AI Empowered CX: Real Conversations, Real Results: Shri Nandan, Comcast

The Gist

• Start small. When implementing AI in customer experience, it's not necessary to make sweeping changes right away. Start with small, practical initiatives that align with your business goals.
• Plan and roadmap. Create a clear plan and roadmap for your AI initiatives. Consider factors like how it can make money for your company, its impact on customer experience, and whether your technology can support it.
• Involve relevant teams. Collaborate with various teams within your organization, including IT, marketing, compliance, and AI/ML teams, to ensure the successful implementation of AI solutions.
• Focus on ethics and compliance. AI implementations must prioritize ethics and compliance. Ensure that AI-generated responses align with legal and compliance standards.
• Continuous monitoring. Continuously monitor AI performance and customer interactions to make improvements and maintain quality control.

This episode is brought to you by Wix Studio.

Dom Nicastro: Hello everybody, Dom Nicastro CX Decoded podcast here. We are in season four, episode two. So happy to have one of our good friends, great friends, I should say at CMSWire, an advisory board member, a speaker at our upcoming events and now a speaker for this podcast so happy to have her., Shri Nandan and VP of AI Experiences at Comcast. And just a quick Editor's Note, folks, this podcast recording took place before Shri started her new role at Comcast.

What's going on, Shri? 

Shri Nandan: Hey, Dom, nice to talk to you. I've known you for the past five years, I just enjoy chatting with you always, happy to be here.

A Journey Into Digital Transformation

Dom: Same and I believe we met at one of our past conferences, is that correct? DX Summit, yes, DX Summit. That was our old brand. And now we got the Connect brand that we have in Austin in May, coming up may 2024. You know, before we get into the meat of the show, we're gonna talk about contact centers, AI, digital investment, all that good stuff. But I like to know more about Shri. You know, tell us about your journey and everything in the world of digital transformation. And what inspired you to get into the roles that you've been in?

Shri: Sure. I have a background in technology, I started as a programmer, I used to type out PHP code by hand. That's how old I am. So I moved from the sort of hard coding side to the more product and project management side of things with the more business side of things. And my journey in midlife was all about digital transformation. And I had so many good mentors who taught me how to understand the customer, how to write customer journeys, and how to use data to analyze what the customer is thinking. And it just opened up this entire new world of being able to serve the customer, it wasn't just about getting the job done. And going home, I'm actually seeing that these transformative initiatives are making a difference to the customer's life. And it just became the thing that I have been doing for the last 20 years is, all I want to do is create customer experiences that are meaningful to the customer. So since MetLife, that has been my goal, I have always been able to lead product, and engineering teams and sort of go from end to end, you know, think of the product as a concept and then execute it. So it's been a really interesting journey, especially as we move into the whole AI realm of things. It's starting to get even more interesting.

Related Article: Digital Transformation in Customer Experience and the Butterfly Effect

AI's Impact: Technology Transforming Industries Once Again

Dom: I know transformative technology that's out there right now. I mean, have you seen anything that quickly, just transcended a whole industry like AI? And we're not even there and what it actually will do someday, right? I mean, it's gonna get better and better. Some people say scarier and scarier, I think better and better. But, I mean, have you seen an impact like this in your role, like the internet was certainly impactful, right? I mean, the evolution of that, I'd say,

Shri: Yeah, absolutely. I think the mid 90s, when we saw the internet boom, was similar. Everybody said everybody's gonna go online shopping, no, no more retail stores and things like that. So that was the first hype cycle. And then about 10 years ago, we had the AI hype cycle, start up with predictive AI. And now we're in the, I think of it as the second AI hype cycle where we have generative AI, which is, it's just made AI available to everybody. It's not like AI didn't exist. It's always been around, you know, robots have been trying to take over forever. So it's not new, but it's become more approachable and available. And it's just amazing how much it is affecting our lives as people who produce products and as people who consume products. It's an amazing transformation. It's a really great time to be a product manager.

Related Article: How Is AI Changing Digital Transformation?

AI Chatbots: Transforming Customer Service and Productivity

Dom: Yeah. And as a digital evangelist, you know that you are and doubling down on AI now, like, what would you say some of the early wins have been for you personally in your roles? Since November 2022? Because that's like the big day, right when ChatGPT came out? I'm sure a lot of the AI engineers and researchers have like, yeah, that's one of the days but for me and looking at that as an editor, I mean, I use it really consistently to analyze large amounts of text. What are some of the big early wins for you and your role as a practitioner?

Shri: One of the most recent things we've done is a live person AI chatbot and it's just had a phenomenal level of success in improving our customers lives, and the data that's coming out of it is just so beneficial to us and to the customers to the bottom line. The other side to it is how it is affecting our team's productivity. We're starting to use AI to write code, to write requirements, and you know, to make our lives easier. So it's had an effect on both how we do the work and the kind of work that we produce. So it's been interesting experience for us to see how it actually has a cost benefit analysis to it. So we're actually seeing an uptick in revenue because we are using AI. That was unexpected. I thought I was just experimenting with it, but it actually had a positive effect almost right away.

Related Article: Super-Power Your Teams With Generative AI in Customer Service

AI Chatbots Enhance Staff Productivity and Loyalty

Dom: Yeah, and implementing something like that, you know, into a chatbot. Has that led to more training? Has that led to additional hires, you know, AI scientists? I mean, that kind of thing, like, what impact would it have on staff? Because people are wondering, is it going to replace me, but you have been doing this, so what's the impact on staff with AI implementations for you?

Shri: So that's the pushback you will get from operations, as you know, is this going to take away my job, so one of the things we did was to build the chatbot with the contact center, not for them. So we involved them from day one, and explained to them how this is gonna make their life easier. So this is how chatbots work, if you take away, 95% of our calls are handled by the chatbot, they are contained, it never goes to an agent. So what happens to the other 5%, the other 5% are the high value customers, the ones who are high maintenance, the ones who have bigger problems, the ones who are yelling and screaming. Previously, the staff, one person was not able to spend more than two minutes for the high value customer. Now they're able to stay on the phone with them for 20 minutes and solve their problem. And they have built loyalty. So it's not like the staff has gone away, or their jobs have been taken away. But they have moved on to delivering more meaningful solutions to our customers, you know, to the ones who really need it. So that is the effect that something like an AI chatbot has, it's not about taking away jobs. It's about rethinking how you treat your customers and how you use the skill set that you have. Even in terms of modeling, we have an AI/ML team. And we have them creating models for us all day long to do all kinds of things. But now that AI is here, their focus has shifted a little bit to thinking about how can we use our skills in that direction, as opposed to just bringing it for underwriting or fraud or anything like that. So it's more of a rewiring as opposed to you know, getting rid of people or hiring people or anything like that. It's more of a rewiring of the skill set that we already have.

AI Call Resolution: Achieving 95% Success Rate

Dom: And you know, something you said about the stat that 95%. That can't be industry standard that 95% of these calls get solved right by AI? That's phenomenal.

Shri: Yeah, because that's how simple it is, right? Like, you don't have to overthink it, you just sort of categorize the customer problems and say these are all the you know, problems we need to solve and put the answers in there. And then the AI picks it up. But we didn't get there overnight. It took us about a year to get to 95%. It's not overnight, and even then we are constantly tweaking, we're looking at the data. And we're tweaking the data that we're feeding into the language models. So it's not that it's easy, or that it's overnight, but it's doable. It's entirely possible.

Related Article: AI in Customer Experience: 5 Companies' Tangible Results

AI Implementation Eases Stress for Contact Center Employees

Dom: Yeah. And one of the things we talk about Shri a lot at CMSWire is employee experience in the context center. Right? And how good customer experience strategies and tools and leadership can help employee experience naturally because that isn't unforgiving job. No one in the context center that answers phones all day gets a call that goes like this. Hi. Just wanted you to know you're awesome. And you're doing a great job goodbye. It never happens. It's always a problem. How has the implementation of AI like directly impacted employee experience would you say?

Shri: I think it has reduced stress levels a lot. I was mentioning the fact that they can stay on the call with one person for about 20 minutes. And that makes a huge difference to the employee and to the customer. The more time you can give them and solve their problems. Instead of saying I'm sorry, you'll have to come back in two days or let me put you on hold or Let me transfer you to another person. You are giving their personal attention to one individual for that long. It makes a difference in the life of both the customer and the employee. We're seeing the data that proves that as well is more and more of the calls that are being handled by the agents are getting contained. They're solving problems. They are arriving at a solution that helps everybody. So I think it's definitely been a positive experience for employees. 

How Long to Achieving 95% Efficiency in AI Implementation

Dom: Yeah. You said one year to really get to that 95%. That's a journey. In fact, that seems fast in the business world to me.

Shri: Not these days. One year, I mean, we should have gotten there a whole lot sooner when we started about 60, 65%. And it sort of stayed that way for a while. And then suddenly it started going up as we started analyzing the data. But sometimes you look back, I mean, hindsight is 20/20. Right. So, you know, you look back and you say, I wish I had looked at the data sooner, but that's okay. I mean, I'll take it at this point.

Data Insights Drive AI Success: From 65% to 95%

Dom: Yeah. What are the specific differences and challenges between the 65 percenters? I'll call them and the 95 percenters? What gets you over the hump specifically, what challenges are happening at the 65% level, and what gets you to that increase like 50%?

Shri: You need a good business intelligence team, you need to have people whose brains work in that way who look at numbers and derive insights. And I cannot do that. I cannot look at data and have it speak to me in that way. So I have a really strong team of business intelligence folks who are constantly mining the data from the AI chatbot. And we also get a lot of information from live persons themselves, a lot of insights coming our way. So you need to be able to do that from day one, which is probably what we did not do is you need to be able to look at that data and mine it and, and sort of get into the heads of your customers. And that way from day one, there's no scope for delay. And I think that's something that we should have understood sooner, and it became a bit of a challenge. But now that, you know, we get that we're sort of getting smarter about how we look at the data and how quickly we react to it.

AI in Customer Experience: Customer Satisfaction Grows, but Challenges Remain

Dom: And those resolution times 95% is huge, like we've been talking about. It has it translated in customer satisfaction terms, resolutions, great, and customers problem solved. But they did have a problem, right? So you know, how has this transferred over to things like customer satisfaction, customer effort score,

Shri: We had an NPS goal of 30. And we got there, but we got there slowly, but not as fast as I would like. The problem with NPS is it doesn't tell you why, it just gives you a sort of score. So we also have a CSAT score that goes along with that. That gives us a better idea about why are they satisfied? Why are they unhappy? The customer satisfaction score has been sort of, it's not as steady as I would like, it goes up and down. And it's a little bit volatile. And we're not sure why. And that's what we're trying to dive into. We're trying to categorize some of the questions that we're getting. And in the 5%, that is not contained. What are some of the questions that we're not answering? Is that the cause for the CSAT scores to go down? We don't know. That's the that's point where we're at right now is trying to figure out why, why is everything else looking so great, but not that, you know, it's a bit of a mystery to us.

Unlocking Opportunities: Beyond Basic AI Chatbot Metrics

Dom: Within an AI chatbot? What kind of data do you want out of that? And do you get, the data mining? Obviously, you get resolution rate, that's great. What other kinds of pieces of data would help a contact center from an AI chatbot?

Shri: I want to know how long the bot takes to contain a problem. Like we dropped from six minutes to three minutes on agent related calls. But what about the bot? How long is it taking? I definitely want to know how long it takes for the bot to sort of it's like, isn't meandering through meaningless questions before arriving at the right one, you know that that's what that tells me. I also want to derive data that helps me solve other problems. For example, we realize that almost 50% of our customers are asking about refinance. What does that mean? I think that's where we want to use the chatbot data, it’s not just to say the agent did this or the bot did that or the containment was this, but what other problems will be solved. So we created a whole marketing stream to target our refinance customers and prequalify them and so on. So it created a whole new revenue stream that didn't exist before because of the data that we saw in the chatbot. So it's going beyond what the chatbot is telling us and deriving solutions to problems for the company as sort of like an enterprisewide solution, as opposed to just looking at the sort of basic chatbot metrics.

Incorporating Marketing Insights Into AI Chatbot Branding

Dom: Yeah, you know, would a marketing team in the situation kind of get involved like, hey, we want the chatbot to be branded or those kinds of things, does that ever come into play, or is it just you know what, we're gonna go straightforward with problem solving. Let's not worry about tone, let's not worry about the color. You know, that kind of thing. I could see a marketing team being like we need a say here.

Shri: Oh, we absolutely have branded our chatbot. We call it Penny, we didn't put too much thought into that. But she, the chatbot is a she. And she has branded the marketing team, everybody was looped in from day one, the marketing team, the underwriting fraud, everybody was looted from day one. So the marketing team has definitely had a say in branding it. And I think that's a good thing. I think it's important to talk about these things from a perspective of how does it appear to the customer, if the customer is not seeing a branded brand experience, then that might be a little bit of a problem. So I think that's a good thing. They also frequently listen in on the data, like I was talking about the refinance piece to see what they can help us with in terms of marketing efforts, paid and social and all that stuff. So absolutely. Marketing is a big part of this.

Key Players in the AI Chatbot Implementation Journey

Dom: In a journey like this, who else needs to be involved to make it happen?

Shri: Obviously, IT, I mean, there are folks that need to be involved from day one, like the contact center folks, the IT folks, the marketing folks, being in the industry that we are, there's a lot of regulation and compliance. It's important for us to make sure that we haven't, you know, put something out there that could get us into CFPB trouble or anything like that. So we definitely have them involved as well. There is not a lot of scope for risk and fraud when it comes to any chatbot, it's just answering your question, but we still keep them in the loop. And then of course, the AI ml team, I mean, they are actively looking at all of this data and you know, taking it and putting it into a data lake and analyzing, and so on and so forth. So it's more than just dropping an iframe on a website, it takes a lot of effort to put a chatbot.

Enhancing Customer Engagement With AI Chatbot Copiloting

Dom: Gotta say, I think Penny has a good name, I could never like scream at a Penny. Penny, you're not listening. The bottom line for me is just, you know, solving the problem, of course, and I'm a fan of the chatbots and self service. I'd rather I don't even care how long it takes I if you know if I can send them a message and then go do some work. Go back like oh, they answered. It's been 20 minutes. I forgot about it. But I'll go back. Sometimes you can't account for that too, with customers, right? Like you never know, if they're just doing work. They're cranking out emails, and they're not fully invested in the conversation that might reflect on the chatbot like, oh, this took 25 minutes, it should have been five. But for me, it has taken 25 minutes because I forgot I was talking to one.

Shri: Yeah, that absolutely happens. And we're trying to look at the data for that and see if we can solve that using, you know, sort of copilot. I was at a conference and I saw this, I forget the name of the company, but they had this chatbot that would just mirror everything that the user is doing. And I thought that was fantastic. Because you're keeping the customer engaged, you're seeing what they're doing, without actually getting into their computer, you're mirroring them, and then you're walking them through it. You're saying I'm holding your hand and help me you know, click on the screen, see what happens. It's as if someone is sitting next to you and helping you. And I think that's fantastic, especially with somebody who has a really big problem. I think that sort of copiloting is hugely beneficial. And we're trying to experiment with that. We're trying to pilot that a little bit. Here. There's a conversational copilot, and then there's an autopilot, where it's just the bot does it it doesn't hand off to an agent, the bot is the pilot. So it's just also fascinating to talk about. We have to see how how it helps the customer at the end of the day.

Voice Chatbots: The Future or a Work in Progress?

Dom: How about voice like is there a future for or even a present day for voice chatting with a chatbot? Like, hey, I need to pay my bill. I can't figure it out in the chat. Because about like, Yeah, let's figure this out. Dom. You know, I know there's IVR and all that like on phones and all that. But I wonder if AI can sneak that in and like a chatbot setting like you click on the chat bot on the website. And here comes a voice.

Shri: I think that's happening now in some places. And and I think that's a good idea. The only problem with that is I don't know that anyone has managed to make the bots sound human. You can tell it's a bot and it's not real. I don't know that we have come to a point where it sounds so natural and so friendly that you can tell the difference. And that's the problem. I hate talking to bots. I feel like you know, what the hell do we type into a machine like where's the real person? That sort of thing. But maybe that technology is out there? And I haven't seen it. But I don't think there's a board out there that sounds human and that might be what stops it from taking off. But I don't think we're that far from it.

2024 Vision: AI in Contact Centers

Dom: Looking ahead, let's say 2024. We're here in 2024 a year from today. What are some of the things you hope as a practitioner in this space of implementing AI into contact centers into customer experience scenarios? What will you have hoped to accomplish? What's going to be the next level AI infusion into customer experience if you will.

Shri: The thing that we we're just talking about the voice thing, I think that would be super nice if we can just get the calls handled with doing a warm handoff, and the customer has no idea that happened. That would be fantastic. I don't think that customer contact centers are going away. I don't think that that's going to happen. And I don't think people should aim for that. I think the human touch is always important. Having an agent at the end of the line is always important, just in case. So I don't think that's going away. But how can we use AI to improve their life not just in a way that like, what is the chatbot doing? But are there AI tools that they can use? Once they have to call? Or do they have tools at their disposal to be able to solve the problem much more quickly? And I think that's where we should be heading? Not, let's get rid of contact centers, but more how do we use AI to empower the contact centers?

AI Ethics and Transparency in Implementations

Dom: Yeah, that makes sense. And Shri, one thing we haven't jumped into very deeply is AI ethics, transparency, I mean, on the day that we're recording the Competition and Markets Authority in Britain set out a whole list of AI principles. And you know, haven't read the whole doc yet. But most of it is centered on transparency, letting consumers know that you're using this kind of technology, does that come into the equation when you're building out these kinds of AI implementations?

Shri: Absolutely. I think we're just talking about chatbots here, but there's so much more to it. Right? We are giving responses to the customer that come under a certain jurisdiction of compliance in legality. So we're saying, Yeah, absolutely. This is our check cashing rate, we have to make sure that we're not making stuff up on the chatbot, we have to be sure that we're giving them the right information that we are giving them information that is helpful. We are using AI to write code. There's a certain level of quality control that comes in because you're using AI because AI is making up stuff. Is it writing code that might be breaking some rule? Is it writing code that is going beyond a certain quality issues or legal issues, compliance issues. So every time you use something that's generated by a machine, you have to think about ethics and quality and the effect it has on the customer, I don't think you can just write AI code and then just leave it alone there too do whatever it wants. I think there's a lot of monitoring and handling that needs to happen.

Challenges and Benefits of AI Code

Dom: Write AI code, I just love how you say that so seamlessly. Like, I wish I could do that. I mean, you write in PHP by hand, now you're doing AI code. Any differences there at the end of the day is coding, coding, and it's just challenging all along.

Shri: Yeah, I mean, the GitHub, it's called GitHub copilot. And he said, we're going to use it. And it just generated so much nonsense. I'm watching my language on the podcast, but it just generated so much nonsense. So we realized we need to beef up our sort of quality control measures. But at the end of the day, it works. You know, it really freed up a lot of time for us.

Start Small and Build a Roadmap: Leveraging AI for Customer Experience

Dom: Good. Well, let's wrap up things with one big takeaway, or just a few big takeaways on if I'm a customer experience leader, listening to this, a chief customer officer, and they just don't feel like they've leveraged AI technology to the fullest yet in their contact center, and or they need some employee experience boosts. They need some NPS boosts. And they say, You know what, I think AI is the way to go. I'm liking what Shri saying, but where do I get started? What's my first move in this big picture here?

Shri: Get started anywhere. I mean, you have 1000 things at your disposal, pick one, maybe just pick something simple that helps your team like writing code or writing requirements, you don't have to install a major lead chatbot or anything like that. Get started by talking to people who have the service like there are so many off the shelf products, start talking to them, understand what's out there. We don't have to make sweeping changes in digital transformation doesn't have to be like, Oh, we install all this great software. It doesn't have to be like that. We can start small, you know, do small and simple things and then have a roadmap. There's all this AI at your disposal? How does it make sense for your organization? There are only three reasons to do something: it makes money for your company, it's meaningful to the customer and your technology can support it. So look at these three factors and see what fits in and build a roadmap so that you have a plan instead of sort of sitting in this. I think of it as an innovation workshop where you're saying I'm going to do all these shiny new things, but I don't know what it means but have a business outcome. Have a plan and start small. I think it's it doesn't have to be complicated. 

The Wrap-Up: Shri Nandan

Dom: Yeah, I'm excited to see that too. I gave a plug, I'm gonna give you a chance to give a plug. Where can people connect with you, talk with you? Are you in any groups or forums or anything like that you wanted to promote? Do you have a blog? Shri Nandan here, of course, we can find you on LinkedIn. But is there any anything else you wanted to tell our listeners about where they can connect with you?

Shri: LinkedIn is where I am active. I do have some Medium articles. But I'm not super busy there. But LinkedIn, I'm always on LinkedIn. I'm happy to talk to people and connect with everyone, and I just, I have a blast on LinkedIn. Everybody's so nice. You ask someone for help, and they show up even if you've never met them. And it's just it's, it's where I am, if you want to find me. 

Dom: Awesome. Well, thank you so much for joining CX Decoded, much appreciated, awesome conversation getting into the how to do things. Thank you so much for joining us here on CX Decoded. 

Shri: Thank you so much. I had a great time. All right. Have a good day.

Thank you for listening today. We encourage you to drop us a line at [email protected]. And if you have comments or suggestions for a future podcast, we'd love to hear from you. Additionally, if you like what you hear, please post a review on Apple podcasts or wherever you may be listening to this. And be sure to share CX Decoded with anyone that you think might benefit from these types of conversations. Finally, be sure to follow us on X, formerly Twitter @CMSWire. Thank you again for joining us, and we'll see you next time.