Why contact deflection creates quiet loyalty erosion - Interview with Pasquale DeMaio, VP of Amazon Connect Customer

0:00So welcome to the next edition of the Punk CX podcast. With me today, well, I have Pasquale DeMaio, who's the VP of Amazon Connect at AWS. Now, Pasquale and I have been trying to have a conversation for a little while now, and we've succeeded in getting it in the diary. And so, Pasquale, it's great to meet you, great to see you, and I'm really looking forward to our chat today. How are you doing, first of all? I'm doing great, and yeah, it's great to finally get a chance to sit down and chat with you. So, you know, it's a beautiful, sunny summer day and I'm enjoying the summer weather and looking forward to having a conversation.

0:34Awesome. So for the benefit of our listeners or people that are maybe continuing and reading the highlights that I'll put together post podcast release, I wonder if you could just introduce yourself and tell me a little bit more about yourself and sort of the work that you do. Absolutely. You know, I run an organization that's Amazon Connect customer. And really, I've been working on this as a project for about 13 years or so. I wanted to start a service that would help folks. And I was interested in doing it at AWS because we were really, you know, obviously at the time, AWS was big in the cloud.

1:11And when I got to Amazon, I had thought I would do something maybe with Maps. I love Maps. They're beautiful. and then I met these folks who had been building out the technology for dealing with customer experience, customer engagement, customer service and I saw what they had built and I was blown away by what they had done and I said I want to go do this with you and I said what do you guys think about making this a public service that anybody could use not just for all the teams in Amazon who wanted to use it, deliver great customer experiences and they were really excited and so these folks who I consider really the founders of the service, some of them the first person who wrote the very first line of code is still on the team.

1:46They were crazy enough to go on this journey with me and we set out to do that. It's hard to believe, but that was back in 2017, we launched this product and we've been through a few revolutions since then and now none bigger than the AI, this new AI revolution is happening right now. It's an amazing time to be alive. And so every day I wake up and sort of think to myself, it's insane that I get to do what I do for a living. I get paid for it. And then we're all trying to change the world here. And there's a lot of folks on this journey now, but but it started just as a sort of a small group of folks a singular vision that we thought ai could do something amazing and it turned out we got very lucky today i got a lot better during the prod the progress as well so the process as well so pretty interesting awesome i mean yeah i mean as i say it's like i've been in and around this sort of service and experience space for oh crumb 17 18 years now and amazon has always been a big player in that sort of whole service experience but one is a provider like a brand and of itself doing his thing and then with the with the you know the cloud services and all the stuff that came to that and now the amazon connects you know came to the play and it's almost a bit like i'm taking a an internal capability and then making it available to the you know to to clients and so you saw me over the last few years exploded into this sort of space and been leading in all sorts of different things but and so that's great i'm really looking forward to to you know mining your perspective let's say and um and there were some things that you kind of said i mean to do that i think there's some things that you said you mentioned to me in the run-up to this kind of podcast that i wanted to to to highlight and i asked you to sort of explain that so we can explore it a little bit and one of the first things that you said that i wanted just to you know to ask you about was you said every customer now expects a concierge level treatment regardless of how much they spend and i was a bit like do they though and so i wanted you to sort of explain that to me if you could well i you know i think every customer wants it and and the folks who who have experienced it are are starting to demand it um i think folks are kind of stuck in this world where they've been And we've kind of, as an industry, some poor actors have abused folks for a while and sort of created lowered expectations.

4:00But what's happening now is that we're seeing that the product is something you sell, but the experience is what people often are buying. And you hear that, you know, the stats back it up that folks will now leave. A few years ago, they're like, well, after three bad experiences, you'll lose a customer. And now they're, you know, now one bad experience will lose a customer is what the data suggests. And I can't tell you every customer is the same because I think we all know they aren't. But the reality is that if you think about almost any good interaction you've had with a company or a brand or people, there's an experience where you feel like it's not just that you got some item that you wanted or you had some service completed.

4:38Yes, that's obviously the core reason why you probably started there. But you think to yourself, oh, I actually felt like I was understood. This was enjoyable. I felt like I was listened to. and you know there's entire businesses you know the cafes exist not because people can't make coffee right and so the reason people want to be out and experience this these interactions for the first time ever now if you think about the price of the item you know there there are objects that people sell where if there's a single customer contact good or bad they would lose money if they had to handle that contact in a meaningful way and so those folks were found themselves forced to try and avoid contacts.

5:15But now we don't have to spend a lot of money to solve a lot of these problems. We can use technology to have great understanding of the customer, provide something tailored to them, that experience that is meant for that person, designed for that person, and the experience is unique to that person. And we can do it in a much more efficient way. They don't have to wait on hold. They don't have to have, you know, someone ask them the same questions for the fourth time, you know, after they go through this process. But instead they can be greeted with something and says, you know, hey, Adrian, I see you had a problem with this item or I see you bought this item last week or I see you have a flight booked and I noticed that the time changed.

5:49Is that why you're calling? And that can change the entire experience. And, you know, I certainly will tell you that the difference between having someone who actually cares about your outcomes or a system that cares about your outcomes, the process that cares about your outcomes versus the process that just seems like it's trying to get rid of you is a very different experience fundamentally. And that's where I think the most forward-looking companies are going, I think it's where customers are asking us to take us. They don't just want to be gotten rid of, they want to be engaged and they want to have you actually solve their problems, understand their problems and solve them.

6:23Right, so it's almost like, it feels like a changing mindset around, you know, realizing that customers will have problems and you have to be open to solving those kind of problems rather than just kind of going, la, la, la, la, la, I wish they all went away, as it were. Well, and if you think about that, it's like customers are probably contacting you because something interesting has happened at best, right? I mean, if they could take care of everything they wanted, they wouldn't necessarily call you. They may have a question. It may be a very easy question. It might be a great question. It might be like, why is your product so awesome?

6:52That's a wonderful place to be in life. And customers have those real questions like, which one do I want? This one or that one? Why, how would I pick? And, or they may have, they may have a challenge where something didn't go as they expected in a way that's negative. And, and in both those cases, the, you know, we always approach the problem by saying, how would I make this person not think, hey, I got whatever I wanted done, but hey, I actually feel like I like this company better. And that's actually possible, even in cases where something wrong happened, if you go take that next step of following through, if you take the step of treating the person as if they're your friend, you do some trust, but verify because you have to make sure that it's not fraudulent and stuff like that.

7:32And you have to protect yourself and your customers for those types of things. But the reality is, it generally isn't any more expensive to deliver an amazing outcome, deliver great customer experiences. It does take more thought. But then once you have done the thinking, the outcomes are disproportionately beneficial too. And you can turn what should have been, by all means, a negative experience for a customer into a very positive one if you handle it right. And I think that, yeah, related to that, I think it's because you also related to the concierge level treatment thing. You talked about deflection is dead, and it's actually all about engagement.

8:06And it feels like actually what you're doing is you're saying, actually, when we address these queries, let's say, whether they're problems or just general kind of queries, you almost have to start with the end in mind and then work backwards kind of from there. If your end is like, I want you just to go away, then you're in that sort of like deflection containment sort of thing. and it's like, I don't want to deal with that. But if you actually start with the end of mind and actually see it as an opportunity and you use the technology and all the data and the knowledge and skills that you have at your behest, as it were, then you can actually drive better outcomes.

8:42Yeah, I couldn't have said it better. The thing that I think is really interesting is your metrics, and everyone loves metrics in this business, and in most businesses, you should understand your metrics for sure, but your metrics might start to look really good. You're seeing fewer and fewer contacts. The contacts are shorter and shorter. The customers aren't calling back. But is the reason they're not calling back because they're happy and excited or is the reason they're not calling back because they've given up on you and instead they've either chosen a competitor or they've chosen a tool like ChatGPT or something to go try and solve their problems because it's actually providing a better experience than what you're providing.

9:18And that may be great on the balance sheet for a quarter because you'll just see costs going down. Is that good on your balance sheet for a year? Maybe. Is it good for three years? I'm going to argue almost certainly it's very, very bad in two years. Your customers are now no longer your customers and you are just option A in a list in a chat window of someone else's application. That's no way to live life as a business leader. You're going to be very sad if that's where you're at. And it's kind of a quiet erosion because it's not like a customer is on the call and says, I'm never going to call you again.

9:50And they wouldn't bother to give you that feedback because they're already feeling like you're brushing them off. And so, you know, you have this and you have what isn't highly engaged customer who's taking the time to get in touch with you. If a customer is talking to you, I always, you know, I like to liken it to going to the gym. You know, the hardest part about going to the gym is putting your sneakers on. Once you've got your sneakers on, getting to the gym and working out is usually pretty easy. It's like getting over that first hump. if a customer has gotten over the hump of having the contract you wanting to engage with you they've helped for a pretty significant investment into this thing and two parts of that i think are super interesting one of them is this is an opportunity where you have someone who is most evenly invested in engaging with you take that opportunity use that um you know in the best way possible the other thing is if this is something where they've gotten to that point it was probably somewhat predictable not always but you also should be looking at these customers regularly as a set of people who you should be thinking, how do I predict what they want?

10:46How do I engage with them first? And then when they call, how do I call it reactive, proactive engagement, which is I proactively understand why I think they're calling and I offer that. And if I'm wrong, nine out of 10 times a customer is glad that you actually had some understanding of them, even if it wasn't exactly right. If you say, hey, was it that flight that was moved that you called about and say, no, I wanted to add a companion on my flight. It's not like you're going to be upset that you were understood. No, you're going to be glad you understood. Yeah. No, exactly. And going back to that, you mentioned metrics, but the other thing I wanted to ask you about is that, and you mentioned the quiet erosion, it's like, there is that sort of thing around, if you don't get this right, there is this quiet loyalty erosion that you've talked about.

11:25And actually, loyalty is the only metric that really kind of matters. and how should we actually be kind of like measuring kind of kind of that or how should we actually be how are you thinking about kind of like that because some people go like oh loyalty and they'll be like oh it's customer lifetime value or kind of so on and so forth but how are you thinking about it well i i certainly don't hate trying to understand lifetime value of a customer because i think that it's an interesting number but i think there's um i think what i like to do and encourage my customers to do is analyze every conversation whether it be with a human or with an AI interaction and actually say, don't ask, did the problem get solved?

12:07I mean, ask that. But say, did that seem like it was a good experience? You can actually ask literally, did that feel like a good experience? Did it feel like it got solved? I would also say stop measuring some things. Don't measure average handle time as a metric you care a lot about. It's one of the classic ones of like, if you want to have great average handle time, just hang up the phone every time someone calls, you'll have a tremendous handle time. And that happens sometimes. It seems facetious, but it's not as facetious as it should be, maybe. And so when I look at things, there are some legacy metrics I think are interesting.

12:40Transfers is a great one. I've yet to meet someone who's like, thank God you transferred me, or at least not in a good way.

12:48And so some of those are really good. If I look at that, a number of our customers have seen dramatic reductions in transfers. I know DoorDash saw 49 % reduction in transfers. And I think that's a good legacy metric because those are calls that you just weren't treating right to begin with. I don't look at that as a metric of transfers. I look at a metric as defect elimination ahead of the transfer, right? Sure. And so what did you solve before that person ever got there? And as a person who is offering customer software and AI and human interfaces and engagement tools to deal with when a customer contacts you, I'm always trying to help my customers have fewer contacts, but only because they're having fewer things that the person would need to contact them for a bad reason.

13:29And so everything I'm always saying is like, okay, let's move upstream. So let's understand why the person contacted you. Could you have avoided that? Let's understand why they're contacting you now. How would you make that not just a solution but a magical moment? And then if we have a magical moment, how do I solidify that? How do I follow up? How do I go get ahead of this in the future? And then how do I measure not did this customer stay with us after the call? You know, maybe you convinced them not to cancel a subscription. you know that's better than having them cancel this prediction probably although if they call back the next day now it just costs you twice as much for them to cancel so you know don't make them feel guilty i've seen that one before you know and uh but you know engage them help them see the value but they wouldn't need to see the value if you're showing casing the value to them before they got to that point and then see if they are still a customer in a month in six months these are analytics you actually have a pretty good handle on today and then individual to your business, you know, I'll often talk with customers and say, what do you want to measure in terms of how you think about your business?

14:26And usually they're pretty smart about this. They've been thinking about this problem. I mean, they're not, these people are smart, but they didn't have the capabilities they needed to deliver on it. And the people running a lot of the programs, the systems, the solutions to deliver the experiences weren't able to cross the chasm with them and have the conversation together. So they're disconnected. The obvious one is the one where you get the pop-up bad for the item you returned yesterday right i mean that's the kind of the canonical bad example but we have this all over the place where we you know where where people are having experiences where the person the left hand is clearly not talking to the right hand and getting that data into places where people can access is incredibly important and so that's a big part of what i do with folks is to say hey where is your date and the answer should be everywhere how do i get it together how do i unify it do we have the matchmert capabilities you need to be able to tell when it's the same customer?

15:16And then how do you reason on top of it? And so I've actually brought a few teams into my organization who build AI just around predictive analysis for this. And they will end up costing me money because they will drive down some of the things that people buy from us in other areas in a customer. But I'll be overjoyed with it because those customers are going to love the fact that they've improved their business. And again, we get back to loyalty. Those customers will be very excited to keep working with us because they know we're on a team to go improve the outcomes for their customers, not a team to just extract every dollar from them, which I hope is how they're thinking about their customers.

15:48And honestly, it usually is. Yeah. No, it's usually about having that. So you kind of, I just wanted to pick up on the, you talked about the, you know, everybody's talking about artificial intelligence and all sorts of different kind of flavors and agentic capabilities and, and how, and autonomous enterprises and all of these different sort of like kind of things. sometimes the role of the humans gets lost in all of that but you actually kind of like despite all the kind of the power that's inherent within all these the the this technology that's getting developed you believe that humans have superpowers too and and i wanted to ask you kind of what do you think they what are they and and and how should we be leveraging them kind of like kind of more because i just want to because sometimes they get lost in that mailsrum of kind of conversation about technology.

16:38You know, I think that's right. And as someone who loves technology, I also have a deep love of human beings. And, you know, it's an interesting thing that I've kind of built my whole life around wanting to work on technology that helps human beings. And a lot of people, I think, have come with this idea that AI should just eliminate the human beings from the equation. I think we're getting over that first hump as people realize that AI is good for some things and not good at other things. And also, you know, AI is not free. a lot of people thought it was just going to be something where you could throw an internet out of the eye at something and even if you got kind of crappy results you could do it you could kind of beat the problem of death and i and that's not a strategy that i would ask anyone to employ and and that that is that that is where humans come into this is the the judgment areas the places where emotions matter the places where you want to be understood the times when things just aren't working the way they should and and you know i'll give i'll give some great examples i'll say like If you are getting your bank balance, you probably don't want to talk to a human being.

17:36If you want to reset your password, talking to a human being is not part of making that process better. In fact, it's slowing it down and the bank balance, maybe it's making you feel nervous or shy about people having information about your personal life you don't want to have, right? And so no one sits there and says, boy, I wish a human being was there. Now, if something is not working the way it's supposed to, or you're seeing that something's showing up incorrectly and AI is sitting there telling you, oh, I can solve it for you. you see it's not getting solved or you're wrong that that thing is not what you're seeing is not what you're seeing which ai will often hallucinate and do or it just doesn't have the ability to see at all right and it does not have eyes and human things then a human being is in you know incredibly value in that moment to make a judgment call say no i don't believe that what we think is happening is happening we need to go figure this out in the middle of all this i think there's a gentle continuum where you know with with our products looking at customer you can go a hundred percent AI and you can go 100 % human.

18:33And I never recommend anyone do either of those things in almost any circumstance. Instead, I like to think about a general continual process where you have agentic AI might be great for some of these early things, particularly when someone has a complex problem they want to explain to you, you want to boil it down to something, they can just blurt that out. And AI can actually do a pretty good job understanding that. It's really understanding conversations, understanding what's happening in a conversation. And then it can say, it can do a pretty good job of saying, hey, should I try and fix this problem right now?

19:00Password reset? Yes. Person wants to talk about, you know, their wife or husband has passed and they want to talk about their estate planning. No. And so let's get that to a human being who's going to be able to really understand the challenges here and recognize that this is a very complex emotional aspect of this. And in the middle, we're going to bring in, you know, I call it deterministic AI and deterministic workflows to ensure the same things happen the same way every time when that makes more sense. I'm going to give you another example there, which is if you were going to process a payment, you can certainly have AI generate code to write payment processing procedures on the flop.

19:37It'll be expensive and it will hallucinate occasionally, screw up. It won't work the same way. It's just a silly thing to do. But there were a set of folks when we started on this journey who were very much like that's the way all of AI is going to work. Eventually, AI will just write every piece of code as it goes. and then on the other side of that um you'd say you know i want my payment processing to be exactly the same if i find a security issue there i want to change it once and have to go to every part of my company i want to use the same procedure every time and whether it's a robot an ai that's making the process of payment whether it's a human being process the payment whether it's at an atm machine or some other place process the payment i want to be the exact same one and i think people are kind of coming to that realization now that's where we started on our side because we we were looking at this problem from as like, how do I make customer experiences better?

20:24Not how do I make AI better? How do I throw AI at it? And what we found is that we handle tons and tons of different interactions with customers with human beings, not because we have we have to, but because we want to and because that's what the customer wants. And humans are still pretty awesome. I spend a lot of time with human beings every day. I love I, I come into the office every day. I really enjoy that experience. I like going to get coffee. I find human beings to be actually quite delightful, despite occasionally there's one or two I might not enjoy as much, but we should bring that to bear.

20:55Yeah, no, indeed. I mean, I think the way I like to think about it is, because you mentioned this idea that some people started from this perspective that they saw that everything, that AI would be generating code on the fly kind of like in a continuous basis. And I think about that and go, well, that feels expensive and it's proven to be. But also there's variety in that, Particularly if you're in a part of a process like you talked about, organizing a particular thing which might be sensitive or it might need to be completely accurate. And you can, there's no room for maneuver in any of that.

21:32And so I guess breaking it all down and understanding kind of like what's emotional, kind of like, you almost have to pick the right tool for the right job, right? But you have to break down a particular task that is made up of multiple different stages. And you have to almost see it and break it down into its composite parts and understand what is the best tool for the best part of that task, if that makes sense. Yeah, you make a comment there that I talk a lot about the conversation you just had with my team. I spend most of my time with my team talking about ideas of how we want to approach stuff.

22:08And then these are amazing technologists. And I say, let's you guys figure out the technology, but let's work back from what that is. And when I talk to them about that, I say, we want GenetCustomer to be a teammate to the people who are using it, not just a tool. And tools are awesome sometimes. Like, I'm a fan of a drill, believe me. I like to work my hands and stuff like that. And some really nice tools. And there was an old expression that said, you know, there was an old expression when I first started working in product management years ago. And it said, people don't want a drill, they want a hole.

22:41And I said, no, some people just want the drill. Let's not get confused about that. The drill feels good in your hand. You know, it's fancy. It's bright colored. Like people like the drill. You know, it's like they don't just want a hole sometimes. But when I think about the best outcomes I've had in my career, the times I've had the most joy, the things I've done, I was working with teammates. And teammates were making that great. And I think that is one of the things that we can use AI to do is, you know, is to make our the products we're delivering, the solutions we're delivering to customers, act like a teammate, act like something that's proactively understanding the situation with you.

23:14and then helping you make some suggestions, understanding your business with you, not taking over the business for you, but just being there for you and giving feedback and keeping track of things and knowing things in ways that are superhuman because that is something that you can only hold so many thoughts in your mind too. And there are places where AI is really good at that. And so it's sort of a teammate that's good at what it's good at and leaving you to be great at what you're great at. And so Connect Customer is fundamentally changing the way you interact with it to be an interaction that is a back and forth, one with memory, one with understanding, one that looks at what's happening and then gives you suggestions.

23:55And these interactions are actually quite powerful for people. You know, you show stuff to this and their sort of jaws drop because they're like, we're not just going in and saying, how do I change the way your end customers interact with your business? But we're saying, how do I change that with you? And one of the things about this is it's a really good lesson in learning for us because Because I think the scenarios, while different at some level, are the same, which is I've got a human being who wants to achieve something. They may not even know exactly what they want to achieve. They have an idea they want to achieve something.

24:24How do we get them to the place they want to be at? And it may not even be the place they thought they were trying to go to. But by the end of this thing, we'll have worked with them to get them where they want to be. I guess sometimes it's as simple as a very simple case. But a lot of times it becomes a more interesting and more enjoyable experience if you're really flexible in the way you're engaging with folks. So I think the teammates thing is absolutely the future of how we interact with technology, not in every circumstance, but in a lot of them. Right. And, yeah, I mean, I think we could probably kind of like debate kind of like the, well, maybe it might be more of a semantic debate than anything else about how we could refer to kind of the, whether it's tools or teammates or kind of whatever.

25:05But I think there's a broader, like, kind of feels like a broader shift kind of like happening. or you believe there's a broader shift happening that that customer experience or the whole contact center sort of space is moving from being a cost center to a strategic growth driver um what makes you say that i mean is that because of the the thing about the kind of the the concierge level of treatment and loyalty that every moment matters all these different things is that is that is that what's driving the shift or is there something else going on yeah there's uh you know there's a there's a wonderful expression that says never let a crisis a good crisis go to waste and the real to me the real benefit of what's happened with the um the modern you know the first chat cpt appearing kind of to many people just completely out of the blue and then the advancement of all these other technologies and claude and gemini and and um nofasonic from us and and and all these different amazing people out there trying to build these new tools is it's caused people to have to think differently.

26:06And thinking differently is almost always really good for you. And one of the things that's really fundamentally changed is just who's in the room. You know, five years ago, it would be an IT director maybe, and maybe there'd be some director of customer service. And that person was getting just beat up by the business to just save money and not do anything but save money, save money, save money. Oh, by the way, save money. yeah and now the people in the room still include that it director who is being empowered to do much more because we have every the tools he's using or she's using are better and then we also but we and and we still have that customer service person but they've been freed up to say yes instead of no so many more times the things that people want to do to advance something and the cmo is in the room the the the um the chief digital officers in the room they are all in there having a conversation they just weren't having before.

26:59And with that being the case, it's just a delight to think that when I get to have that conversation with them, we can start by saying, hey, we probably are going to save money. That's probably going to happen because good customer service is almost cheap, always cheaper than bad customer service in so many different ways. But we're going to, let's put that to the side and say, what are the things, you know, if you could go to your CEO and tell them you are fundamentally going to change the relationship with your customers, what would you want to tell them? And sometimes they have an answer. Most of the time, they take a step back and start thinking.