From dismissed to disruptive: How IVR will shape the future of CX

So thank you for joining us, and welcome to a conversation with TTEC Digital about IVRs. So first, I wanna do is introduce some experts that I brought along for this conversation. So first, we have Amanda Robinson. She's a director in our AI strategy and design practice. I also have Cliff Martin, who's an executive director in our CX transformation practice, and Caleb Johnson, who's the vice president in our sales and sales operations practice. I'm Mike Vaughn. I'm one of the VPs in our customer transformation technology consulting group. And it seems like we've been talking about IVRs forever. They've been around for decades now, and they're still one of those things everybody chooses to ignore. Like, we all wish they didn't exist. Nobody's favorite thing in the world is to call and deal with an IVR. Yet they're still there, and they're still an integrated part of our customer's experience. And even in today's world where we have all these digital technologies and everyone's trying to push to get customers into more digital channels, over fifty percent of the interactions coming into most clients and even up to eighty percent with some are still coming in through voice channels. So the experience you deliver your customer in that channel still matters. So we want to talk a little bit more about that today. We also want to talk about the evolving role of the IVR. With all the new technologies and all the new AI in place, IVRs are changing. Some for the good, some maybe not so good. So we want to talk about that. We also want to talk about what is cutting edge. Like what are people who are pushing the boundary in this space doing right now? And also, you know, any of that conversation, I don't think helps if it's not grounded in reality. So what are the costs that people are seeing to upgrade or update their IVR experience? And more importantly, what are the ROIs they're seeing back from those investments? So those are kind of the topics we wanna talk about today. We appreciate your time. Thank you for joining us. With that, I'm gonna kinda kick it off to the panel. Let me say, you know, what do you guys view as the role of the IVR in today's customer experience? You know, Mike, I I really look at the the role of the IVR as evolved. Initially, we used IVR to sort and perhaps prioritize. The modern IVR with NLU embedded and IVA on board, is a world where we're with some deep clarity identifying, who am I talking to? What it specifically do you need today? And more importantly, what is your sentiment? Are you neutral? Are you sad? Are you coming in hot? And how can I use that combination of sentiment and intent to inform customer systems and, you know, conversations and support, on down the road in that conversation? So IVR to me is the gateway to personalization. Love that. Love that. Now I I realized in my introduction, I used the term IVR as if everybody in the world knows what that is, Cliff. You just threw out a couple acronyms, NLU, IVA. Let unpack those a little bit for me. What what are the what are those acronyms for people that might not be familiar, and and how does that impact? Yeah. So at the inner at the front of our interactive voice response, the the basic old school push button, IVRs, we've replaced those with speech recognition and voice IVRs. And when we talk about NLU, natural language understanding, we're talking about robotic and AI systems that can pull apart and understand the subject, objects, and verbs that humans speak into, the platform and pull that apart and map intents to it. And as we pull that out, and map those intents downward in the customer systems, we can really use that to inform next best action and other customer facing and agent facing, assist flows. So, Amanda, I know you've done a lot of research and and put a lot of thought into that. So some of the things that you hear when you talk about natural language, you know, background noise, different dialects, maybe even different languages or accents, things like that. What how well is this NLU technology doing all of the amazing stuff Cliff was just sort of talking about? So there's a lot of layers to it. If you're talking about noise and all that stuff that happens around, that happens at the speech recognition level, and those engines have, come a long way since that technology was added to the phone channel. So it used to be viewed as, you know, rudimentary used to be able to say a keyword at a time, and now you can speak full phrases. And not only is it recording and transcribing, cleaning out the noise, it then goes on to the the natural language engine, and Cliff mentioned the word intense. That's you know, you'll train the system to be listening for specific things that the caller is trying to accomplish, And it'll pick it apart and then get them to the right handling, either, human queue or over to you know, he also mentioned IVA, interactive virtual assistant. That's, you know, a much more sophisticated interactive experience that you can chat with using voice. Awesome. And I wonder, Caleb, like, what what are you hearing from customers? Are they seeing this as cutting edge and we don't wanna go there yet? Is it we're already there and we're behind? What what are you hearing from customers? It's a it's a fair question. So and I've actually segmented a little bit based on industries. Right? There's always these leading technology or early adopter type industries. I look at banking and financial services on the retail travel side. We've seen those companies kinda early adopters of they want they're they're using the voice channel as an opportunity to provide a better customer experience. Right? But we'd still see a lagging, industry. Like, you you think about some of, like, the manufacturing or engineering type of industries where it's a very b to b type of business, and the voice channel is truly just somebody's calling in. They have a destination they're trying to get to. The r the IVR is a way to route them, drop them off somewhere. Now that being said, I actually had a really interesting customer, conversation with, engineering company that has a hundred different project sites going on at once at at any given point. And they're actually looking at the IVR conversion into an actual virtual assistant to be able to start giving them more real time data, being able to say, hey. We know when deliveries are coming in. So actually using it almost like an Internet of Things type of application where you're able to capture, okay. Tell the delivery driver to call this number. It's gonna notify our project systems that this delivery's on the way. It's gonna send out other notifications. So we're starting to see this migration, I think, of how do we use the technology, not just the traditional, let me get you to the right destination, but it starts to almost become multimodal where it's becoming I'm utilizing it to trigger other actions, other automations that historically the IVR couldn't handle. But with the advent of AI, the advancements that have happened, you know, with Gen AI, especially over the last, you know, eighteen months, things are changing rapidly. And and how which, you know, helps the company, obviously, in the efficiency of the of the communications and things. How have you seen similar types of those applications when it comes to the the customer experience of an individual caller? If I'm calling my banker, I'm calling, you know, a service provider I'm working with? Yeah. So it's I I look at it two ways. One is you had traditional operating hours that exist in a contact center where I'm a customer. I need to get help. No one's possibly available at midnight, but I still need to actually take some kind of action. So this shift into using AI within the IVR, the idea that I can actually still provide somebody, self-service actions. I can let them tap into back end data that may not be on a portal or it's not in an app potentially, or it's just not accessible. Right? So now I can I can extend my operating hours? That actually really helps from the customer experience standpoint. I also look at it from a a volume or, like, a through throughput concept of everybody maybe trying to get help at ten AM on a Monday for some kind of banking question. There's an event that's happened. Right? So you have this, you know, spike in volume that's coming through. There's just not enough people to answer the phone. And so if we can use an IVA to be able to still service some of those conversations or move them to a different channel, now I can move it into SMS or I can move it into, you know, a mobile app conversation. Right? So the customer experience is improving in that case. It's not we're not even getting into, like, what's the actual ROI benefit on the back end yet. And I think the the other side of that same coin when you talk about the customer experience is also talking about the employee experience in the contact center. I think, Cliff, you know, with with these things we've been talking about in the technology we're putting upfront, how is that impacting operations? How should, clients think about their operation if they have a different sort of front end distribution system coming in? Yeah. You know, operationally, matching intent, to our people, we now can unlock the power proficiency based routing. So instead of dumping, customers into a pool where all agents are assumed to have the same skill set, what if we now match intent and sentiment with our best people? So I know who my best product champions are. And I know every contact center in the world has got folks that we wish every escalated customer went to because their verbal judo is so good, and their empathy is off the charts. And their ability to listen and deescalate is so great that I want escalated customers queued to them, almost because of their high results on experience. So I I think from a people standpoint, we've gotta get really good in contact centers about understanding who are my process champions, who are my customer champions, and how can I use powerful routing to create different experiences to match up great customers with great people? I do I do wanna and, Caleb, you kinda tipped us into the next topic, but I wanna wanna hit on, was around the innovations in the space. And you mentioned IVA and virtual assistance and being able to tap into back end data sources. You know, the idea of the virtual assistant's been around for a while. You know, the the world kinda changed with the sort of advent of, generative AI. So I'm curious, Amanda, like, what are you seeing as you know, what are the cutting edge kind of applications today? And more importantly, where do you think we're going here in the in the short short term future? Sure. So I think we're generally seeing a raising of the bar in terms of the quality of experience we can offer over IVR. You know, back in the day, it used to be, you know, press press one for this, press two for that. You're putting the burden on the caller. What we're seeing now is you put the customer first. You wanna greet them as though you know them. You want to, you know, reduce the friction of having them, you know, tell me your address, tell me your name. You don't wanna run them through all of that. So, and then you want them to be able to say things the way that they would naturally say them. So all the technology under the hood that we're seeing that powers that, you know, things we've talked about already, advances in CRM, getting all that data up to the IVR so you can use it in the experience, generative AI to take scripted responses and, you know, make them more conversational, enhancements in the language, even things like biometric identification. Anything you can do to reduce the friction for the caller just improves the quality of the experience. Excellent. And is that, you know, Caleb, kinda piggybacking on that in the in the future? Like, how much are customers looking to do that? And and I guess, to me, that might be a different skill set than than a lot of companies have in house. So how should companies think about, you know, the the skills and the abilities that they would need to really be able to deliver on those technologies? It is a different skill set and at multiple layers. Right? In the traditional model, you might have a, a voice engineer that's very good at going in and scripting what's gonna be the route that's gonna take somebody from point a to point b. Not that that person couldn't learn the new skills, but that that role exists in most companies. And we're seeing that the maturity of the AI platforms that are out there now have really helped move to a more of a no code, low code type of interface. And so that's much more accessible. And similar to, you know, twenty, thirty years ago, you had the big tech companies coming out with certification programs and free training in in these learning pathways. The AI companies have done a very good job of building that. They're advertising that, I think, even down into the high school level of getting people to where they're understanding how to use these applications so that it is a rescaling of the workforce. Now from a kind of can you do it easily? Right? There is a, an advantage when it comes to you don't know what you don't know. So having the right people around to be able to say, let me help accelerate this for you. Let's skip some of the hurdles that you're gonna run into that you can't see yet. Because getting into things like conversational design, some of the kind of things that are, I think, a little bit more artistic than they are scientific to the approach. Though, that's where we see companies looking for help. They're looking for support. They don't want this to drag into a two year project. They want it to be something that can happen really quick because they are looking for, like, you know, that that return on their investment. And there's I think there's demands at the executive level that's happening where they're saying, hey. I wanna have I want the advantages of Gen AI. Everyone's telling me it's great. It's available. Let's go fast. Right? So so what what kind of timeline expectations are you setting then? If if customers desires to go fast, what's fast in this world? It's it's interesting. So you can get a a virtual assistant up and running in an IVR under six weeks. Now it's not gonna solve all of your problems. It's probably not going to take over every routing scenario that you have, but it is accessible, especially when you get into some of the more, kind of foundational elements of, I just need to do natural language and I need to do intent recognition. That can happen really, really fast. Large language models, prompt engineering, the ability to build things really, really quick, that rapid development, that's all come into play now in the most common systems that are out there. So we we are seeing that happen fast. Now there are very mature organizations that that have many lines of business. They have very complex IVRs. You know, that is gonna be a longer project. But if we're seeing it shift into much more of an agile, continuous improvement optimization type model to where you don't have to build everything on day one. You're able to go continuously update and bring in new functionality, new features into that experience. That's interesting. I Amanda, how how do you kind of counsel your clients on on that sort of starting small or starting, maybe small is not the right word, but start in a prudent fashion, you know, kinda test and learn and grow. Wrong approach. So there's a lot of ways to contact with. And, actually, we've got packages that we've been putting together recently for these clients that have very complex IVRs. They know they need to modernize, but they're terrified of touching the spaghetti code and breaking everything. So there are ways to do it. What we do is we come in and we take a look at, you know, what are people calling about? What are the, what would be the best fit in terms of ROI that we can give you? The high volume calls that don't need to go to an agent, and we pull those out. We put together, you know, sort of a parallel thing. And so you start with one, you prove the value, and you get more training data from that. You put an open, prompt. How can I help you today? People will ask for whatever they want. You get that data. If it's the one that you've set up the intent for, you route it to that one. Everything else can go back to the existing experience. So it's really low impact. You put it to the side, and then you've got that stream of data you can analyze going forward. Amanda mentioned data in there. And you might be going there in this conversation, but I'll just kinda open it up to the group because I think it's a really important thing. As you were talking about, like, can you go fast? What are what are the project links? Those kind of things. I'm curious what you folks are seeing from companies trying to make more data driven decisions around this as they're looking at moving towards using AI, you in the voice channel, what like, kind of an approach or what you've seen recently. And I'll I'm opening it up to the floor, so whoever wants to jump on that one. So it's funny. I've often seen the opposite. Companies will come in with an idea in mind that they think is gonna give them the most ROI, And their first instinct is not necessarily to look at the data. Oftentimes, they don't even have the right kind of data. So that's the first thing we counsel them on is how do you get the right data, and then what do you glean from that? Well, how do you use that to drive your roadmap going forward? And I was gonna say the same thing. Speed to market is speed to data lake. All design work is intent driven, and intents are informed, and routing decisions are made by access to personalization data that is embedded in our core systems. And your speed to market on a lot of this advanced automation is simply defined as your speed to servicing and linking this data so that advanced, systems can route from it. That is I I I look at that as the longest poll in any tent. I I think the thing I would add to that is data is not limited to bits and bytes stored in a CRM or a policy management system or something in the back end. What I see to a lot of clients have you know, if you go sit next to their agents and you see where their agents go to get answers, It's a lot of tribal knowledge of well I know this answers in this database. This answers in this policy document. This one's on a SharePoint site that hasn't been updated for five years. So a lot of times too I think the the when you talk about the data it's also where's where are the answers? Like, where do your agents go to get the answers to the questions people are calling about? And, you know, until we get that and have that cleaned up, we we don't really have a good base to train a virtual assistant on either. And a lot of times, you know, we rely on the human agents to cover up the fact that, yeah, the knowledge is in five different places. The agent know where to go. Yeah. And and that's because they've had the battle scars of of, fighting to find the data. I do wanna come back to one thing, though. So so all of us have used the term intent and and how, crucial intent and intent identification is. So I'm I'm curious, like, Amanda, can you can you kinda tell us in your world, in in the IVA, in the in the automation world, intent means something a little bit different than those of us dinosaurs are used to looking at call disposition reports and and call reason codes that the agent assigned at the end of the interaction. So talk to me a little bit about why it's important and how that differs from, you know, the reason for the call. First of all, the reason for the call, these disposition codes that are assigned, first of all, they're they're applied by agents at the end of the call, if you're lucky. And the schema for them is defined by the business. And, you know, oftentimes, they're lagging behind. The agents can't go and just add a new one. So that's one thing. It's from the viewpoint of the business. An intent for natural language purposes, is anything that the caller is trying to accomplish when they call you. And it encompasses things we call utterances, all the different ways they might ask about that. So it's think of it as sort of like a cloud of intention, what someone wants to do in IVR. And so you can't take those call reasons and just put them at the top and think that that's going to work. Even if, you know, the spirit of what they wanna do might map to those, oftentimes, there's many more of them, and they're a lot more complex to manage. One core thing that gets overlooked, like, it was said earlier that, you know, conversation design is somewhat artistic. There is a lot of science to it as well, especially around the language side of things. You often you know, companies will especially when they're designing their own IVRs or trying to get something live quickly, they overlook the fact that their customer base is very diverse. They speak differently. They're from different geographic areas, different levels of education. And so they come in and, you know, especially if you've got a development team throwing in those training phrases and not thinking too much about it. It's from their point of view, and people are saying things in a completely different way. So making sure that you're studying how people say things and incorporating that is a key factor in success and performance. So so sort of round out on on the topics here we're talking about, wanna get go on and talk a little bit about cost and and ROI. Right? Because anytime we're talking about new technologies, the first thing a lot of clients think about is how much is this gonna cost me? And and, you know, how do I make a business case? How do I go to my my CFO or my CIO and make a business case that this investment has a solid return behind it? Caleb, how do you counsel your customers who now need to go put together a business case for an investment? Yeah. I mean, there's just like a lot of software projects. Right? There's two core elements when you look at the pricing. There's gonna be a cost to get it stood up, get it running, some kind of initial implementation piece of that. And so we we do go through work with clients on those types of things. We see a desire for companies to actually bring some of that skills, in house over time to where they're not necessarily going out and having to pay someone every time there's a change to something. Right? And then there's obviously the the ongoing run rate, what I'll call the consumption cost for these things. And so at the end of the day, if you take a traditional call, let's say it's, you know, a five to six minute call that's gonna go through an IVR, it's gonna get to a human, they're gonna have to answer something. Most companies can calculate what that cost is gonna be. If you're able to move some volume into being able to self-service that or deflect it back into a portal or into an app where the call is not having to be answered, that cost of that interaction is definitely lower. And so it really comes down to some pretty basic calculations of how much call volume do you have coming through your system, how much of that do we think can actually be automated or moved into a self-service type of interaction. And that's a pretty easy calculation. And then just simply building out over time, this is projected. Right? This is what it's gonna look like over the first year, the second year. And most likely, you're gonna have some kind of ramp and additional use cases or intents that are gonna come into the virtual agent that you're gonna get a continuous benefit from. So, most people are able to build that business case, tie it back into either a a direct cost savings or a bigger business initiative to say, hey. We really want people to use the portal that we've built for them to be able to service their loan or service their purchase that they have. Right? That's a that's a bigger initiative, and you'll be able to enable that by utilizing something like a a virtual assistant. Alright. Those are great business case. I think there's there's probably also underlying telco costs and things like that that also go into there that you should be getting it. And you mentioned before that some of these can be as fast as, say, a six weeks for a limited, you know, a deployment with maybe some limited capabilities and things. So what kind of cost do you think in general? I know your the answers that depend. They would need requirements, but if you tell someone, hey. Roughly budget how much? Less than a hundred thousand dollars. So a six week project. Let's go get an integration done into an IVR. Let's have a a data source that we're gonna tap into. Let's do five intents. Give give you some, you know, natural language in the IVR. That's that's anywhere from, you know, seventy five to a hundred and twenty five thousand dollar project. That hundred thousand dollar range is is pretty safe when you're looking at that initial production, but we have something small. We have something running. That that that's also gonna get benefits. So I can start getting some of those benefits and and start learning and and going further like Amanda was saying. That that's awesome. From a crawl, walk, run standpoint as well, starting off with that high volume, low complexity, low risk of emotionality intent that's highly transactional, highly repetitive. Let's go automate there first because that's where there's the most lift for the contact center organization, the most static in our capacity plans. So there's also an approach to where should I automate first, so to keep to to almost always throw great cost, find volume and low complexity first, and you'll you'll you'll never make a mistake. So I'm curious on, if you have thoughts around it, just to round all this out. Where should someone start? If if someone hasn't already started down this journey, but maybe they've done some research, they know they're right there and need to take the first step or things. What are there some strategies for success that that any of you would recommend? Yeah. I'll I'll I'll start with this because I think that Cliff actually really articulated this well. Understanding what's actually happening in your customer journey, that's a great place to start. And as simple as that might sound, a lot of our the companies that we work with, they don't have it well documented. They don't necessarily know volumes. They haven't gone through the disposition reports. They haven't gathered why are people actually calling us. Right? And so just gathering that kind of information as a starting point, it's a great place to be able to whether you're gonna do it in house or you're gonna work with a partner to help go build this out so that you can walk in the room with some kind of general idea of, we understand where the volumes are at. We understand where some of the complexity is coming from. We have a business case of we wanna reduce call volume by twenty percent. Right? Being able to have something like that that you can bring to the table is a huge advantage that will definitely speed things up. I don't want companies to feel like if they don't have that data that they're dead in the water. There are services where you can take call recordings or even transcripts, run it through we even have something called conversation intelligence that can do that crunching for you and not just get the code that an agent would put on it, but track everything that gets discussed over the life cycle of that call. And that would give you a lot of insight as well. But it is really important to Caleb's point. We see a lot of these, initiatives, launching either from the business side where the contact center wants to reduce volume or increase efficiency or the IT side where there, you know, there's a technology consideration to be had and they get a little too far down the road before those two paths converge. So I think one of one of the best practices I would say is, you know, make sure your your IT and your technical team is working very closely with your operations and and your business teams, so that you're going down that path together. Again, we see a lot where they've gotten pretty far down the path and then realized, you know, those paths did not come together the way they had hoped. So with that, I wanna absolutely thank the panelists for for taking time out to share their expertise and their opinions with us. And equally importantly, I wanna thank you, the audience, for spending some time here to to think about the IVR. We'll have some things popping up here where if you want more information or next steps or where do you go or to contact any of us, we'll have some links coming up here that can, direct you to those resources. Ttechdigital dot com is always a great resource and we look forward to, speaking to you and and please feel free to reach out. So thank you everybody and, have a fantastic rest of your day.