Innovations in contact center efficiency

There’s a lot of change within the healthcare contact center market. Tell us about the trends you’re watching.

Marsha: Most of the platforms are at the end of their life and will not be supported. So, it’s important to move to more integrated systems in the cloud; certainly, systems that support application programming interface (API) connectivity and integrate with workforce management and quality programs.

Then there’s artificial intelligence (AI). What can it do? There are some great things AI can do, not just in terms of intent gathering, call routing and sentiment analysis but also language. This is meaningful for states because English is not the first language for a lot of the Medicaid population. Trey can speak to updates on some of the different language opportunities.

Trey: Currently, AI is a race between vendors, and there’s a hype cycle around it. It’s hard for our clients to understand what’s possible today, what’s in the future, and whether it will save them money.

At Optum, we stay current with what’s emerging, and as part of that, we attend industry conferences and talk to new vendors. We study the architecture and understand how the technology behind the new technology works.

For example, for years we have relied on interpreter services with our clients. Now, there are generative AI models with large language modeling that allow us to have conversations with people in different languages without an interpreter.

We’re not using this technology yet with a client, but we will within the next year. As a company, Optum has used AI technology for years and we help states figure out how to navigate the AI hype cycle and get to a product that fits their needs.

How do these trends impact data privacy and security?

Trey: Many of these emerging technologies are at various levels of IT security compliance. There’s FedRAMP and MARS-E compliance and a new set of CMS standards coming soon. One thing we are closely following is making sure it is as easy as possible to keep the security strong for our clients.

As you add more into your overall stack, like with large language models, you want to be sure it is as secure as the rest of your environment, Optum shares these same concerns. For our own businesses, we understand what needs to be done, and we know how to protect ourselves as well. 

What are key benefits to consolidating contact centers for state Medicaid and HHS programs?

Trey: In a previous job, I ran one state’s Medicaid contact center, and we transferred almost 40% of our calls. Some states have multiple vendors or multiple teams within their own state call centers. They each handle specific things or are contracted to do certain types of work.

With call center consolidation, you can decrease the number of vendors or teams, thereby reducing the number of transfers and improving the customer experience. You end up doing more with less. It’s a rare opportunity to decrease costs and improve the customer experience at the same time.

Certainly, there are challenges to consolidation. How can vendors help ensure a smooth transition?

Marsha: There are strategies, and in fact, one of our clients is taking the approach of a soft launch rather than a big bang, all-at-once migration. We’re structured in how we approach our project management, requirements gathering, recommendations, the rationale, system integration testing and user acceptance testing.

We offer top-tier training and software solutions, including AI-driven simulated calls that mimic real-life scenarios. These dynamic simulations adapt based on the agent's responses, ensuring they are fully prepared before going live. Trey, what are your thoughts?

Trey: Some of our competitors sell solutions that only allow their employees to be on your platform. That’s hard because states may have multiple contracts with different vendors or they may have state employees in their own contact center on a platform.

A differentiator for Optum is that we negotiated with our software providers so that anyone can be on our platform. Maybe a state has different contracts that expire at different times. Or the state is ready to reduce the number of vendors and consolidate the contact center.

We can accommodate both of those models.

In each case, we reduce the amount of transfers and improve the customer experience because of the front-end phone system. When the call comes in, we gather the caller’s intent and route them to the correct place rather than having the agent answer and transfer.

What are the best practices for a state agency that wants to consolidate a contact center?

Trey: I recommend having a conversation with us and we can go more in-depth on things they can think about. I also suggest attending contact center conferences and seeing what is going on in that space and what is new.

It is no longer a commodity type of business. For 20–25 years, things hadn’t changed much in the contact center world. The changes going on now are the biggest I have seen in the 28 years I have been in the industry.

Marsha: Yes, I agree. States should select a vendor and a platform and have language in their contract that helps keep them up to date with advancements. Vendors often implement contact center as a “set it and forget it.”

At Optum, our approach is different. Both Trey and I have worked in contact centers in our previous jobs, and we experienced the technology limitations. But now, technology is evolving at a pace where you need to keep up or you will be significantly behind.

States should seek out a vendor who is committed to staying current and then have that built into their contract so they remain up to date with different features that the market is starting to expect and also reap the benefits of the features the technology can offer.

Trey: Another best practice is to consider getting rid of your IVR and replacing it with an IVA — an intelligent virtual assistant.

I worked as a consultant for a consumer insights company. Time and again, the biggest detractor to customer satisfaction scores was the IVR, having to go down the phone tree, pushing buttons. People get lost in them.

Nowadays, they can be replaced with a greeting as simple as, “Thank you for calling. How can we help you?” Based on the response, they’re routed to the right place.

There are fundamentals as well, like workforce management and effective staff scheduling. Don’t be afraid to use part-time staff. At Optum, we have part-time staff who log on and take calls according to the arrival pattern of the calls.

Usually, calls come in a bell-shaped model. You’re busy in the morning, busier in late morning, and then in the afternoon it slowly starts to taper off. We help our clients manage costs by staffing to the volumes rather than putting full-time employees into every position, to avoid underservicing.

Let’s turn the focus to AI. Talk to us about its role in the consolidation process.

Marsha: When you think about a state that has several different vendors and telephony platforms, many of them must layer in AI within each platform. So, if you move to a consolidated model, you can leverage a single AI layer across multiple solutions.

Back to Trey’s point, if you have a vendor who currently has routing and that sort of thing, you can have an IVA installed and route to existing places or Optum can do the whole approach as well.

Trey: I often hear people say, “We are going to save so much money by switching to AI.” Yes, you will save money, but it’s not as much as people tend to think.

AI computing is expensive. It can do a portion of the tasks and even make the call shorter. But you will spend, depending on the budget, 60–80% on software, the implementation and the ongoing support to make it work correctly. By diverting all these calls, you will not get 100% of that savings.

What about bias within AI?

Trey: Yes, bias within AI is a topic people are talking about. AI learns from the internet, so there could be inherently biased reactions exhibited by the software.

As part of our process at Optum, we require that our vendors have undergone bias testing to be sure the software is not reacting in a way that is biased. So, we take some of that legwork out of the process for our clients.

How do you balance AI with human touch in contact centers?

Marsha: We’ve all experienced getting stuck in an AI-type loop where we can’t get to a person and our question is not being answered. It’s important to have a deep understanding of what calls can be effectively handled with AI and what calls truly do need an agent.

People often try to do everything they can with AI and that’s where you run into misdirected calls. It’s not a great user experience and creates more problems than it solves.

Adding empathy or more human types of interactions with AI can have a positive result. When the user interface is done well, people can enjoy the interaction with AI.

What role do human agents play in a contact center that utilizes AI?

Trey: How involved you want the agent to be is entirely up to you. You can automate a lot, but it takes some time to do that. If you want a high-touch contact center, there are quick wins with AI that make the call shorter and save money.

Simply having AI do the up-front authentication of the caller so they can be passed along to the agent means the agent doesn’t have to do that repetitive task. The call is routed correctly the first time, the load is lighter for the agent and the experience is better for the caller.

You also can use AI as a triage tool where it listens on the call and gives agents information on the next best action or point them to an article that helps answer a question.

Or maybe you go a more automated route for after-hours help, taking appointments when agents have gone home to provide service for people who can’t call in during regular business hours. Each client has a unique automation threshold, and the right tools accommodate that.

What future trends do you see in the use of AI and other technologies in contact centers?

Trey: I expect that 10 years from now, AI will have automated most contact center interactions. Not all of them. But I think it will have learned enough to take the next steps. There will be a need for more technical and policy writers, roles that create the materials and provide the data AI needs to improve conversations.

There will be advantages to that in the long term. AI will be less likely to miss the details that need to be asked on the call. For example, maybe we want to ask the caller about their experience with a provider and the system triggers AI to go down a line of questioning that an agent may miss.

Marsha: I think biometric authentication will be a normal state. We will call in, and it will recognize your number, and you will say the first few things you need, and it will recognize your voice. And then, adding to what Trey said, I think we will see AI and data analytics do more.

For example, we’re learning about social determinants of health and how AI, outbound communications and that sort of thing can help people lead healthier lives.

Trey: Yes, or asking the right questions on that call so that AI has a better picture of what treatment or what things we need to recommend to that patient.

How can states prepare to meet these future needs and stay ahead of the curve?

Marsha: We touched on it earlier, but partnering with a vendor who will do a lot of the legwork is an important strategy.

What support and training programs will help staff adapt to new technologies?

Trey: The software does a lot magically by itself, but it doesn’t do everything by itself, that is without tuning of the software, particularly in the AI space.

So, I think data analysts are more important than ever. I mentioned that there will be a big future space for technical and policy writers, and continuous updates so that your AI consumes the correct policies and information that it conveys to either call center agents or patients and providers.

Documentation will be key. With AI, the output is only as good as the input.

How do you ensure that solutions are user-friendly and accessible to all?

Marsha: It’s possible that AI will learn our behaviors. Biometric authentication is just one layer. But if it understands that it is speaking with someone who prefers to speak with a human, why not send the call to a human right away? Or, if someone who doesn’t do well with technology and can’t navigate something, let’s get them help in the way that works best for them. So not only is it learning to identify you, but what are your patterns, your preferences, so that the software supports you as an individual.

Trey: Yes, and also being sensitive to what the person is saying. For example, if the caller says, “I have cancer and I need…,” we would want to be more high-touch with that person than a caller who calls in about a podiatry appointment.

We use tools that have interfaces that allow us to set things up quickly and rapidly. It’s one of the things we look for in our solution stack and what we use on behalf of our clients — user-friendly tools that are easy to set up and adjust rapidly. We use tools that allow us to adjust faster so we can get things into production faster thereby reducing the cost to our clients.