Why Non-Linear CX Needs Emotionally Aware AI

Customer journeys are no longer linear, and AI alone won’t fix the friction customers feel when they need fast, empathetic support.

In this CX Today interview, Nicole Willing speaks with Mike Clifton and Max Schwendner, Co-CEOs of Alorica, about how real-time data, emotionally aware AI, and human support are reshaping customer experience.

They explore why businesses need to move beyond generic personalization, how AI can help agents understand customer sentiment in real time, and why the best CX strategies focus on outcomes rather than technology for its own sake.

The conversation also covers multimodal customer journeys, data orchestration, global CX delivery, the role of human agents in complex interactions, and the importance of knowing when AI should step back and let a person take over.

For more Customer Experience tech news visit https://www.cxtoday.com

Transcript

SPEAKER_00 0:07

Hello and welcome to CX Today. I'm Nicole Willing. There's no doubt that customer experience is being reshaped by real-time data, AI, and deeper understanding of human emotion. But research indicates that while customers place a high value on empathy in their interactions, it often isn't what CX leaders prioritize when designing experiences at scale. As expectations continue to evolve, businesses are rethinking how they engage their customers across every touch point. Joining me today to explore that are Max Schwendner and Mike Clifton, co-CEOs of Valorica, who are going to help us unpack what this transformation looks like in practice. So welcome Mike and Max, thanks for joining us.

SPEAKER_02 0:46

Welcome, thank you very much. Thanks for having us.

SPEAKER_00 0:49

So let's start with the big picture. How do you define a dynamic customer journey today? And what's changed really with the rise of real-time AI-driven experiences?

SPEAKER_02 0:59

When you say dynamic customer journey, I think uh Max and I would have categorized it as this. Most clients that we work with have sort of started journey mapping by looking at the path by which clients enter a specific use case or a function. So a return, a bill. And those journeys orchestrated that way have been personalized for them in some way. So I think most clients feel good if they have a transaction, but where they have a problem is where those journeys aren't connected across multiple functions, right? So I think when you say the word dynamic in that word journey, we would look at it as don't think of it as a journey as a linear path through something. It's now with, I think, as you said, AI, it's now a kind of an ecosystem of a map where it's really jumping around between functional things in order to fulfill the requirement of the transaction or the need of that person. And that might be multiple things: a build to a new phone, to a new heating system, to a new X, right? It doesn't really matter what it is, but those things are now interconnected through kind of that orchestrated diagram of a more of a circular planet model versus a linear end-to-end.

SPEAKER_01 2:23

So yeah, no, I think that that's well said. I think the only thing I'd add to that is I think what Mike and I have spent a lot of time with both our teams and our partners on the technology side is trying to make sure that all of our interactions can be handled with our clients on a multi multimodal basis, which is, I think, Mike, what you're going for, right? Which is can someone text or or chat with an AI bot? Then if there's an issue, can they escalate to a person? Can they go back to a chat? And having the ability to kind of meet customers where they are, I think is probably the most important challenge for business, our clients today, and what we provide for those, for those clients, is you know, people expect very different things from technology, different things from their uh experience. And so I think the one thing that I would say is that there is no one journey. I think that's what Mike was going for with the journey mapping, is that every journey is somewhat a little bit different and has a degree of personalization to it. And we're certainly seeing that across almost all verticals that we we operate in today, where clients are increasingly looking for the way to engage people where they want to be and have the level of personalization that they want. So it's not just generic personalization, it's Mike may want a different degree of personalization than Max may want, and having the ability to kind of flex and toggle, I think it's gonna be the future of kind of what makes it a good experience versus you know hitting pounding zero as many times as you can to get to an agent. You know, I think no one really enjoys that experience as much. So that's that's how I would say.

SPEAKER_02 3:50

I'd add one thing to it. Um we've seen predictive versus reactive, right? So predictive analytics used to be back office functions. Now it's predictive actions, which is AI-enabled outcomes, right? I think where we're going as a sort of orchestrated map is really about how you think about the client and all the things you may know about them and what they're going to ask next, and being able to predict with the AI assist and a human touch point how that actually would uh would help them get to where they want to go, might solve the problem or solve the question.

SPEAKER_00 4:25

Sure, exactly. I mean, that word personalization you know gets thrown around a lot, and um, you know, some customers are not experiencing that to the extent they want to, they might still be encountering you know delayed interactions or generic responses. Um, but you've kind of touched on where that is going. So, um, from your perspective, what does that truly look like in practice where you have this real-time personalized experience that works?

SPEAKER_01 4:49

Well, I think the the best versions of not everything is the same, right? And I would say customers are at all different stages of their technology journey and what they want their design of their CX experience to be. So there isn't, unfortunately, there's not one truth. And so whether you experience it if you're uh talking to a bank, to a utility provider, to a health insurance company, to a uh a telecommunications provider, all those are very different. I think what the best versions of real-time CX look like, you know, has a degree of sentiment analysis, historical buying patterns. So let's just take a real example. Let's say we know that Mike goes in to buy a new cell phone every two years, uh, and he wants these kinds of features. And we know that Mike lives at this part of the country, and if he wants to go try that device in person, hey, here's the here's a location, you can go see it in person, and then if you have any issues, you can call me back. That's a best case version, right? Which is we know about Mike, we know Mike likes to be engaged that way, and we can use location data and all the other data points to Mike's earlier point on how to connect that map of information. The worst case scenario is kind of the other end of the spectrum where Max has already put his information, it's always Mike gets the good stuff and Max gets the bad stuff. So Max has to put in his information three different times to three different systems, and then he gets transferred six times, and the person before doesn't talk to the person after. Everyone has those experiences in their own way, shapes, and form. But I think what's increasingly becoming the norm, and what's really cool about a lot of the technology we're investing in, is that that allows us to help our clients bridge that divide with next best action or sentiment analysis that our people can help kind of provide a human in the loop for those AI experiences where that isn't as tightly woven together. I think that is what I would say is that I think people want the mic scenario, and they unfortunately deal with the max scenario way too often. But there's a journey to kind of getting there. And I think what I would suggest is you know be patiently impatient, right? Like the the fat we will move along that journey and our we will move with our clients at pace, but it's one of those things where I think clients themselves have different perspectives on how they want to engage with their customers. So I I know I'm kind of giving you a no no one size fits all, but what what's cool about what we see every day is the whole spectrum of you know, from a fully personalized to the you know, not just the multi-bajillion dollar client that Mike is to the lowly max that exists, it was made one by with you. There are clients that do that. There are also clients that don't have that yet, but they're on that journey map. That's how I would answer that question.

SPEAKER_02 7:25

I I'll only add one. I mean that the clients that are doing this well, uh, and we we're AI enabling our company in a very vast amount of different functions, um, and I'd say we're doing it well, but they're not throwing more AI at it, right? I think I think the one thing we've seen uh as a kind of patent is originally when all of the AI launched and ChatGBT hit the markets and all the revolutionized network 15 models every two weeks, it looks like everybody was adopting tech for tech. What can we do with it? What's the use case for it? Uh and use cases were great because it gave the AI companies and technologists an understanding of the business to be able to figure out how to solve problems. I think we've moved, and the focus was eliminating people. And that focus is sort of pivoted now. It's about making people more productive, and it's about giving them the expeditious side of delivery functions where both an agent call, in the case of CX, can be a sort of wrapper with the right AI functionality for the right journey, as you mentioned, in personalized outcome, or it's a coder doing that much more in terms of being able to build an enterprise system that much faster. It's more of the same assistive technology to human capacity. But I think what we've seen is those folks deploying more AI, the limiting factor becomes where's the data? Can I get access to it? And how can I get access to it over a long period of time so the models can learn, right? And I think that last piece is the piece where AI will be the most effective, which means it's got a maturity to it over time, that we're probably just seeing the beginning of as companies adopt in a large-scale way because they're in they're implementing a ton of data ingestion right now, but it has no context yet, and the legacy systems weren't orchestrated correctly. So I think orchestration becomes more of the important outcome than the AI model itself as you go forward.

SPEAKER_00 9:25

So sure. Like you say, it's more than just the data, it's the the other things behind it as well. And um to bring it to you know, this growing interest in emotionally aware AI, you know, like as you mentioned, or just AI, throwing AI things for the sake of it. Um, but how does that actually work in scale? And then how does it also help to reduce customer friction? Because that's always the goal, right?

SPEAKER_02 9:49

Yeah, we we've spent a lot of time on this. Um when you think about personalization, you think about linguistics and language, and the ability for us to be at a global scale as a company, um, and where most connections with clients fall apart is the ability for you to understand, as Max said, sentiment, but also empathy, uh escalation, and being able to really have a set of models and an orchestration layer that can interpret and understand and respond so that the person on the call with you, or even if you're running a virtualized agent, be able to understand it and react accordingly and pivot uh so that those intents are delivered with an equal response, right? And I believe more of this is maturing faster, where language and you know, all the dialect and language capabilities have progressed exponentially uh over the last year. That's only helped with an understanding of the models interpreting all of the ingestion data so that you can be responsive, but you can understand the dialect tone, the volumetric issues that clients would have as they raise their voices, or just that they're frustrated by the number of questions they're asking and not getting an answer. I believe the models are getting faster, the sentiment is getting faster, the personalization is an understanding now that when you called the last time, you didn't get an answer, you're calling again, and now I am pull, I pull that forward and understand that context from the last discussion. Now that model's aware of it and can assist the human in pushing the right kind of content. It says, by the way, last call didn't go well. Here's some things that really were you know on the mind of the customer. More of that's becoming a reality and worked into the desktops of every agent. And they're sitting there active as the agents having that discussion so that they can sensitize that and give the right responses. Now, the agents on a global scale may not understand it, right? They might look and say, Well, I don't understand why they're yelling because it's four seconds into the call. But now they knew they called the fourth time, right? So so that you know it's a little bit of context to where.

SPEAKER_01 12:10

I think the thing I'd add to that as well is I think what we're seeing increasingly, to Mike's point before, about um tech just being thrown at tech, we're seeing a lot more outcomes driven by technology as a kind of focus, right? So you know, just having a bot sit in front of um, you know, a customer and then asking them questions at nauseum without a degree of an emotionally aware um kind of you know bot um is is counterproductive to the experience, right? And I think we're seeing that like kind of a customer reflection of the just let's call it blind utilization of tech uh coming forward. And so what we're very focused on is that to Mike's point, that if you've been routed through something for a period of time, having the first interaction you have with a person can be a very positive one. Um, and I think you know, emotionally aware, AI will be eventually so aware to Mike's point about its continuing maturity that it will know when it's time for it to disengage. And I think we're getting there. I don't think we're there all the way. Um, but but that you don't want to necessarily have a bad customer experience because it uh a bot wouldn't essentially stop, right? Like that is, I think, that is a bad scenario. And so I think at scale, having a bot be let's call it, I don't want to say the word sentient, but at least a little more self-aware than I think the earlier versions are to know when it's time to um you know engage a different mode of engagement, right? And I think we're seeing that increasingly come to the fore. And a lot of what our customers count on us for is let's call it that higher-end complexity um resolution management, whatever you want to call it, right? Where it's not the hey, where's my package, what's my bank balance? It's um, you know, I am suspected as fraud on my account, I've gone through this, I've verified this, let's talk through this problem. And I think the ability for humans to work in the loop for with sentient, or let's call it, uh, emotionally aware AI, I think is going to be the true differentiator to make mass customization or mass personalization a possibility.

SPEAKER_00 14:21

Yeah, that makes a lot of sense. Um you mentioned there, like you know, the financial example, but could you um share an example maybe from a customer um how of how recognizing though that change in in the customer's emotion affected the outcome of the interaction?

SPEAKER_01 14:37

Sure. So um uh I I think the one I would use is probably let's call it revenue cycle management, which is a nice way of saying uh if someone's not current on their bill, um so we are currently deploying AI alongside humans in the loop to kind of go through that process with someone. And in some cases uh AI works better in that circumstance, in some cases it works worse. What we have seen though is that there's a point in that journey when it flips over from being uh helpful to being disruptive. Um, and so what I think we've continued to tone and refine our models is around how we can be that, you know, kind of the the escalation curve is not the same. Mike might go from zero to a hundred very quickly, I might go from zero to ten to twenty-five to fifty and be fine. And I think what we've started to train our models on is that, right? Is to recognize that pattern management where someone is already called in because they've already gone through the self-help bot on the website, they've already gone through the user manual, and they don't know how something works, or they don't have a solution to the problem that they're being presented with. Getting that escalation point faster and accurately diagnosing where the curve is is I think going to be the real differentiator.

SPEAKER_00 15:55

And when it comes to um you know customer journeys now, where are the um biggest friction points that you're seeing where potentially AI can help to eliminate them but without feeling overly automated?

SPEAKER_02 16:08

Um then I think the best example I'd give there is the if you were in a legacy mode of CX, you were given an IVR with a number of choices and paths to go down as you got into your sort of functional swim lane to get an answer. I think then we the CX platforms of today started layering in um, I'll call it rules and other things that became a way to try to figure out predictively where the client wanted to go and maybe even built in a little bit of preceding you know, phone lookup and then who the client might be as a preceding pop to the agent, right? I think where we've seen the journeys now, and to your point, I think the question, I think we're seeing a lot more of the complexity go because the predictive nature of AI and the likeliness of having a context-aware situation means that there's lots of less choices. So the streamlining of the content for the end user, the consumer of the call or the interaction, whether it's chat, an email, a bot, or whatever else, isn't an endless desire of pick lists. And I think that's simplified a lot of this right now. Um at the same time, as you're given the option, you're learning. So the model isn't static, right? So the the ability for it to learn as you're choosing your options and then remembering the choices you made by individual options only helps the journey the next time. Um and you see it, right? When you call a call center today or a support function phone number or chat, they they know who you are. You know, welcome, uh happy to see you, Max. It's great you're calling for the 850th time this week. You're our number one user. Um, and here's all the things you called about. But but I think that simplicity is coming. I think the other part is um there's no dead zones, meaning the fluidity between having an interaction with a system to a human is getting better. Um, to Max's point, the ability for the AI engines to assist people and be able to actually give an answer is speeding up an average call to answer time. It's also handle uh average handle time for the call is becoming, you know, I'll call it less on the simplistic side. The benefit we see is we see the complexity of calls going up, which means that more and more is going to come to the business that we support because we're seeing the ability for the smaller stuff to go away, which is fantastic for us because that's that's a that's even a more stickier business for us. So I think those are the two places we've seen most of the impact. Max, you gonna add something?

SPEAKER_01 18:55

No, I think that's exactly right. I mean, I think you experienced it yourself, I think everyone experiences themselves in terms of their interactions with um technology, right? Like I think the ability to Mike's point to really get you to an answer as quickly as possible uh or engage you in a way that you want to be engaged, I think is removing friction from the system, right? Um there isn't a you know way down. I mean, there there are countless examples I know that every friend of mine will tell me about their individual customer experience. Well, I can't solve all of them. I would say certainly that the the most common set of feedback I do here is about how technology is helping, to Mike's point, really simplify out the simple transactions. And when you need to speak to someone, that person is knowledgeable, they know what they're doing, they know how to solve your problem. Uh, and sometimes you know they they do so in a way that makes you actually enjoy your experience with that that brand or that company more. Um so I would say that stripping away the volumetric calls uh or engagements that are very, very straightforward, right? Where's my package, what's my bank balance, um, and and moving up the chain in terms of complexity, we're we're seeing that to Mike's point in spades across verticals, across geographies, across clients, you know, and I think we we take it with a tremendous amount of humility and respect, but those are some of the most important calls that can make or break the customer experience with that brand, right? Like if you called us and we were calling to ask a question about your bank account and you suspected something that's not going well there, and we didn't handle that well, um, you might think differently about whether or not you want to keep banking with that bank. Um, and I think that's what I think Mike and I think is so bright about the future of our particular end of the CX business, which is that complex, high uh engagement um with the most important customers, um, is really what I'd think our our clients want to spend the most money on, right? They want to have the best service possible for those things and try to simplify out as much as possible to simplify it. And I think that would be the best case scenario for I think the overall health of the CX industry and for the ultimate you know kind of goal of increasing the customer experience, right? At the end of the day, that's what this is. People want to have a good experience with the brands they engage with. Sure.

SPEAKER_00 21:19

And you mentioned there, you know, working across different geographies, and when it comes to, you know, because of course AI can kind of help with that, you know, dealing in different markets and languages and that kind of complexity, but what does it take to do that effectively? Um, you know, while still respecting, you know, local nuances, differences in customer expectations across regions.

SPEAKER_01 21:41

That's a that's a really interesting question. I I we could probably spend all day talking about it. Um I I think one of the things that I'm most excited about, I think Mike and I spent a lot of time thinking about, is how technology sentiment analysis, let's call it technology broadly defined, whether that's AI, sentiment analysis. Um, next best action, whatever you want to call the kind of engagement of technology with people, is increasingly removing barriers from people around the world to enter into the CX space, right? So you can have more situational awareness, provided you, if you, for example, let's say you're taking a call about a ticket to an American football game and you have never seen American football in your life, you have no idea what it is. But the the tools that are increasingly being developed helps you develop a sense of familiarity and nuance that someone otherwise wouldn't have. And maybe that sponsors you to think about that more in your own personal life. I think the amount the let's call the democratization of this space, I think will continue to show better overall experiences because what Mike and I are looking for are smart, capable people that are looking for a way to increase their knowledge uh or their ability to service our customers globally and not be restricted by language or understanding of a particular customer. Um so I think what's cool about that, it takes investment and prioritization, I would say, number one. Um, but two, it takes a willingness to think very differently about what it is that we provide, right? We don't provide a location, we provide um people that are skilled, able, and willing to help with the most complex interactions, is what I would say.

SPEAKER_02 23:20

Um yeah, I'll I'll I'll put a I'll put a spin maybe in on how hard this is for companies to do. Um and we've seen quite a bit of this, right? Uh and we ask, you know, people ask us when we get into boardrooms or C-suites, they say, you know, what are you guys seeing out there and who's being successful? And uh Max is spot on with sort of the business perspective of it. And I would I I I will I'll say it technically, always be architecting. Um, the reality of of these technologies and the advancements of them are so fast-paced that yesterday's view of how to solve a problem globally is not tomorrow's view. So the architecture has to be adaptive enough, and it is not a build once, deploy many like enterprise systems used to be. So you've got to really think about it differently, right? You know, the idea of forward-deployed engineers is really meant to get at the fact that you're always architecting. Um, and that that's it's just because the pace of technology change is rapid, right? The ability to you know architect a model with a number of nested models, with neural networks becoming the thing that guides you between them is not an easy concept for most companies where they want to see an end-to-end map or a process map. Now it's about a star map or some other way of interacting from a neural network. But the other part is on the data. I think most companies are siloed in data and inclusive of division, so when you think of a company at scale globally, they run into these siloed data and sovereignty issues of that data and regulatory issues of that data, which if you don't architect at front, where you at least have a harmonization of that data, the orchestration, and how it will be accessed and used, and then the models by which they could be nuanced by region, then you're in a spot where I think you can really see some benefits in a shorter amount of time. Um, and the only other spot we could we say is don't buy a tool and put the tool as the answer to the problem. Um, the the right answer would be the tooling that you're gonna provide is gonna change so dramatically. The best thing you can do is have a common architecture that can plug many things in and don't get caught into just the pilot will prove everything's right. That's not the case. The pilots will prove you've got a concept that's right. Now the question about applicability has to be thought about across the globe. Um, so we we run into quite a bit of that.

SPEAKER_01 26:00

I I would say just to add to Mike's point there and to tie the two together, there are real examples in our business where we've invested in pilots, technologies, and we've ripped and replaced, right? The speed with which we need to be able to move to the next best technology, large language model, um, age and assist tool, um, performance management device, like you know, kind of performance management module is increasing. And our our need for continuous improvement, I think, has never been higher. To Mike's exact point, the idea that we would architect one process around one tool one time and then never look at it is a thing of the past.

SPEAKER_02 26:43

I think in our space, what will eventually emerge is a platform of tomorrow. It is not an omnichannel system where voice and chat and email were handled. It's not um uh a ticketing system that was produced for an internal routing and outcome model. I believe it's an outcome model. And I think the way the industry will mature is a new wave of fully architected AI-enabled capabilities that will serve clients' purpose with an ability to ingest a lot of data, an ability for it to rationalize it quickly, and provide an assistive technology to human interaction, a assistive technology to online interaction, but plugged into the ecosystem to learn much better. And I I I everybody always said to it, where do we start with AI? And I said, You start with the outcome you want to achieve. Max and I preach this quite a bit. So, what's the outcome you'd like to achieve? And then the question becomes the how, and those are sometimes a tool, but they're not just tools, they're an architecture that has to learn and self-heal and self-regulate. So I I would that would be the only thing we'd probably add to the Max's great points there.

SPEAKER_00 28:00

Absolutely. Well, thank you so much. That's a great point to end on, and thank you, Max and Mike, for joining us and sharing your insights.

SPEAKER_02 28:08

We're welcome. Thank you. Thanks for having us. Yeah.

SPEAKER_00 28:11

And thank you to our viewers for watching. You can find more videos and articles on cxtoday.com where you can subscribe to our newsletter to keep up to date, and you can also join us on LinkedIn to keep the conversation going. Thanks for watching, and see you next time.