Definitely, Maybe Agile
Definitely, Maybe Agile
How AI is Transforming Banking Customer Service with Rick Delisi
In this episode, Peter Maddison and Dave Sharrock welcome Rick Delisi, Lead Research Analyst at Glia and co-author of "The Effortless Experience" and "Digital Customer Service," to discuss how AI is transforming customer service in banking and credit unions.
Rick reveals why the future of contact centers isn't about eliminating human interaction; it's about automating the routine so humans can focus on building real relationships. Learn how banks are breaking the age-old trade-off between efficiency and customer experience, and why starting with internal-facing AI tools is the safest path to transformation.
Discover the surprising truth about which customer satisfaction metric actually predicts loyalty (hint: it's not what most companies are measuring), and why customer expectations for AI are shaped more by bad experiences with other companies than by anything your organization does.
THREE KEY TAKEAWAYS:
1. AI for Everyone, Not Just Customers: AI can transform your entire organization, from helping frontline agents with real-time guidance, to giving managers instant analysis capabilities, to enabling executives to make data-driven strategic decisions. The most successful implementations use AI across all levels: customers, agents, managers, and executives.
2. Start Internal, Then Scale Outward: Begin with internal tools that help agents, managers, and executives first. This builds confidence, allows teams to experience the technology firsthand, and creates incremental improvements that build organizational trust. By the time you roll out customer-facing AI, your entire team understands and trusts the system.
3. The best predictor of customer loyalty isn't satisfaction scores or Net Promoter Score, it's the Customer Effort Score. Ask customers, "How much effort was required for you to get what you needed?" after each interaction. Low-effort experiences drive loyalty, and this metric gives you actionable insights into where to improve your service processes.
CONTACT US:
Email: feedback@definitelymaybeagile.com
Definitely Maybe Agile explores the complexities of adopting new ways of working at scale, covering digital transformation, agile practices, and DevOps in enterprise environments.
#AI #CustomerService #Banking #DigitalTransformation #ContactCenter #CreditUnions #CustomerExperience #Glia #FinancialServices #AgileTransformation
Peter Maddison: 0:04
Welcome to Definitely Maybe Agile, the podcast where Peter Maddison and Dave Sharrock discuss the complexities of adopting new ways of working at scale. Hello and welcome again. Hi Dave, how are you doing today?
Dave Sharrock: 0:16
Peter, doing very well. And I'm visiting your fair city as well, so I've been enjoying the fact that it's not as cold as I thought it was going to be when I was checking the weather apps.
Peter Maddison: 0:30
Yeah, it's been beautiful for this time of year. And today we're joined by Rick Delisi. Did I get your name right?
Rick Delisi: 0:41
Yeah, thanks Peter and Dave. Really fun to talk to you guys. My name is Rick Delisi, and I'm the lead research analyst for Glia. Glia is the number one platform for intelligent banking interactions, and we specialize in creating excellent interactions between banks and credit unions and their customers and members.
Peter Maddison: 1:01
Fantastic. I think there are a lot of interactions with banks that I wish would go a little better, so yeah, I'm looking forward to exploring that. That's why we exist in the first place. So tell me a little bit more about it - what makes Glia stand out from the other contenders in the space?
Rick Delisi: 1:20
Well, first of all, from a branding perspective, because we work only with banks and credit unions, everything we do is entirely specialized and customized to that industry. There are plenty of other providers offering CCAS services and various AI functionality that work horizontally across a bunch of different industries. But I'm sure as you guys have been discussing, AI works so much better when it's more vertical - when it's designed to help customers who have the specific problems of people in a specific industry. And I think it's fair to say that whether you're talking about a 10,000-member credit union or some mega bank like Wells Fargo or Bank of America, the issues and reasons that customers contact a bank or credit union are pretty much the same, regardless of the size of the company itself. So we specialize in creating exactly the right interaction for each customer every time they contact the bank. And as we get into our discussion, what we'll see is those reasons vary quite a bit - not from bank to bank, but from customer to customer. And there's now a way to identify what that person's issue is and match them with the exact right type of interaction.
Dave Sharrock: 2:38
And maybe, Rick, if we explore that a little bit - you're looking at presumably consumers, but also businesses contacting the credit unions and banks, right? So very different needs when they're reaching out. But there's some way of segmenting or tailoring the experience based on either demographic or the experience they've had with the bank, the years of interactions that you've got tracking through that as well, I'm assuming?
Rick Delisi: 3:07
Yes. I'd like to clarify though that while it always seems like B2B and B2C are completely separate worlds, the reality is even in a B2B contact with a bank or credit union, it's still a person - still an individual person. And that person's expectations and their reaction to the interaction they're having with a bank or credit union are shaped not just from their dealings with that company, but from their experiences with all companies they've ever dealt with. So the expectations are more similar than different, and people's reactions are more similar than different, even though the nature of the interaction might seem different if I'm representing a business versus just myself or my family.
Peter Maddison: 3:51
I guess some of it would depend on how much they care about the organization's money.
Rick Delisi: 3:56
We could say that in some ways they're similar because, you know, if I'm representing my family, I'm concerned about our whole financial well-being. If I'm representing an organization, it's all of us. So there's always more than just me here.
Dave Sharrock: 4:12
Now, I think this has a lot to do with contact centers, call center interactions. And clearly there's a lot happening in that space. The reason I'm kind of cautious around this is AI is promising many different things in that space, but AI is such a nebulous term. There's a little bit more to it than that. And there are experiences where people are talking about effectively eliminating the need for human beings - or at least that personal interaction that we might actually be looking for - versus the ones where there's a lot of data and alignment of exactly what the messaging is as you come into that conversation. Can you talk a little bit about how AI is shaking that up?
Rick Delisi: 5:04
We couldn't disagree more with this strategy of "you'll never need to speak to a person ever again." That just doesn't turn out to be true, certainly in financial services. What we've done is an analysis of the various reasons why a person would contact a bank or credit union. And what's really interesting is that for almost every organization, if you analyze just the top 20 issue types or needs that a person would have, those 20 comprise 80 to 90% of all interactions. So if you were to sit in the contact center of a bank or credit union, it's pretty much the same questions and same issues over and over again. Now let's break it down based on AI's role. What we've learned is that about half of the issue types or needs a person would have fall into a category that can be absolutely automated to the great satisfaction of the customer - simple answers to routine questions. What's my balance? Did this check clear? What's your routing number? The kinds of things that come up time after time. Again, if you're sitting in a call center, you'll hear those questions dozens and dozens of times a day. The other 50%, though, are issue types that are more complex or more emotional, and those are way better handled by a human being. But what Glia has developed is a platform that identifies this customer's issue - the reason they're contacting us right now - and automates the process of matching the simpler issues to automated service or virtual assistants, but seamlessly transfers that person to a human being in situations where that's better. So again, companies that are thinking "we should automate everything" - that's wrong. But "we should automate nothing" is equally wrong.
Peter Maddison: 6:55
Yeah, I kind of think of it - because we've had the ability to find routing numbers and other pieces of information fairly easily with various different mechanisms...
Rick Delisi: 7:06
I'll just interject here, Peter. You'd be amazed at how many phone calls there are for these simple issues that people could have easily figured out themselves.
Peter Maddison: 7:14
Yeah. And being routed through some sort of IVR system that's going to direct you down a particular channel into whichever automated piece gives you the information it thinks you want - which invariably isn't what you need - and you spend time screaming at the phone going, "give me a person." I wonder though, as we've had mechanisms that could expand on that to help maybe better navigate that interaction, have you seen opportunities where AI can start to solve the simpler, if not the most complex, issues from customers?
Rick Delisi: 7:56
That's exactly what we're seeing. Again, on average about 50% of all the phone calls and live chat sessions that a bank or credit union gets could have been easily resolved without any human assistance at all. But the key here is that the AI has to know what it can solve and what requires a human being, and then immediately transition that person to the right form of interaction for the thing that they need. If you think about it this way - just about every organization has some kind of marketing strategy around personalization. We have to know our customers, we have to know their needs, we have to know their history. That makes total sense in the marketing world. But when it comes to service and support, the one thing that really matters is: what is this person's need at this moment that they chose to contact us? And so personalizing to the person actually makes less sense in service and support than personalizing to the need this person has right now. The same exact customer might contact a bank or credit union two days in a row, have completely different needs or issue types, and would be well served by two very different interaction types.
Dave Sharrock: 9:12
Yeah, I'm just thinking there are some really interesting areas to explore here. The customer service support space has in many ways been one of the very top priorities in terms of application of AI and agents around that, and models that, like you say, start shifting as many calls as makes sense into a lower cost format that we can navigate through, hopefully as close to natural language as possible. But there are many other processes a business will have within their organization at lots of different levels. You mentioned marketing is one area that has plenty of processes - sales, HR support, things like this. What would you say in putting together what Glia is offering - what really stands out in terms of looking at business processes with an eye to improving and using these new emerging technologies? How do you get those into an organization without sort of tripping over and forcing everybody to talk to an automated agent? What are you seeing? What are the takeaways? What would you do if a new credit union or bank came to you and said, "Let's have a look at our support process"?
Rick Delisi: 10:42
One of the things that's really interesting for anybody who's studied this space - and I've been studying customer service and customer experience for, dear God, a quarter century now - there has historically been this conflict, this tension, this trade-off between efficiency and experience. In other words, you can drive a significant amount of cost out of your customer service operation if you're willing to sacrifice the quality of those interactions. And vice versa, you can create a five-star Ritz Carlton kind of experience if you're willing to spend to that level. And for virtually every executive who works in the space, whether they've ever thought about this or not, every decision you end up making is some kind of compromise between those two, some sort of trade-off. "We can only have so much of this and so much of that. Let's come up with a compromise that we can live with." And what we're seeing now in the AI era, as organizations begin to utilize AI at all levels of their organization - which we could certainly expand on - what they're seeing is they can create efficiencies far greater than any they'd ever been able to achieve before AND improve the quality of the customer experience. This is the ultimate in being able to break that tension or that trade-off. It really hasn't been possible before. It is now. And when an organization realizes "we can be saving significant budget on what we used to spend for customer contact AND create a better loyalty-building experience for customers," that becomes a no-brainer.
Dave Sharrock: 12:23
And maybe as I'm trying to explore that - I think that the opportunity is very seductive, exactly to your point around that tension. And I'm trying to imagine if I'm a manager trying to bring something into an area that I'm responsible for - are there specifics that you would say look for a particular type of return? Are there problems that are really the obvious low-hanging fruit that we'll often talk about? Categories of problems that really stand out as being "go after these first"? Are there categories of problems equally where you say, "Look, get some experience first, don't go trying to tackle that"? And part of why I'm asking is that everybody in this field has heard about or read the MIT study where they're arguing that 95% of enterprise spend really hasn't delivered on its ROI. And there's this element of concern that comes with that, because I think Peter and I have alluded to that in some of our conversations. We work with organizations - they go through organizational transformations, cultural shifts all the time. We know that there are so many things that can go wrong right at the beginning that means it doesn't matter how well the transformation might go, the actual return on investment becomes incredibly difficult to really get. So are there areas that you say "go after this space for ROI" or "don't go after this, at least in the early days"?
Rick Delisi: 13:58
Yeah, I mean, one of the great things about operating in a contact center environment is that everything is measured. There's no question, there's no subjective judgment. Every second of every interaction, every dollar spent is all immediately accountable. So it's not like something like marketing where you invest a million dollars in a campaign and hope it returns 7x six months down the road. So one thing that's really important to note is - here we are in 2025, and still to this day, the vast majority of organizations are taking way too many unnecessary phone calls. Phone calls in customer service are, of course, the most expensive types of interactions. And even though most of us have conformed our behaviors to becoming more digital-first and we live on our screens, and our first reaction whenever there's some kind of issue or problem is to go right to our screen - whether it's a mobile device or a laptop or desktop - there are still way too many phone calls happening. And those are phone calls that don't add any value to solving a customer's problem. I was recently talking to a woman who's head of customer experience for a bank in Connecticut, and she said something so interesting. She said, "There's no inherent value in hearing your balance from a human being versus being able to see it on your screen." And so one gigantic opportunity for every organization is to eliminate the live phone calls that add no value either to the company or to that customer. But what we're also seeing is that customer-facing AI - which is one of four different flavors of AI, and I'll elaborate here - is seemingly the riskiest. Many companies are very hesitant, and banks and credit unions are both risk-averse and heavily regulated, so certainly very concerned about any information that might ever go to a customer that isn't already vetted, or hallucinations or bias, or all these obvious potential flaws of AI. So what we're recommending to companies - if you're not ready to go all in on AI, start with the applications that are only internally facing. And there are three flavors: agent-facing AI, manager-facing, and executive-facing. And each one of them brings extraordinary benefits in terms of efficiency.
Peter Maddison: 16:31
Yeah, that makes a lot of sense. I can see that - getting people to get familiarity with it helps build trust, helps build confidence that these systems are going to work the way they're expecting them to. There is still, of course, the problem with AI that it is non-deterministic. It will potentially produce absolute garbage from time to time or hallucinate, especially if, for example, a chat session runs on for long enough and you start to even get partway through the context window, you can start to see a great degradation in terms of the responses you're getting from it. When you look at things like that, how have you been looking to tackle those types of problems in your ecosystem?
Rick Delisi: 17:17
So you guys have probably heard this term "responsible AI." It seems to be a phrase that almost every organization is using, but with variable interpretation. And at Glia, the way we define it is: nothing will ever be seen, shown, or spoken to any customer that hasn't already been approved by you. So there is no potential for hallucinations. There is no potential for misinformation. If at any point the AI is in the middle of a conversation or interaction and it doesn't have the answer, it doesn't know what to do next, you're seamlessly transitioned to a live human being who can pick up the conversation in context. There's no starting over, there's no re-explaining your issue. If a human gets involved in the interaction, they already see everything that's already happened in the interaction with the virtual assistant and they pick it up from there. That is truly meeting your customer where they are - not just in whatever channel they chose to start the interaction, but in the context of the interaction that already started a couple minutes before the live person ever got involved.
Dave Sharrock: 18:27
So it sounds to me, Rick, like there's a sort of natural and elegant handoff into that live person, so that the conversation is continuing rather than the classic "hold on, we're going to transfer you to another department," which is almost a meme in some cases.
Rick Delisi: 18:47
But that's a high-effort experience when you have to start all over again and throw your hands up in exasperation. We've all experienced it, and it's terrible. It's just a degrading experience that makes you feel disrespected as a customer. But picture the difference between "Hi, I'm Rick, how may I help you?" versus "Hi, Dave and Peter. It looks like you guys are trying to apply for an auto loan. I can totally help you with that. And in fact, I've already started the process." So it's still the same two people - one agent or rep on the company side, one customer or member on their side. But even though they're resolving the same issue, it's a completely different feel and totally different experience when it's all handled in context.
Dave Sharrock: 19:33
So Peter and I spend a lot of time talking about organizational change and the people involved and that sort of cultural shift. What you're describing opens up the opportunity that now, as a call center agent, I'm not going to have to deal with the run-of-the-mill routine, maybe soul-destroying "it's the buttons on the right-hand side" type of conversation where people are asking for things which are very easy for us to navigate and find, or at least for an agent to walk through. Does that mean that there's higher engagement or more opportunity for those in-person conversations to actually lead to - like you said - loyalty, customer satisfaction? Do you see that being something - you mentioned earlier about everything being measured - so there's both the customer satisfaction, but presumably also agents that have a more rewarding experience because they're not dealing with the mundane tasks nearly as frequently as they may have been in the past?
Rick Delisi: 20:44
That is exactly what we're seeing. The experience of being a frontline rep - let's be honest - can be an absolute grind. Eight hours a day, call after call or chat after chat, many customers have some kind of problem or issue, so they're already a little emotionally amped up before the conversation even begins. You know, "today's a great day because I get to call customer service," said no one ever in the history of mankind. But when that job of a frontline agent can be made easier, ironically, the AI is enabling humans to be more human instead of being human robots. Instead of having to start every conversation from scratch, if you already know who this customer is because they've already been authenticated, and you already have a strong idea of what their issue type is, you can really create something that feels more like a VIP experience. "We already know who you are. We already know what your issue is. We're here to help. We're working on it right now. We're already a step ahead of you. We're on your side. We're advocating for you." It's not "you, the customer" versus "me, the company." It's us together as a team working to resolve the issue. And with agent-facing AI that provides all this information upfront before the agent even says hello, and then even on their screen provides tips and suggestions for how to resolve this specific issue type, as well as potentially even related issues that we might want to discuss while we're still having this conversation - it sets frontline people up for far greater success and it makes their job that much more rewarding because they can just concentrate on being a person.
Peter Maddison: 22:30
Does it also change the makeup of call centers and the type of people that you have there, the skills that they would have? And make it similar to what we're seeing in software development, where the more experienced people are the ones who get the most benefit out of the tools, the knowledge that's being presented to them? But then how do you bring in the more junior people and get them up to the level they need to be at?
Rick Delisi: 22:57
Are you guys familiar with the basic concept of contact centers - tier one and tier two? Tier one is entry-level people, tier two are more like experts, or sometimes a call or contact would be escalated to a tier two agent. Well, what if every agent was a tier two agent? Because all the tier one stuff is all automated. What if even a person who's been in their seat for just a couple of weeks has the benefit of being able to work with the notes and bullet points from the best agents who've been handling that same issue for years, making it that much easier for even a relatively inexperienced person to come across as a true expert because all the information they need about this person, their issue, and how to resolve it is all presented to them automatically?
Peter Maddison: 23:46
Yeah, it's upskilling the entry-level tier one people to tier two faster, getting them through that by providing the information. There is a little bit of "you can't teach experience" that comes into that too.
Rick Delisi: 24:02
But that may be the case, but what we're seeing here is that in a unified channelless platform where all interactions are being stored in a central repository, all information about every interaction is available to anyone at any time. And so the most experienced people are - just through their daily interactions - teaching the system how to teach everyone else how to act in the same way that they do. And also providing that confidence that you're not going to be in a panicky situation as a new person, immediately thinking, "How do I get out of this interaction and how do I escalate it to somebody else?" Because the very same information that a more seasoned person might know in their head is now available for me on my screen. And so the AI helps your people get smarter, your people help the AI get smarter, and what you see very quickly is this virtuous cycle of continuous improvement.
Dave Sharrock: 25:05
It's a really interesting area that I'm just trying to explore deeper on, because in theory it's possible that this kind of just keeps cycling round and round, where those tier two experiences train the AI agents to be able to cover more and more of the tier two work, so there's less and less need for tier two. But I wonder also - you talked about call centers being really heavily instrumented so you can really see what's going on. Fundamentally, this is about frustrated people calling a service provider, a bank, or a credit union to say "something that I thought would be quick and easy and I could come online and do is not quick and easy, at least with the experience that I have right now." How are the experiences - the customer satisfaction numbers, time to resolution, whatever the data might be - how is that coming through? Because ultimately, we would expect it to get shorter and shorter and quicker and quicker to get a resolution if I'm calling my credit union or bank. So how do you see that coming out? Do you see shorter resolution times, higher customer satisfaction, fewer repeat calls, whatever the metrics might be?
Rick Delisi: 26:29
Yeah, those are all valid metrics. And what we're seeing is substantial improvement in all of them. I know this sounds almost too good to be true, but that's what we are genuinely seeing. In all categories of metrics - efficiency and experience - there are improvements when you're able to automate the simple issues and elevate the performance of your frontline people who are handling the more complex and emotional issues. It's one of these things where every category of metrics ends up improving because there's no longer this trade-off.
Peter Maddison: 27:07
AI is obviously well known to everybody now. Why is it that not every single bank and credit union that you have access to has this in place already?
Rick Delisi: 27:17
What we're seeing is that the greatest hesitancy comes from a couple of reasons. One is just the general hesitancy to use new technology in a heavily regulated and risk-averse industry. So we certainly understand that there might still be - as we're rounding into 2026 - some lingering hesitancy. But there's also this concern that because the reputation of AI is "you'll never speak to a human again" or "it eliminates people from the equation," organizations that are community-based or relationship-oriented will lose those relationships or lose the opportunity to build a deeper relationship with people. And what we're seeing is not every interaction is a relationship-building moment. Checking my balance, simple transfers, "What's your routing number?" - those aren't relationship issues. But "help me with my first ever mortgage," or "my credit card got stolen and now I'm concerned about fraud," or "there's a death in the family and we have to deal with a probate issue and deal with our relatives' finances" - those kinds of issues should absolutely be handled by a human being who's in the business of providing a relationship and being on that customer's side. But being able to instantly determine at the moment of the beginning of an interaction "is this an automate moment or is this an elevate moment?" allows those situations in which a relationship matters to become that much richer and to create that much more future loyalty.
Peter Maddison: 28:55
Yeah, it makes sense. I mean, I completely agree with that. I would never call into my bank to ask what my balance is unless they for some reason had lost the ability to provide it to me in other ways.
Rick Delisi: 29:07
But again, spend a couple of hours in a bank or credit union contact center, and you'd be amazed how many phone calls are for these routine, everyday issues.
Peter Maddison: 29:19
If we can get leaders to buy in to this, and it's going to build those better relationships with customers - I mean, is it sunshine and roses all around? Are there no negatives to this? Should we just move forward and get AI into all of our call centers?
Rick Delisi: 29:36
Yeah, I mean, there are potential negatives, there are risks and concerns, but those have now all been eliminated from the system that we're operating. Because, again, this idea of responsible AI - nothing ever gets sent or seen or spoken that isn't already approved. And we're relying on the experience of every other bank and credit union to understand what the issues are that customers have. There's a tiny fraction of issues that a person would ever have that are one of a kind or out of the blue or "we've never dealt with this before." It's the same stuff over and over and over again. And so once organizations realize "this is already working really well at lots of other peer companies," it's way safer and in many ways even more rewarding to start with internal AI and then expand outward to what you can do that's customer-facing. It provides a much more gentle path to achieving the ultimate success of way greater efficiency and improved experience.
Peter Maddison: 30:46
I think that makes an awful lot of sense. So Rick, you mentioned before we started the call that you have a couple of books as well. Would you like to tell us a little bit more about them?
Rick Delisi: 30:58
Sure. I'm the co-author of two books. The most recent one is called "Digital Customer Service," which I wrote with Glia's CEO and co-founder Dan Michaeli. And it's all about how to create five-star experiences in the on-screen, digital-first world that we now live in. We wrote it just as the pandemic was coming to an end, when people's digital behaviors were really solidifying. A book I was part of writing 10 years prior during my time at Gartner is called "The Effortless Experience." And that's where we learned that out of all the questions you could ever ask a customer right after an interaction, the one that best predicts that person's future loyalty as a result of that interaction is not the standard customer satisfaction question or even the net promoter score question, which is "how likely are you to recommend this to a friend or colleague?" But rather - think about this just from your own perspective - when you ask a person right after an interaction, "How much effort was required for you to get what you needed from that interaction?" Their answer to that question correlates at the highest possible level with that person's future loyalty behavior. And so what's the point of measuring the quality of an interaction? Not just "how did we do?" - that seems like a very self-serving thing - but rather what's happening as a result of the interactions that we're doing with customers that will influence whether or not they're going to continue to be loyal to us in the future? And learning from those low-loyalty or what we call high-effort interactions, to be able to minimize the amount of effort that customers experience when they're interacting with you, particularly when they're trying to solve a problem or resolve an issue.
Dave Sharrock: 32:51
I really like what you're describing there. I know Peter, we just asked this question at the end instead of the beginning, because there's so much more we could talk about, right?
Peter Maddison: 33:01
Yes, I was going to say - if I'm going to wrap this up with three points, I'm going to pick that one.
Dave Sharrock: 33:08
But maybe before you do the wrap-up, Rick, I wanted to ask about the second book, the one you just wrote after or as we were all coming out of the pandemic. Because I think the digital customer experience expectations have really - there's sort of a significant change in those over the last few years. Not purely because of the pandemic, but it certainly generated a very different expectation or experience. Can you maybe summarize two or three things from that? Like what did you identify as part of writing that book that stand out as being significantly different to prior to the pandemic?
Rick Delisi: 33:55
I'm not sure if we coined this term, but we certainly relied on it heavily, and it's a concept called the "A to Z phenomenon." And that is when it comes to customer expectations for a digital interaction with your company, their expectations don't come just from their interactions with you - they come from their interactions with every company they've ever done business with, from A to Z, from Amazon to Zappos. And so if your organization - if you're a regional bank or a community-oriented credit union - if you're not providing the same kind of digital interactions as people are now used to dealing with digitally native companies like Amazon, you've already fallen behind. You already aren't meeting expectations. But I'll tell you something really interesting. We're seeing now almost exactly the opposite when it comes to interactions with virtual assistants or AI, in that a customer's expectations about an AI interaction with you don't come from just their interactions with your organization, but from every other horrible, crappy AI interaction they've ever had with every other company whose bots don't understand what you need or give you the wrong answer or can't immediately transition you to a live human. So we've seen the whole phenomenon work in reverse in the AI era.
Peter Maddison: 35:18
Yes. I know what you mean. Sometimes it's like, "Could you just give me a button now?"
Rick Delisi: 35:25
And where does that come from? That comes from the fact that dealing with some other company - or in your memory are interactions where the bot didn't understand you or gave you the wrong answer or thought it knew what you wanted and gave you the answer to something that's completely unrelated. We've all had those kinds of experiences, but those don't have to exist anymore. And certainly not when the bot is purpose-built for the exact industry that company is a part of.
Peter Maddison: 35:53
Yeah, and personally I find the ones most irritating - the ones where it wants you to... "What question do you have for me?" And it is absolutely, invariably, completely incapable of getting it anywhere right. Because by the time I'm calling a call center, it's usually a pretty technical problem, and it can't help.
Rick Delisi: 36:11
Right. You call an airline and it says, "Where would you like to fly today?" And you say, "I'd like to fly to Newark," and it says, "I think you said Tegucigalpa, Honduras. Is that correct?" No, it's not! That doesn't help.
Dave Sharrock: 36:23
I was going to say, I think we can all list a few experiences that we've had around that.
Rick Delisi: 36:28
Yeah. But again, that negative baggage is carried by that customer into their interactions with you. But what's interesting is that in general, I think it's fair to say we're still at a point where if you asked a hundred consumers, "Do you like interacting with bots? Do you trust AI?" you'd get a very high negative reaction. But if you then, as a follow-up question, ask those same people, "Do you like instant answers to simple questions?" you'd get a very high positive. So it's almost like there's this contradiction in the zeitgeist. "I want what AI gives me, but I don't think I like AI." Well, that's the last generation of bots and AI, which have now been supplanted by much more intelligent systems. And again, when it's targeted specifically to that company's industry, the understanding rate is extraordinarily high.
Peter Maddison: 37:25
Yeah, context is king when it comes to these AI tools, for sure. We've discussed that a little bit on some of our previous podcasts - the more you can narrow the focus and the more context you can provide, the better the responses you'll get.
Rick Delisi: 37:41
So for technology companies that are offering bots and virtual assistance horizontally across a wide variety of industries, good luck. That's really hard. Obviously, there are all kinds of challenges in providing a vertical solution, but again, it's within the context of "it's the same issues over and over again."
Peter Maddison: 38:03
Yes. And there's this almost underlying premise that "well, everything's the same across all of these and there isn't any difference," but there is. There are differences in languages, taxonomy, and the way that the types of problems you need to solve, the back-end systems that need to be interacted with - which differ very much across industries - even though at a very surface level, it's kind of the same thing. So with that, that feels like a good point to wrap this up for today. And as we've discussed, we normally wrap this up with three points, one from each of us. And I'll suggest Rick, you can go first as our guest. What point would you like our listeners to take away from this conversation today?
Rick Delisi: 38:46
Yeah, I hope people will broaden their perspective about AI and customer interactions and think not just about AI that serves customers or automates simple issues, but think about AI that helps your frontline people to do their job more effectively and in a more fun and engaging way. Tools that are available for supervisors to be able to analyze any interaction, all interactions, interactions about a certain type, interactions from a certain team or people who work for a certain supervisor versus another. The range of analysis that you can do just at your own keyboard or just by even verbally asking simple questions is completely different from what used to have to happen when if you wanted to do some kind of data analysis, you'd have to assign it to a team and submit a ticket and wait a few weeks. It's all instantly available to you. And AI for executives allows for strategic decisions to be made at a far more data-driven level, for you to be absolutely sure if you're going to attempt to do something different or do something new, it's based on an analysis of all the interactions that you've done previously. And so this idea of "AI for all" - customers, agents, managers, and executives - is a far broader view than I think most people have in terms of how AI can help in customer interactions beyond just bots that respond to customers.
Dave Sharrock: 40:19
Yep, I like that. I'm going to touch on what you were talking about, Rick, when you were describing how to go about enabling a particular business workflow. You talked about using your internal systems - so whether it's executives, managers - but providing that internally first so that you can learn, but also everybody gains an experience of what it means to interact with these agents and tools. So I thought that two things came to mind. First of all, it relates with everything that Peter and I talk about - incrementally small modifications and changes so that you can learn from them in the next round, and the next round gets better and better. But secondly, also the different audiences, because I think bringing the executive along, bringing the managers along, bringing the frontline staff along as they begin to experience it from their own perspective, just gives a lot of confidence across the depth of the organization. So that's a takeaway for any sort of AI introduction that might be planned. Have we left anything for you to pick up, Peter?
Peter Maddison: 41:31
Well, I was going to go to the one I said I was going to go to, which is the concept of asking "how much effort was involved in this interaction?" as a way of correlating to whether somebody would be a customer who would come back - the retention piece. I think that's a very interesting question. I've heard similar before, but when we're talking about it here, it brought to mind like a coaching conversation and at the end of a coaching conversation, we would ask "How was this conversation of value to you?" And there's a correlation piece there. It's that retrospective of having somebody think about what the interaction was and what was involved in that interaction, and then relating it. At the same time, in this particular instance within the customer service perspective, it's giving you insights into where should I focus, where should I improve, what are the things that I need to do next to improve how customer service interactions occur. So I did like that. It was good. So with that, I'd like to thank you, Dave, and thank you, Rick, for joining us today. And I look forward to next time. We'll wrap it up here. And if anyone would like to reach out, they can at feedback@definitelymaybeagile.com. And I look forward to next time.
Dave Sharrock: 42:48
Rick, again, thanks for the conversation. Peter, always a pleasure.
Rick Delisi: 42:51
Thanks to you guys, this was fun.
Peter Maddison: 42:54
You've been listening to Definitely Maybe Agile, the podcast where your hosts Peter Maddison and Dave Sharrock focus on the art and science of digital, agile, and DevOps at scale.
.png)
.pdf.png)
