Pragmatic AI, Real Business Value: A Q&A With Chad Dunavant

Communication service providers (CSPs) are under pressure to uncover and pursue new applications for AI—whether that’s to improve customer experiences, streamline management of their networks or apply myriad other potential use cases.

But it’s a complicated road ahead for CSPs trying to draw value from AI implementation. Gartner predicts that by the end of 2025, 30% of generative AI projects will be abandoned after proof of concept. The main reasons: poor data quality, inadequate risk controls, escalating costs and unclear business value.

Which AI initiatives will bear fruit for CSPs? The ones that take a pragmatic approach, according to Chad Dunavant, CSG’s EVP and chief strategy and product officer. He sat down with Doug Green of Technology Reseller News to discuss proven AI use cases that cut through the hype and create real value for telecom operators.

Catch the highlights of their conversation below or jump to Technology Reseller News to see the full video.

The transcript has been condensed and lightly edited for clarity.

Finding AI Use Cases With Real ROI

Doug

First of all, can you even measure ROI [for implementing AI]? Can a CSP do that?

Chad

I think so. And I think it starts with defining the specific use case you want to solve for. Let me give you an example. We launched a product at the end of last year called Bill Explainer. And the intent around Bill Explainer was to use AI to [improve] what has typically been a pretty static process. If you think of electronic bill presentment and payment EBPP [electronic bill presentment], you direct a customer to a website, they pull up a bill, they view the bill, they may have questions about it. If they have questions about it, they may engage with a chat session or pick up the phone and call a call center agent.

We wanted to be a lot more proactive about how you engage with customers around that bill confusion that happens during a monthly statement process. And if we proactively outreach to Doug and say, ’Hey, Doug, what we’ve noticed on your bill is a change—a $20 change and fees this month—and it’s a result of you watching a sporting event.’ Or maybe you bought a movie, or maybe a rate change happened because you moved out of the promotion. Having that conversation upfront can actually drastically reduce the conversations that happen inbound with a customer.

You can measure ROI with use cases that are very specifically tied to an outcome. In this case, the outcome is we want to create a better experience with the customer because we are proactively communicating something that would have likely led to a call in the call center. And if we could then see the reduction in call center calls, or in the time it takes a customer to pay—speeding up that process—we can put a very tangible ROI in front of an operator based on those results.

Doug

And this takes us to the opportunities for the CSPs. Where do you think they are?

Chad

A lot of times we start with cost savings. What is the ROI based on the cost savings that you have with a specific AI use case? I think more importantly, you should start to think about ways that you create value. And I think the differentiation for operators is, how do they create a much more personalized, rich experience with my customers by leveraging this data? Forever, they’ve developed products and solutions that have kind of been one-size-fits-all. They develop offers or promotions based on large groupings or subgroupings of customers, on profiles of customers.

I think they can get much more personalized in the way they engage with customers now that they leverage this technology. If I go back to that Bill Explainer example, that outreach could be very personalized. How I want you to pay or interact can be very personalized, and then that can lead to really dynamic bundling. If I know more about you; if I just ask you the question, ‘What do you like to watch? How much do you want to spend? When do you watch specific shows? How do you engage with your friends and family across this medium?’ I can then start to tailor a specific offer, a specific promotion that’s based on your specific attributes versus maybe a collection of people within my database or within my processes that don’t match your specific wants and needs.

I think that’s where we can start to focus AI technology, is creating that unique customer experience with each individual customer, and that will then lead to increased customer satisfaction, increased retention and increased NPV [net present value] scores for the operators going forward.

The ‘Right Ways’ for Telcos to Implement AI

Doug

Do you feel that the CSPs have challenges in integrating AI into their operations?

Chad

I do, and I think it starts with data. Data exists across lots of different systems and subsystems, and normalizing that data has likely been on their product roadmap or their radar for the last decade. But what I’ve seen in conversations is they’re not at a place where they can actually take some of those large data lakes or those large data initiatives and really turn them into an ability to garner returns. So, there are subsets of your data that are ready now to take advantage of in training these models. They’re normalized.

As a BSS supplier in this space, we sit on a lot of billing data—customer care data that already is normalized within our systems. So rather than trying to start a large data normalization project that might take you five years before you then engage in some of these AI use cases, I would implore the operator to think about how you can take those datasets that are already normalized and get the benefits of AI now, while you’re working towards a long-term goal that normalizes maybe all of your data down the road.

Doug

I think there’s a danger, as we move into the world of AI, to lose some of the human touch. Do you see that in the CSP world?

Chad

I do. I think what we would advocate for is giving some of the same tools you’re giving to your end consumers to your call center agents, and to your staff that’s supporting customers or technicians. And if you do that, you can actually create efficiencies both ways.

I’ll go back to that same use case we talked about around Bill Explainer. If I’m giving the end consumer insights about the changes in their bill, I should be giving my call center the same tooling. If a call does come in and a call center agent has the ability, then, to see those interactions and understand why things have changed, they’ll be able to communicate that in a much richer way with the customer. It creates a better conversation with the customer when they do call into the call center, number one. And number two, it reduces the training time.

And we’re never going to cut out entirely the need for human interaction. But if I can use those tools to make that human interaction richer and more fulfilling, then I think again, it increases that customer satisfaction score. And it actually enriches the employee satisfaction score because they feel more empowered to have those conversations because they have the right information in front of them.

Doug

Hasn’t that always been the case with any digital transformation? That there’s these sort-of wrong migrations, and companies sort of do it in a kind of oafish way? And you just sort of know someone out there is going to do it that way?

Chad

Yes, but I think we’ve all been through this long enough that we have to be pragmatic about how we make these changes. And I don’t think it’s a one-size-fits-all, especially for an industry like telecom that’s been around for 30, 40, 50 years. There are customer expectations that we can’t just turn off overnight. And so, I do think having a pragmatic approach, managing that evolution over time, is the right way to do it. We’re hearing this all the time: ‘I want to talk to a human; I want to talk to somebody that can help me.’ And so, right, let’s embrace that, but let’s make it more efficient for everybody in that process.

Getting Pragmatic With a Blended Approach to AI

Doug

So part of your vision is to use the AI effectively to get you to the right human and to prep that human?

Chad

That’s exactly right. And, in fact, we’ve used it. One of the use cases we’ve used within our customer experience division is a large organization—I cannot use the name—but they’ve got lots of different departments. And so even when you call in to, say, the helpdesk, you may be prompted with 15, 16 different prompts about where you want to go. And you may end up in the wrong department, which then you have to get routed back to another department and sit back in a queue. We’ve been able to use AI and intent-based natural language processing to identify what your challenges are, what your problems are, and get you right to the right human the first time. And so that efficiency gain is not only for the operator to make sure that they’re talking with the right customer at the right touchpoint, but from a customer experience perspective, it’s a huge gain because I’m not sitting on hold and I’m not getting frustrated as I get transferred around an organization.

Doug

As we look into the future, I have a feeling a lot of people are just going to look at this as maybe a magic cure-all, a magic fix.

Chad

AI is not new and it’s going to continue to evolve. And I think we have to take that same approach when you look at this technology and the use cases, and find the use cases that are most suited for this type of technology. I don’t think you can just apply it and assume you’re going to cut out your call center representatives or your call handle time. I think there has to be a blended approach. And it’s definitely how we’re thinking about how this technology can be leveraged and how we can deploy it to create real benefit within the community.