous interactions. But not all of the technology is the same:

• Automationis the simplest form of computer intelligence, in which a computer system completes one specific task. For instance, a contact center program might automatically send an alert to a manager if an agent’s call exceeds a predefined time period.
• Orchestrationis the next step in the process. Here, the system completes a few tasks autonomously. An application might, for instance, pull a customer’s purchasing history from a database and present it to an agent as a call begins.
• Artificial intelligenceis a term that emerged in the 1950s to describe actions that computers take that possess the same characteristics of reasoning as human intelligence. Natural language processing (NLP) is one application.
• Machine learningis a field of study that enables computers to learn without being explicitly programmed. This type of programming is often referred to as “unsupervised,” as the technology finds patterns, builds insights, and automatically acts on those insights. NLP solutions that recognize local dialects are an example of such capabilities.

While the various terms are helpful for understanding how these solutions work, they lack precision, so overlap is common. Also, new terms are sure to emerge as technology further takes shape.