5 Best Practices for AI Self-Service Without Compromise
Brian Cantor: Ladies and gentlemen, it is time. Welcome to today’s webinar, Five Best Practices For AI Self-Service Without Compromise, brought to you by SmartAction. I’m Brian Cantor, principal analyst for CCW Digital and your host for today’s discussion. Now you likely won’t find any contact center leader out there who doesn’t trumpet the theoretical importance of letting AI powered self-service handle interactions. You probably won’t find a single customer who isn’t asking for a more convenient, more autonomous experience, particularly as we look at the omnichannel world. You definitely won’t find a technology provider who isn’t confident in the ability for organizations to use conversational AI to transform voice and chat automation. But here’s the problem with all that, you also won’t find many organizations that are actually taking advantage of this new normal. They’re still so reliant on agents for even the most basic, simple, redundant transactions leading to greater customer effort and more business inefficiency.
Quite simply, this is not where we want our contact centers to be. It’s not where the customer journey should be in today’s era of competing on the customer experience. So it’s time to break from this trend and SmartAction, who I’m joined by today has actually been helping or more than a hundred organizations transition to conversational AI. Nearly a dozen of them have been Fortune 500 providers with high stakes interactions, massive customer bases, and they’ve really been able to make that transition work in their organization. But the key here and the key for today’s discussion is not just to talk about why SmartAction is awesome, but it’s really to share the best practices that they’ve learned from these experiences of working with so many different brands and really understanding what it takes to go from really promising and exciting technology, to really amazing and customer centric experiences. And so of help us get there, we have Tom Lewis of SmartAction.
Now, Tom, he’s so knowledgeable in all areas of the customer contact space and with good reason, he spent 13 years as a partner at Deloitte Consulting and he actually led their customer service advisory practice. He advised their powerhouse clients on everything from contact center strategy to technology and operations, the customer experience, and pretty much everything in between. And he’s now the CEO of SmartAction and it’s absolutely my pleasure to welcome him. So Tom, how’s everything going?
Tom Lewis: Fantastic. Brian, that Good Morning, Vietnam invocation at the beginning. I appreciate it. And looking forward to this discussion today.
Brian Cantor: Well, we’re finally going to move from the theoretical importance to AI, to actually getting there and not having to give up anything operationally or customer centric wise, that’s where my excitement is coming from. It’s a phenomenal topic, and I know you have a lot of real world practical insight to share. So very thrilled about today’s discussion now, before jumping in, I just want to give some housekeeping notes for the audience. So we’re aiming for about a 30 minute discussion, but we want it to be interactive. I know Tom has plenty of content, but if questions do arise in your part, please type them in on the Q and A box on your screen and we’ll do our best to get toward them whether it’s as we go along, we’ll also try to save some time for Q and A at the end of the discussion. But for now, let’s get into the content that Tom and his team have prepared here. And so Tom, before diving in, I just hope you can give the audience a little bit of an overview about who is SmartAction and what are some of the things you work on?
Tom Lewis: Yeah, sure. Brian, and I’m not really going to spend a lot of time on the company overview. I’ll make it quick, but I think it’s important to know that we started as an AI research company back in 2002. So we’re not the new kids on the block as it were, when it comes to AI development, machine learning and natural language processing, all the things that are hot and being talked about today. Where we fit is with companies who are already using contact center platform or have an IVR. And it could be on Genesis, Avaya Five9, inContact, whatever, on prem or in the cloud. And they lean on us to expand the self-service automation in those environments, essentially automating the conversations being handled by live agents today. And so you can see our awards and recognitions on screen, like the Gartner Peer Insights, where we happen to be top ranked on that. So that’s by customers of all sorts of different platforms. So interesting to go look at the Gartner Peer Insights.
Brian Cantor: Now one of the things that we want to start with right is really this idea of a perfect fit or balance because we really, as we’re starting to build our workflow, we need to understand that proper interplay between where conversational AI fits into the organization, as well as where and how it can coexist along live agents. No one is necessarily advocating for getting rid of the human component to our contact center, so how do you figure out what are the areas where you really want to leverage conversational AI and how it can therefore starts to compliment the actual live workforce that you have in place?
Tom Lewis: Yeah, sure. For most organizations, it’s understanding where to start. Almost every organization we talk to has probably two or three call types that are perfect candidates for natural language automation. And that’s going to deliver a great customer experience and frankly, a near immediate ROI, given the cloud-based nature of it and then they would sort of grow from there. But it can be hard if you’re trying to do, let’s say, a big bang implementation and automate practically everything from the start. It’s not to say that we don’t do big bang implementations, we certainly do that. We’re in fact, we’re doing one right now with one of the largest hotel chains in the US including 13 different call types, going into production at the same time. And so the point is that we tailor the approach to the client and prioritize the fit based on the client needs. And we’ve got customers where we started with one call type, but now handling upwards of say 25 different call types. And for this reason, we have an entire layer of solution consultants that do nothing except help organizations build a business case, the roadmap, basically to help figure all of that out.
Brian Cantor: And now, as I mentioned earlier, a lot of organizations are talking about AI, thinking about AI, but there’s plenty out there who really haven’t taken that step to actually implementing it. And so for those who are listening or those organizations out there who have really never even dipped their toes in the water when it comes to conversational AI, what’s really involved with getting started. Is it something that they can kind of take the initiative on their own?
Tom Lewis: Well, I can explain it at least how we do it and then to the extent they can emulate it. Sure. What we do initially is as something that we call a call study. It’s an analysis into what are the top call drivers coming into the contact center, which includes assessing each one for its fit with conversational AI. And we literally sit with live agents, supervisors, we review reports and we talk with the business. What we’re looking for is the perfect marriage where automation is basically a no-brainer, but previously hasn’t been able to be economically deployed. And how you get there is by knowing what can be automated, projecting the automation success rates, and then map that volume in costs, not just the cost to build, but the cost to manage, maintain, upgrade. And what we’re looking for, the ideal Holy Grail here is the self-funding business case that’s obvious to everyone.
Brian Cantor: Now let’s dig into that a little bit deeper because that assessment portion sounds really, really interesting to me. So when you’re kind of assessing those capabilities and where you want to start, what are some of those things that you’re looking at?
Tom Lewis: So probably considering four basic elements, one is really looking at how natural language is going to work in the environment, both the recognition and the cognition components of that. We want to understand what the availability of data is to personalize and otherwise handle the conversation and execute the transactions to the extent that we can use that data to enrich a prediction of the call. Predicting customer behavior would be another one. And then finally, how adaptable is it to an omnichannel or at least a multichannel approach? So those are some of the highlights around that.
Brian Cantor: Yeah. And I think it’s so important and so valuable to really start to understand what those capabilities are. And that’s simple assessment right there is going to give you a great picture of what you need to make the most of conversational AI, but also just as importantly, where it can go and what impact it can make for your organization. But before you really capitalize on all those benefits, it’s really important to go back to the call study, because you have to understand what’s driving these interactions, where are these calls coming from and why are they coming into your contact center? And what does that mean for the types of interactions you can handle with an AI driven tool versus other resources within your organization? So I’m wondering if you can kind of walk through some of these key call drivers and let us know kind of how they fit in different buckets, for instance, starting with simple calls.
Tom Lewis: Yeah. So the end of the ones that follow the same repetitive, linear, or pretty near linear process. Humans almost feel like robots doing that job and one of the questions we ask is who or what will be handling the simple call types now. And if it’s a touch tone or directed dialogue we ask, are you and your customers happy with that experience? Or would it be improved with the natural language capabilities? Moreover, could you contain more or get the job done quicker with natural language? And oftentimes the agents are handling simple call types that are very processed driven, and they’re doing it because touchstone won’t work or just as in the good experience. And so let me give you an example, the second largest insurer in the country has auto body shops that call in to get approvals on claims. And in that flow, there are 17 back and forth turns in the conversation that involves capturing lots of different pieces of information, including a long, alpha numeric policy number and the system’s got to accept it as fast as the body shop is saying them along with other things. And so that’s hard for a human to do correct the first time. And it’s impossible obviously for an IVR. And that’s just one example of a process driven call type that could not be automated until the advent of a good conversational AI.
Brian Cantor: And now, Tom, I really appreciate you bringing up that example, really illustrating what a simple call looks like. That said, I do feel like the theory of using AI driven tools to handle symbol interactions and that’s kind of the cliché, that’s what everyone kind of expects. That’s what everyone at the high level was looking for. And I think then when it comes time to those more complex interactions, that’s where we start to see a lot of hesitation, because we know we have a lot of context and our leaders just flat out ignoring the idea of using AI for complex interactions, because they feel that they’re just too difficult or too random or too unpredictable to really be able to automate. They’re afraid that there’s just going to be too many exceptions and without a live agent kind of at the helm, it’s just going to go off the rails. And so what I want to ask you though, is that still the case? Is AI just for simple interactions or does conversational AI have the capability to handle these more complex conversations?
Tom Lewis: Yeah. It’s these complex call types where the power of the conversational AI can really excel. What we’re looking for is the percentage of calls that follow the same repetitive process, including the exceptions but it’s well-defined. In other words, although the exceptions could occur that need a live agent for critical thinking or judgment, is there a happy path, even a broad, happy path at a large number of calls take that is process driven, even if it’s complex and that’s the swim lane for the virtual agents. And so an example, just very simple example, maybe it doesn’t get super complicated, but Office Depot wanted to automate their order returns, but returns can be super complex. These are the exceptions you were just talking about, Brian. And we help them find a path that was very process driven and as it turned out, the vast majority of customers wanted a refund instead of an exchange. So as long as the customer chooses the refund path, then they stay in automation with a virtual agent and everything else outside of that gets transferred to a live agent. So for Office Depot, this meant deflecting a large number of calls that live agents would otherwise have to handle. And the perception out of the gate was, well, we can’t handle returns. Well, you can get a large number of them through this approach.
Brian Cantor: What we just went through, I think, were two types of say interactions. We had the more simple conversation, a more complex one, but they’re ultimately like kind of the full on interaction and problem solving or processing transactions. Now the different call types and the different opportunities to leverage automation to improve efficiency, those go beyond these simple and complex kind of interactions. And so, Tom, I’m wondering if you could maybe share some other use cases of where a conversational AI tool is going to create a stronger connection with customers or lead to a more robust context in your operation.
Tom Lewis: Well, so first you should probably greet the caller with natural language and if your IVR touch-tone menu is working, then maybe you should keep it and just use the virtual agents to automate some of the calls as we discussed. In other words, instead of transferring to a live agent, you’re just transferring the call to a virtual agent. And however, if the experience would be better with natural language greeting or help to contain more on that upfront piece, then we would consider the virtual agent for the front door routing instead of, the default of the IVR that may be in place. And so, as an example, we had a national online retailer that uses virtual agents to greet the customers with an open ended question, how can I help you? And like a lot of our other customers, the virtual agent is listening for a number of intents that keep the customer in automation and then if anything outside of that, is transferred to the appropriate live agent based on what we heard.
And the second thing you need to take a look at is what you’re doing from an outbound standpoint. And maybe there’s some outbound calls you’re not doing because it would be too expensive with live agents, but, frankly, it’s also it’s very often too hard to do with a robocall or not effective because of the lack of transaction capabilities. And so an example, there might be Penske Truck Rentals. They actually used robo dials and texts to remind renters of the upcoming reservation. But if the customer needed to cancel or change that reservation, then they would have to call in the call center and it really didn’t create a great customer experience. But by using the AI powered virtual agent to do those outbound calls, then they handle the cancellation or rescheduling right there without the customer having to call in and it deflected a ton of stuff from ever going back into the call center.
Brian Cantor: No, and I appreciate you going through the different use cases, Tom. And certainly just in that five minute kind of introduction to some case studies, you’ve shown how conversational AI is doing more than maybe we sometimes give it credit for, or some context and a leader’s thinking it can achieve. But at the end of the day, there are still scenarios where it makes sense to have live agents leading the interaction. And so there it’s really important to identify, I think, that distinction between what should be in the hands of conversational AI, what should be in the hands of your human workforce. And I’m wondering if you can give us some guidelines for really navigating that balance and making sure that we’re really drawing the line in the right spot so that our organization benefits and our customers benefit as well.
Tom Lewis: Yeah, well, sure. And so the way we refer to this is guard rails, and when we talk about guard rails for the virtual agent, what we’re really talking about is handling rules and what should be handled by the virtual agent where we know where the CX is going to be fantastic. And when we should transfer that call to a live agent. If the virtual agent has access to customer data, you can use that unique knowledge of the business and that data to establish handling rules at any point in the conversation flow. And so just for a couple of examples for the largest pizza chain in the US, they have this concept of rapid reorders and it means the caller is ordering the same thing they ordered last time. And so this means that the caller, if we’re able to execute against this, meets several pieces of criteria, such as a previous order that can be duplicated, assuming that the store has the right inventory and that it can be delivered in the next 90 minutes, for example.
And so maybe another one is in the medical space where orders covered by Medicare have a very complex process, as you can imagine. And virtual agents aren’t handling all of it, but two thoughts. One is if the ship dates within the next 30 days of the call, or if it’s one of these automatic recurring orders that will be shipped the next day, then the rules around that are perfectly capable for the virtual agent to handle. Otherwise, it would go onto a virtual agent. And so, maybe another one is around AAA. So AAA clubs use virtual agents for emergency roadside assistance calls. And after the authentication, the virtual agent will capture the intent using natural language and in their case, it’s how can I help you today? And it’s listening for one of basically seven things, seven intents that the members can ask about it.
Wasn’t always seven, but it’s seven now. And it’s things like, I got a flat tire, I’m out of gas. I got a dead battery, need a tow, those sorts of things. If that, if the member says something outside of those intents, the ones that we know how to handle, like let’s say, I’m locked out of a vehicle or worse yet, maybe a baby being locked in the vehicle. Then the virtual agent will quickly transfer those calls to a live agent with all the information it captured and the reason they’re calling. And this transfer is an example of what happens when you go outside of that happy path. And so AAA identified the seven I just talked about were perfect for virtual agents and everything else basically goes to those live agents.
Brian Cantor: It makes perfect sense there. And so as we start to look at the next best practice, and that’s really the idea that virtual agents need data, and that’s very zippy. A Great kind of catchphrase for the best practice, but Tom, I want to give you the burden right now of actually walking through what that actually means within the context of your typical customer experience function.
Tom Lewis: Sure. And to be clear, the focus of this best practice is to access customer data. Live agents require access to the customer data for the purpose of reading and recording data and so forth to support their processes. And it’s no different with the AI powered virtual agent, except there’s a little more to it. And so first it’s used for the purpose of guard rails, which we already discussed, and then beyond that, it’s for the purpose of personalization. And so in this day and age, people expect that the machine or the bot should be more intuitive. It has all that data and so forth. It should know more about you. It should also streamline that authentication process. And so one of the best uses of customer data is for prediction. And this is where the cutting-edge aspects of AI self-service is going to reduce the friction to complete the call, more of this effortless experience.
And the best situation is when a company has enough of the data on their own customer, that they can help with that prediction and sort of convey it over the AI. And so here’s an example in the healthcare space specifically around EOB, so explanation of benefits. These can be confusing documents, customers often have questions about these, and if the virtual agent matches the caller ID and they’ve authenticated and so forth, and it sees that there is an EOB recently went out to the caller, then it can offer a different kind of greeting something like, “Hi Bob, are you calling about that recent explanation of benefits we sent or something else?” And the customer realizes we have the right data and they’re much more likely to engage with the system just because of that.
Brian Cantor: Yeah, absolutely. It sounds like you’re really getting to the two basic pain points a lot of people have had with traditional self-service. One is that they just, they don’t feel like they don’t trust it to solve their problem. They don’t feel as if it’s actually going to lead to anything and being able to show that, Hey, we know why you’re likely calling. We know why you’re getting in touch. That goes a long way in building credibility for the platform. Then the other thing is kind of that broader customer centricity benefit. When your system shows, hey, we’re committed to knowing you, the customer that even our digital tools or even our bots, or even our IVR knows who you are and why you’re calling. It really sends that message of, you matter to us, you matter to our organization and we’re going to do everything in our power to help you.
So it’s a great mixture of that practical education. Hey, you should use this channel, but also that broader idea of, hey, as a business, we really support who you are and we want to make the most valuable experience possible for you as one of our loyal customers. Now, speaking of, kind of moving away from the idea of that simple IVR, the idea of the IVR just kind of being there as almost like a waiting room before the customer really got to someone who could help solve a problem. Those days are over. Those days of kind of an independent chat bot where yeah, it’s cute technology and it’s fun, but it ultimately doesn’t go anywhere. Those days are over as well. Because while today’s customers are demanding self-service, they’re demanding that against the broader backdrop of an omnichannel, a consistent and seamless experience, no matter where, when or why they’re contacting. And so given that need for consistency and seamlessness across the entire journey, what does that mean for the state of virtual agents right now? How does this tie into this best practice that virtual agents need omnichannel capabilities?
Tom Lewis: Yeah. And it’s the headache of managing and training all these disparate technologies, independent of each other that have different capabilities, different integrations, different customer experiences, and maybe even different vendors to manage. And so these applications, they fork further and further away from each other with time. And often data isn’t shared, reporting isn’t consistent. There’s no unified management. I could say more on that, but the use cases and demand will only continue to expand and I think people can probably relate to that. But I’ll give you a specific example. So TechStyle Fashion Group, and they’ve got brands like Fabletics, ShoeDazzle, JustFab, they show the shift happening when they examine their customer preferences and the landscape of conversational AI technology that was available to them. And they deployed the AI powered virtual agent first in voice, and that deflected thousands of repetitive calls related to order management. But then they scaled the solution to chat, to provide a consistent CX to the members, same backend integration, same business rules, same everything.
Brian Cantor: Now, as we move into our fifth best practice, and this is one that so often gets overlooked or misunderstood, but it’s absolutely important to the success of your overall operation. And that is that virtual agents need a team of CX experts because what’s funny is that a lot of times when we’re talking about the impact of conversational AI or sort of any automation effort we’re talking about, hey, what human labor can we eliminate or where can we use an AI tool to help with customers and not use a live agent for that task? And that’s great, it’s efficient, sometimes it’s better for the customer as well, but that often ignores the fact that you still do need a human workforce. You need a team behind the scenes to really build, run and optimize your conversational AI tool. Now, this is something, Tom, that I know you and your organization are very passionate about. It’s something you take very seriously. It’s a huge part of who you are as a business and what your stance is on the conversational AI topic. So I really want to give you a chance to walk through this very important best practice.
Tom Lewis: Yeah. And I, to some degree it’s the best for last. Now I’ll skim through some of the items on here for the sake of time, but first off starting with the CX consultant and solution expert, those are the ones that do the call study and they get the scope and the approach and sort of the roadmap set out. And then you’ve got the project managers day to day making sure that things go live on time with the right performance. But the key really comes into that CX designer. And it’s one of the most important roles. They’re not simply a conversation flow designer, and they’re more of a behavioral scientist and your biggest customer, end customer advocate to ensure conversation flows are effortless and effective and have the right guard rails and handling rules like we were talking about earlier. And so for lack of a better word, they help manipulate the conversation flow to lead the customer down the best and most effortless path.
And this all seems very logical and common sense, but our experience has shown in some ways it’s neither logical or common sense, and it’s somewhat of an art and okay and a science. In fact, just anecdotally, I had a discussion with a number of our project managers recently and asked how often does our CX team recommend something and then the client pushes back hard and we try to dutifully oblige and they suggest something else. And ultimately, now here’s the kicker, months after going live, you end up re-going back to that decision and the client finally agrees with you and says, “Yeah, we should probably should have done it that way.” And scary thing is they said, 98% of the time, that’s what happens. And so the intuitive nature around conversation flow is not as intuitive as you think.
And we definitely have a very experienced team that intimately understands that human centric aspect within the conversational AI. The CX design is perhaps the most important thing, the most important best practice we’ve got here in this discussion and the technology is awesome, but technology alone is not going to make for a great experience. I have an example here and just looking at the numbers on here, it’ll give you a little of a perspective. This was from a national travel agency and they wanted to use our technology. They are using our technology, very happy with it, but it’s purpose was to qualify potential bookings, customers. They were looking for people who are actually going to book a room. If they weren’t going to book, they treated them differently. And so they played a very heavy hand in the design, frankly, of the conversation flow.
And you can see the results on what they asked for, which is what we delivered. And then that month later we came back and sort of re-did the way that it was designed. So I’ll spare you the before and after recordings for the interest of time, but you can see the difference that the rearchitecture had and the results speak for themselves. And it’s important as the technology is, again, it’s important to have the best CX design. So let me go back to that circle of the different components and move on to the customer insight manager, which is an interesting group to design this ongoing monitoring and improvement of the application, which means pouring over data and examining containment rates against outcomes, and a lot more. It’s not a tuning exercise, it’s a lifelong journey around perfecting the outcomes. And then finally, the success manager who helps drive that vision we talked about earlier. So, as you can see, there’s a lot of people required, a lot of players here to orchestrate that best CX outcome.
Brian Cantor: Now, as we’re approaching the end of the half hour, Tom, and I really want to do our best here to get one or two questions in. We’ve had so many from the audience already, but I also want to give you just a chance to give some final thoughts and maybe why SmartAction really is so passionate about this topic and perhaps more importantly, so knowledgeable about what it takes to go from the theory of conversational AI, to the rewards of a great customer experience that leverages automation correctly.
Tom Lewis: Yeah. Thank you, Brian. I appreciate that. And I’ll try to make this quick, just hit the high notes. And when we developed our AI, in particular the advanced speech recognition and all the backend AI cognition to extract content, we found out very quickly it wasn’t just about having the best AI or the best AI engineers, but you need the best CX professionals across all different disciplines to deliver the CX services along with that technology. And one thing we do, that’s a little different is bundle both the technology and the service together in a subscription model and it’s all based on usage. And so we have this saying that we call, life less hard, and because we’re trying, everything we do is in the obsession of how we can make life less hard for our clients, for their customers, even for our employees to automate more and how we can make life less hard from a self-service perspective.
And so that’s really been the driving force behind our technology and business model, where we come in as a partner instead of just a technology provider and just selling licenses. And so the best way to check us out is by reading our reviews on Gartner, as I mentioned earlier, and under the virtual customer assistant category. Last time I checked, we were still number one, didn’t have the most ratings, but number two, I think, but definitely had a better rating than number one. More is not, not better. We had better ratings. And as far as the next steps are concerned, most starts by engaging us for a demo or a free AI assessment. These things are pretty low tolerance, low pain, short duration. And if you’re wondering what an AI readiness assessment is all about, it’s just sitting down with you to discover which call types or chats are in your contact center and perfect candidates for AI automation, and then build out a business case and an ROI model. And obviously there bunch of ways to get ahold of us. The easiest one probably is to just email us at firstname.lastname@example.org.
Brian Cantor: Now Tom, as I mentioned, we did have a ton of questions and any that we don’t get to in the next few minutes, I know that you and your team will be thrilled to respond to. They’re all great questions, but there were two that I don’t think we can end without asking, because these really do speak to, I think, a huge factor for the audience. The first is really that I spoke earlier about how a lot of people may not have even dipped their toes in the water of automation, but many companies have as well. And they do have existing self-service automation in place, whether it’s in their IVR or certain chatbot. And so this person asked, “If we already have some automation in place right now, and we switched to SmartAction, what would happen to those existing rules we’ve built, do we lose the existing automation?”
Tom Lewis: I think the simple answer is no, you don’t need to get rid of that. In fact, I often say, if you’re happy with your IVR as it is today, then keep it. But if you still have live agents, then that’s the conversation we should have because the live agents tasks can be automated. And so from as a philosophy, we sort of don’t have sharp elbows when it comes to what other technology you’re using, because our belief is that as a call center leader, contact center leader, you have a toolbox at your disposal of the various tools you need to get the job done and you need to pick the right tool for the right job. And there’s probably a reason why agents are handling so many of the calls today. And some of those very simple ones, you’ve already automated, fine, keep them that way. And then what actually ends up happening is we start doing these complex ones, clients see good results. And then they’re trying to unify that customer experience from a voice or a messaging perspective. And they may go back and revisit that original assumption, but the ROI of even just starting on the next thing that you haven’t automated usually is compelling enough to support that self-sustaining business case that I was talking about earlier.
Brian Cantor: Yeah. And you mentioned unification, that actually ties very nicely into this other great question from our audience. And that is, do you find that your customers typically implement voice and chat at the same time when it comes to your conversational AI? Or do they kind of go one channel before the other?
Tom Lewis: Well, I’d give you the standard consultant’s answer to that one, which is, it depends. And usually what it depends on is where’s the biggest ROI coming from. And today, it may not be true tomorrow, but today for the most part, it’s still coming from voice. And so clients typically start on the voice channel, but there’s one other consideration that we’ve run into occasionally is that they want to do chat or chat bot, AI chat, what have you, whether it’s through SMS or Facebook messenger and sometimes they don’t actually have live agents doing live agent chat as well. So our philosophy is very clear that if you got the voice channel and your backup is live agents, that if you do other digital channels, your backups still should be live agents otherwise it doesn’t contribute to a very good customer experience. So it could start with either generally speaking, we prioritize it based on where we’re going to see the highest ROI for our clients.
Brian Cantor: And you gave me an opportunity to squeeze one extra question in there, Tom. I know that we’re a few minutes over, but you mentioned ROI there. So when you compare, when you look at how your clients are comparing things like customer satisfaction of their virtual agents versus their live agents, what are they seeing?
Tom Lewis: Yeah, I think it’s really pretty consistent that when our clients do their own customer satisfaction surveys and then sort of report back to us, what they tell us is that where AI automation has completed the call in its entirety, the satisfaction and sort of NPS and other metrics are higher with the automation than for that exact same call type with a live agent. Now, if the call only partially completed through automation and then gets transferred to a human, the results are somewhat mixed. And what we’ve concluded is it really depends largely on how the live agent transitions from automation to that conversation. If they forced the caller to repeat things or don’t pick up where the automation left off, then that tends to have a negative impact. But if they do that right, then even just making that more efficient in the upfront authentication or more effective routing, what have you, tends to drive to a higher satisfaction as well. You know, it’s a human machine interface combination here that has the right combination of things, but definitely in today’s day and age, people would just like to do it the easiest way possible and very often that’s through automation if it’s capable.
Brian Cantor: Yeah. Well, Tom, I want to thank you very much for walking through everything today, as well as all of our attendees for their great questions, for their interests, because this is, again, going to be so important to the success of everyone’s contact center when we can really maximize what’s going on in the world of AI. And I think with your last answer, Tom, you really hit on what was so valuable here is that is, thanks to conversational AI, it’s no longer a case of, okay, we probably should automate a few things, but we can only automate those things and that’s it. We’re really at a point now where we can start to look for what’s most valuable for the customer, what’s most valuable for the overall experience. If it makes sense to be handled through automation, if it makes sense to be handled through virtual means, then we should be doing it.
Not because our agents are just bored with that task, or because it feels like it may save us a few bucks, but instead, because it makes the … Technology is at the point where it can lead to a better experience for the customer and therefore you also decide, hey, where there’s going to be a better experience for the live agent, that’s where you keep them in place as well. So start to chase value, not just acceptability, because an acceptable experience is not good enough. The other point is really thinking about how it integrates with the overall people of your organization. So between your recommendation to have a strong CX team in place, as well as thinking about things like handoffs and escalations and making sure that if it starts with AI and moves to a live agent, that they’re very much in sync and that they’re delivering a continuous and consistent and seamless experience, that’s going to be so important as well.
So first of all, it’s about defining value. Second of all, it’s about really defining that workflow in that mechanism. And when you think about those two things in tandem use the best practices Tom shared today, use the great technology that’s out there. You’re going to have a great experience. You’re going to see efficiency. You’re going to see better customer satisfaction. You’re going to really get to that point where the idea of competing on the customer experience is not some empty cliché that you’re chasing, it’s a reality for your business and something everyone’s proud of. Your agents are happy. Your customer’s happy. Your business’s happy.
On behalf of CCW Digital, this has been Brian Cantor, I’ve been thrilled to join you here. And again, thank you so much to SmartAction for having me and for putting together this great discussion.