The Customer Experience Podcast
The Customer Experience Podcast

Episode · 9 months ago

195. Customer Voice and Sentiment Analysis w/ Luis Angel-Lalanne


How does an iconic brand like Amex evolve to stay connected to their customers? By driving their CX Experience with innovation.   

We take a deep dive into customer surveys and look at how Amex uses the latest technology to collect and distill relevant data.   

In this episode, I interview Luis Angel Lalanne , Vice President, Customer Voice at American Express , about improving customer listening using technological tools that allow for higher quality customer feedback. 

Lewis and I also talked about:

  • How Amex handles CX internally 
  • Best practices for executing customer surveys for high quality data 
  • A playbook on balancing and processing qualitative and quantitative feedback
  • Using NLP to uncover patterns from customer surveys 
  • What implementing new CX technology looks like at Amex   

Check out these resources we mentioned during the podcast:

Subscribe, listen, and rate/review the Customer Experience Podcast on Apple Podcasts , Spotify , Google Play or Google Podcasts , and find more episodes on our blog. 

Listening on a desktop & can’t see the links? Just search for The Customer Experience Podcast in your favorite podcast player.

The single most important thing you can do today is to create and deliver a better experience for your customers. Learn how sales, marketing and customer success experts create internal alignment, achieved desired outcomes and exceed customer expectations in a personal and human way. This is the customer experience podcast. Here's your host, Ethan Butte. Where have we been, where are we now, and where are we headed? With Voice of the customer, customer listening and sentiment analysis. That's what we're talking about today here on the customer experience podcast, and we're doing it with a gentleman who spent more than twenty years at am x, or American Express, in risk management, compliance testing, customer communication and other disciplines. From more than six years now, he served as vice president of customer voice, Louis Angel Alan. Welcome to the customer experience podcast. Thanks very much. It's great to be here. Yeah, I'm excited to have you and I'm especially excited to dive into sentiment analysis, you know, perhaps doing it from text and transcripts, perhaps doing it from voice, perhaps even doing it from video. But before we get there, we've got a few things to cover first, and we're going to start where we always start, which is customer experience. When I say that to you, Lewis, what does it mean? So I try to take a very generic like view of customer experience and that's just like the customers perception of what they've been through, and I definitely want it to I add to like the unfiltered customer perception of what they've been through. Like I that's really important to me, you know, it's one of my other responsibilities right now. is also complaint reporting, and a lot of the complaint reporting it recorded in our internal systems as like we take the complaint from the customer and we record it. So I always refer to that as like that's customer voice, it's customer feedback, but it's it's, you know, a little bit indirect, you know, because because we we wrote it down where I think like customer experiences that direct, you know, customer perception of what they've been through. Yeah, I really appreciate several things that you shared there, and one in particular is the idea that the customer, it belongs to the customer. They get to define it. We can try to capture, we could try to talk about it, but even what you offered. There's so often we mediate it through some other filter or system or circumstance and come up with our own treatment of internally, but it really does belong to the customer. Do you feel like customer experiences like new language for old things, or is it new language for new things? Like, you know, I just feel like we're talking a lot more about customer experience today than we were three or even five years ago. Like where do you think we are with customer experiences? It just kind of driven by the market now, like we need to be better at it. Like what do you think about the use of it now? I think it's it's it's really interesting question, you know, because one thing I really enjoy about being in this space is that that there's an industry. We have podcast we have articles about it and I feel like we're developing it as we go, you know, like this isn't a well developed like feel that's had thirty, fifty, seventy years of experience and smart people thinking about it, and so I really like that we're developing it as we go. So I would I think I'd answer you like my perception of where we are with customer experience. I've said as an industry or a field or a discipline. Like we're still evolving, you know, we're still like it, I would say, probably early in our journey, but I think like the concept of customer experience is up. You know, I would expect to be like really, really old and like you know, something we've as humans have been concerned about forever. We just don't necessarily talk about it as a discipline like we do now, so that they think that's what's changed. Is the like the forming a discipline around customer experience. That probably didn't exist fifteen years ago, but obviously smart, good business people were definitely concerned about it. Yeah, I'm with you on that and and your answer makes me feel more assured about my own lack of confidence about customer experience and why I love talking with so many different people that are approaching it from different angles. Just this this permission to know that it's...

...alive, that it's evolving and that it's something that no one really has a perfect handle on. I know it isn't exactly your discipline, but just for context for listeners, to the degree that you're familiar with it, how is amx organized around customers? Means is there a customer experience division or team that you're kind of partly aligned with? Like how have you all been tackling it like high level? Yeah, it's a really interesting question and and something like you know, we talked about internally we're right now. The custom streens measurement piece of it is kind of distributed across the different functions and business units of American Express. So, for context, I sit within the servicing or the operations team, so this is like the group that manages all the call centers and all that. I've got peers who sit within the marketing groups and they run the the product and PS surveys, you know, and I run more of a transactional customer service experience program so we all have a community of practice inside mx where we get together, we talk, we share best practices, but structurally we're kind of embedded in each of the different business units or functional areas that American Express. But, like I said, as a community of practice, we make sure we come together and are aligned and in our looking to share best practices and just make sure we're standardized. As you would imagine, when we first came together we had groups measuring, you know, satisfaction on a seven point scale, some on a five point scale, and so that has been some of the benefit of coming together as a community of practice, even though we're all embedded in in the organization. So, you know, a similar question I've received is like who owns the customer American Express, you know, and there is no chief customer officer. You know, I'd say, like it's responsibility, it's distributed. Everyone would say they own the customer, which, you know, there's there's. I think the positive to that is that, like that's a better outcoat. It's a good outcome and someone saying they don't own the customer. You know. So that's the way we're set up today, but we still talk about is it optimal? Are there other things we can be doing differently? But for the time being we're all, you know, embedded in each business units function really good. Thank you for sharing that. And I would also guess, like who wants the customer a precursor questions. Who is the customer and how many different customers, types of customers, are there? But did this let last question here. Yeah, did this kind of getting together across teams, like you know, you mentioned, like you know, getting together on it. Did that happen organically or was that something that, you know, someone was really just woke up when day I was like I need to know who's doing the see the other deaths, like how did you get together? And is it like a monthly thing? Is that an informal things? That like a slack channel? Like what is that? What does that look like? Yeah, so it started like as a grassroots effort. You know, when I moved into this role six years ago, we went out and did an RFP for our survey program like the platform infrastructure, and we moved to a new vendor. When we after at the end of that and as we're getting to know the new vendor, my leader and I were sitting down thinking through all the relationships American Express has with all the different survey providers and I think we stopped counting we got to double digits, and so that really that like so that was five and a half years ago or so and that really came to impetus of like we should clean that this up, you know. So we kind of just started talking to peers, talking to leaders in the company and when you put that on paper, no one's going to say, oh, that's optimal that's what we want. So we then started working toward an enterprise survey platform. So, you know, we went out with a new RFP for that the whole enterprise, and we got all the different survey program owners together and everyone had an equal voice at the table. You know, if you had a small program or big program, we all had an equal voice and that kind of started that. That was a genesis of US starting to work together as a community of practice. And you know, we try, we try to run like a monthly meeting of just getting everyone together. That's ebbed and flowed. There's Times onlike we miss a couple months, but yeah, it's really just us, like me and my peers deciding to make it happen and my team. When we went to an enterprise contract with with a our qual tricks, who's our survey provider. You know, my team kind of took the lead and providing some dashboard support for the whole enterprise. And so we take the lead and getting the getting the group together. But it really is meant... be be a peer exchange. Really good. I love that language to community of practice. That's the word I'm struggling for and asking that follow up question and I won't forget it now, community to practice. Speaking of practice, give me a little bit of definition around customer listening or customer voice or Voice of the customer. I mean your title is Vice President of Yeah, customer voice. Talk about that practice a little bit and you just draw some boundaries around it. Yeah, so when I first moved into this role it was to manage our transactional, you know, customer satisfaction survey that goes out after you've had an interaction with the service network. So this is you call the number on the back of your card, maybe you have a dispute, maybe you apply for a card. Anything that's run and owned organizationally by the service organization would get a survey from my team and so when I joined that was that was our scope. We added to it complaint reporting, recognizing that complaints are another key, important piece of customer feedback. And then, like I said, when we made the decision to go for an enterprise survey platform relationship, I touched on this, but the what we realized is, rather than having each group across mx develop expertise in survey creations board management, my team was already in there. We were already doing it with qual tricks. So let's you know, it's easier to expand my team just a little bit. So my team also has kind of center of excellence responsibility for supporting all the other email surveys across American Express. So you know, the product and PS surveys, or my team helps execute those. You know, we helped with their dashboard, survey creation, file management and we've also helped with some infrastructural changes to help, you know, make analytics more efficient, etc. So that's kind of my scope right now. As I sit within the service organization, I still have the primary responsible of running our servicing transaction survey, but I've complaints reporting as well as this enterprise center of excellence. Awesome. Share the basics of when you survey people, how often, how sensitive are you to how often, why do you survey these people at these times and perhaps I know you've experimented a lot and really driven up response rates to some of these surveys. Share what you can just from a really practical perspective for people that are, like you know, maybe much, much less mature in when to survey, how and how to get people to fill them out and how to make it something useful for both sides of the relationship. Sure you know our our survey poyam today, is is based on triggered by the Trans Action. So if you call or, like I said, if you have even if you start something online you don't talk to a human, will still survey it. Basically what we do is we collect all the interactions from the day and then that night send a file at call tricks for the surveys to go and will de Dupe and all that. If you contacted US multiple times or through multiple channels, like maybe you had a fraud interaction as well as a regular call the number on the back of your card interaction, will make sure we only send one survey, and so that's that's kind of our process now. Is the goal is to or like the hot the starting point is we're going to survey every transaction, but then we will we've got a thirty day suppression where we'll rest people, for our customers, for thirty days, whether they responded or not, and that's yeah, that's just to give you know, our customers a break. If they're there have to interact with us more frequently and then, particularly we'll have things like some of our corporate card member, you know, who manage the corporate relationship. They might talk to mx every other day because they're managing like, you know, five hundred corporate cards and so like literally on the phone us every day. We we don't need to survey them every other day. So today, like I said, we're trying to rest people for thirty days to make sure that we're giving people a break and also, theoretically, we don't necessarily want feedback from the same exact people every day in our scores. You know, we'd rather get a more representative distribution. And so the scores come back in or the surveys come back in and they're used in frontline incentive as well as score cards going all the way up to the top of the Servicing Organization, for Denise Picket, who runs our service organization, and so that's we've got. You know, they're used in score cards and then, of course they're used in process improvement,...

...where I've got an insights team that's looking for process improvement. And you know, when I do to the question of how did we improve our response rates and get that up, you know, when I joined we'd had a program since two thousand and seven and I think so. I think it was like thirteen years before I got here, and so we had a really well established program. It had we had implemented it across all twenty four markets where American Express those business. It touched his every touch point in service. Saying so the you know, my predecessors had built up this global infrastructure and when I got here, I think it was we realized that's great that we have this global infrastructure, but we haven't modernized the survey or the experience in years and years and years, and so that that became kind of like an opportunity to say, Hey, I bet we can get something out of it if we modernize it. On the the other piece of the opportunity is one of my one of the leaders, one of the partners that I support, coming to me saying hey, because we use this in frontline incentive, we stick, we always are chasing more volume. So he challenged us to not just get incrementally better, but to get revolutionary better. So that became our our our mantras is, you know, revolution, not evolution. And so we started with the survey invite itself, like, let's change the look and feel, let's remove words, let's, you know, put some pictures in there, let's find some ways to make it more engaging for customers, and then we shortened the survey dramatically and I think you know we only had eight or nine questions in there before, but we shortened it to two questions plus an open end, and that was that was tough because we got rid of some data that some people still wanted. But, you know, I think what helped is as the world had evolved in our internal data at Americas, I said evolved, we've found ways to collect most of that data internally without having asked a customer. And then if there are a couple little pockets where we still didn't have it, we just decided, like you know, we're not we didn't want to have little one offs here and there. We want to have a standard survey. And also, you know, one of the questions that we took out was asking the customer why they called, you know, and that our decision at that point was, hey, we know most of why they called from internal data and if we don't know a hundred percent, we don't want to put it in front of the customer that we don't know why they call. Like this is supposed to be top like, you know, high quality servicing. We know who you are, and then the next day I send you an email it says why did you call me, like you would ever do that to a friend. So that was the other rationale. You know, rationale like that. We used to help really shorten the survey and it worked. We got our response right up two and a half to three times and it's stayed up. So we're really, really proud of that. So many good tips in there and obviously thing in it. Thinning it out is critical. Putting it upon yourself to answer the things that you are that you can answer yourself. Instead of the Jack the easiest thing is just to add it as another question and get someone to tell you again, and then it's like right there. It's a layup, but you already have it, and so why not put that together? A lot of really, really good advice in there and and I love the revolution, that evolution challenge in mindset. Let's talk a little bit about you said you know there's an open ended question. Let's talk a little bit about qualitative and quantity tative feedback. How do you balance those? What are you doing those? Qualitative feedback, and I assume that's going to kind of lead into the conversation around customer sentiment in particular. But share some thoughts on balancing and processing qualitative and quantitative feedback. Yeah, and it's a really important one to us because the quantitative, you know, their scores. We ask an overall satisfaction and a transactional MPs question. So you know. So we ask your overall satisfactory that, based on this transaction, would you recommend am x? And those become the two primary scores and we had, like I said, many other questions behind that which we've pulled out. So next is an open end that you know, if you give us positive ratings, we say you know what went well and if it gives not positive ratings and what can we do better? And when we did this we had to really up our game on our ability to analyze the open ended verbatim, because that was our only source of insights now really other than scores. And so what helped us is just, you know, the improvement in n LP and categorization that had been happening over the year.

So we felt it was good enough now that we could rely on getting consistent categorization of what's going on. So now we've got data and trends on the categorization and how is that changing over time? The ability to like search for key topics has been a big one for us as well. You know, we'll have a peer, maybe a marketing contact, reaching out to us saying, Hey, I manage this product or this feature where we want to go refresh it. What can you tell me about customer feedback, and we can go search for that keyword or set of keywords in give that, you know, give that marketing manager the feedback they're looking for. So that's that was a key part of our ability to really reduce the survey and get in the extract more value out of the the qualitative. The other thing we've done with qualitative is, you know, depending on the needs, just share it unfiltered with leadership, you know, so if there's some particular event going on in the world where mx is in the news, we might go pull that down and send comments on it. If there's even like a hurricane rolling through a particular part of the country, we might pull the commentary that comes in the week after that and send that to leadership and say here's what people are telling us about the hurricane, what they've experienced in our response, in our way to support them. So that's the other thing we've had more of a focus on over the last three years and obviously covid was it was a great example of that of being able to take verbatim on a weekly basis and we did that for a think a year. Every week we sent out verbatim related to Covid and then we expanded it to non COVID, but at first it was just covid to really help, you know, leaders in all parts of the organization understand what a customers telling us in this time of like tremendous uncertainty, tremendous change. So yeah, the the the focus on the qualitative and the ability to extract it insights from it has definitely increased dramatically over the last three or four years. Really good stuff, really specific sub question. But just for folks who are listening to tell whether things are going well or poorly for them, you like what is do do you have any idea? And you may not. Do you have any idea about how many surveys, like what share of surveys come back with something useful in that qualitative open field, because I know it's someone who feels out surveys in my in my own life for variety of different companies and products and services that you know. Sometimes I'll, you know, answer the three or four questions on the scaled response and skip the box and sometimes I'll feel it out. You like, you know, what have you seen in terms of when or why or how often to people give you quality, to feedback? Yeah, I don't I don't remember the exact numbers, but I remember the scale and what was interesting is when we did the when we shortened the survey, we actually got less of a smaller percentage of customers leaving US verbatims and leaving US commentary, and I think it went from like forty percent to twenty five percent, fifty percent to twenty five percent. And we're I think we're around twenty five percent now, maybe twenty maybe twenty eight, I don't remember, but I think we're in that range today of customers leaving US commentary. But the thing we noticed is even though we lost a large percentage of customers responding with commentary, the quality of the commentary has gone up a lot. So I think we lost the people who write. It's good, it's fine, you know. And now what we're left with it but like the people really have something to say are still leaving US commentary. So we initially, when we saw those stats, when we did the survey transformation, we felt we didn't feel good about losing that much commentary. But then as we dug into like verbatim length and other measures, we realize like, Oh, I think we're potentially getting the same amount of genuine insights, even though we're getting, you know, a much smaller percentage. So it's somewhere in the twenty percent of customers are leaving US commentary, I believe. Awesome. You already mentioned it once. You mentioned n LP, natural language processing. Assume that machines are helping. How quite a, quite a great deal there give us from your perspective, in your experience and really your expertise, you know, where have we been, where we now and maybe what are you excited about in the future with regard to sentiment analysis and the use of NLP, machine learning, etcetera, to really get even more value from these sources of customer feedback? Yeah, we've been and...

...we've been spending a lot of time exploring the space within my team because, like I said, for you know, about three four years ago, we really when we shortened the survey. About three years ago we shortened us our survey and really started to rely on NLP within our survey dashboards, you know, in the qualitrix platform, using their engine to categorize complaints. are kind of complaints with comments, and that's been great because now, like I said, we've got stats, we can do trending on to this topic going up or down over time. The other thing we've done, though, over the particularly starting two years ago, we start to dig into it ourselves. So we've got we've been hiring for that skill set and people on the team have kind of upscaled themselves. So my insights team now has that modeling skill set and so we've been using our own, you know, homegrown skills and tools to dig into commentary. So sometimes if we have a varied, like precise query, we want to go stop find a very precise thing in the commentary, will go do that ourselves. As we get all the data back into our date environment, will go right our own queries and go look for very precise things. The other thing we've done, like you mentioned around sentiment, is we started to model customer sentiment off from the phone call. This has been a multiyear effort. We started, think in earnest, in two thousand and twenty, but it started as a grassroots effort. My team was just exploring, like no one came to us and said go do this. They they said, Hey, we've got transcripts now, which we didn't have from our phone calls. Let's go see if we can model this. Les Can see if we can understand customer satisfaction. And so for two thousand and twenty the team worked on it. We looked at it, looked good, like Oh, this is interesting, and that was about it. And then last year we realize what we had was good enough to do something with. So last year the goal was let's make it inevitable and let's start socializing this with the organization around us. And what they really entailed was looking at or building a model that takes the input from the phone call and models it against the survey score and basically the particular the satisfaction question in the survey. And the idea is we can now replace the the survey in our frontline agents incentive, monthly incentive, with this score and there's a lot of positives that come with that. You know, you you move from a sample based population of surveys to now I can score every call and you know with that comes more consistency. You know, when we look at when you look at agent performance month over month, you know the model score versus to survey. As you would imagine, with with the more consistent measurement, the model score is much more stable and so far we've seen really good results in the correlation between the two. It's about seventy, you know, anywhere from like sixty eight to seventy eight percent correlation between the survey and the model and we feel really good about that because we know the survey also pulls in experiences outside of the phone call. So for the percentage that don't match, we've done a lot of call same like listen to hundreds of calls where the survey and the model differ and ninety percent of the time we agree with the model and it's because the survey has pulled in other experiences, whether it's been multiple phone calls, whether it's been fulfillment issues after the phone call or just you know how you feel about the American striss brand. You know you might call, have a bad call and give us a benefit of down back. You know the call wasn't great, but mx, you've been great to us all along. So the model does a better job of really distilling what happened on the phone call. And so that's so now this year the goals to start to roll it out. So we've got our first team of CARE professionals, as we call our frontline agents, going to be switching to this modeled incentive in April. So we're really excited about that. Super I hope that goes really well and I expect that it will, certainly based on a the the patients that you exhibited in experimenting with this, putting this into play, validating it, correlating it, etc. And I think something that shouldn't be missed by anyone listening is this really interesting layer that you added there toward the end of that response, which is if you can find out how people are feeling and reacting and speaking in the moment, it's probably given pee a lot more accurate than their recall twenty...

...four, thirty six, forty eight, seventy two hours later, where it's kind of bundled in with what else is going on for them in their life at that time, in that moment, their past history with you versus what was truly happening in that moment. And so it's really interesting to apply technology in a way that helps us understand the person in the moment much better than they can understand themselves, you know, removed from the situation. I should have asked this earlier, but for the ignorant among us, give us a basic definition or tell us a little bit about what's going on with NLP and in the machine learning behind it, like just give us a little bit of lay person's terms around what's going on there. Yeah, so natural language processing, you know, we think of it at a really high level. As you know, computers being able to take language and get meaning from it, you know, so understand the words and understand the connection between the words. And then the capabilities are getting so much better. Now they're understanding tonality above it, so they can really start to layer in sentiment. So what we're using in our model is the output from our call recording system. That gives us a sentiment score and it gives us the transcript. So we're using the NLP, the NLP engine embedded in our call recording system to give us that like the transcript, as well as that sentiment. And then, you know, machine learning, again, I think in a generic sense, is the machine using the input you give it to really kind of like muscle through lots of iterations to find the most meaningful connections. You know, I think people think like machine learning and ais is all like magic and I often think of it as like, no, it's a lot of it like is brute force labor where the the computer can turn through hundreds thousands of iterations and combinations of the data that you've given it to find the pattern much faster than you could, you know. So I that's one of the things we talked about internally at mx is like it is not magic, it is sometimes just brute force, but it can do it's super, super fast and that's what gets us to where we are and I think that helps people understand it a little bit better. And at the end of the day it's still a black box. But if you understand that, it's not doing some wizardry, it's just going through combinations of of the data that you've already fed it to find the patterns and find what really predicts you know, I said, I think it gives people confidence and it helps people trust it a little bit better. Yeah, and I think the key there, of course, this quality of the data, and you have such a large volume that I assume that that's something that's years passed in terms of worries or concerns of yours. And I guess for those of us that are probably not as data matures as a company like mx, give any tips on making sure that good information is going in we're enough information is going in it, like any basic like tips or recommendations for people that don't have the quantity and probably not even the quality of the data, knowing of course that, yeah, a the ability to find the pattern and be devalidate the pattern is as useful and accurate is dependent down that that sweet blend of quality and quantity. Yeah, and unfortunately there's not much you can do to shortcut that a quat on the quantity side, you know. I mean we discovered that when we first started exploring this model. You know, we built it with a couple months of data, like Oh, that's interesting, and then we refreshed it, added a couple more months and then added a couple more months and added a couple in it and of course, you know, with with that like dramatic scale, it's get it got better each time we refreshed it. So I don't I don't know anyway really to shortcut, you know, our ability to deal with with getting scale and quantity. But in terms of like quality, that's a really interesting one. You know, I having been in my role six years, I've you know, there're certain exercises and insights or, you know, regression models to understand drivers. I've seen done three times now, you know, and the first time we pulled a bunch of data and in the second time, like Oh, we've pulled some different types of date and third time pull some even different types of..., and each time you get a slightly different outcome because of the data or feeding in, you know. So one thing we try to be mindful of when wherever we're doing analytics or insights or is to try to think through what data do we want to pull into this that's that's actionable, you know. So we obviously have a ton of data. We know, like a customers age and demographics and their their credit risk profile. Depending on the analysis we're doing, I might not pull that in because as a sert using organization. I don't have the ability to control the demographics of who calls me. So if I know, look, we get better scores for for, you know, this generation, verse generation. Like okay, I don't know what Going to answer the phone when that generation calls? So so where? That would be huge and that is hugely important to our marketing organization when they're doing their insights and servicing. That's not so I think that I don't know if this is a super useful tip, but I think what's important to me, at least in my team, when we when we're driving insights and just collecting data about our customers, is let's make sure we're collecting data that's usable to the problem we're trying to solve, because I think, you know, I sometimes joke that we have a bit of a curse on this team that most of what we do is interesting. So you go share your insight, are so wow, I love this, this is great, and then you go back two months later say what did you do with it? And I'm like, Oh, I didn't do anything with it. I'm like there's your curse. Like it would be better if they told you during the meeting like I'm not going to do anything with this, it's mediocre. But no, they tell you it's great, so you walk out thinking it's really you know, like wow, we nailed it. So I, you know, to counteract that that curse is I you know, I we like I said, we try to make sure we're really thoughtful about the data we pull in and is it actionable to the to help solve the question we're trying to address. Such a good tip. Two looks to the future and the first one that out the asked you to maybe speculate on it some at some level, you know, in the past. I think we're manually cranking through a lot of this stuff to the degree that we were collecting it. Obviously you've brought us into this. Let's do some NLP and let's do some sentiment analysis off phone calls and LP reading and categorization on the on the quality to feedback. Now we've got transcripts from phone calls, are we going to tone of voice or urgency in actually truly listening to the calls and not just reading the transcripts? Are we may be going to video and perhaps even facial expression and body language laid over top of tone and pace and urgency in the voice. Like where are we relative to reading and turning those things into maybe useful information? Tier to the degree that you're familiar with it? Yeah, so an mx, we've like just we've does done some little exploring in video servicing, but but not much. They're so I don't have much of an opinion yet. But on the voice side, I totally think we are going to that's where I think the future is going, is to continue to get better and better at understanding like tone, pacing, urgency, emotion and how that flows through, like when we're decent at it now, but I think the future is going to be all of that getting much more refined, you know, because I think of like our sentiment model is turning into a terriffic coaching tool. We've seen, you know, because when we give someone to sentiment score, we're giving that on every call. We can then filter for different dynamics and if you want to coach for like three things, we can help filter for that and give you those three things with the lowest sentiment and really target the coaching. We also give the sentiment score for each quarter of the call. So the first twenty five percent of the call, second, third, fourth, exacts. I bet it turns sometimes right. Yeah, and we've seen timmy, the leaders, using that, like we didn't ask them to do it, but like they've seen that in like Oh, I'm going to coach on particularly if it's like a credit call that you know, where the customer I call come up angry, like where are you good at d escalating the call and where are you not? And they're using that to help find those moments. So I think the future is going to be get beyond, to get more precise within space like that. You know, and I've already had my modeling teams already trying to look at the transcript and trying to find the word combinations that change the experience. I think that's going to be the future. So,... coaching is no longer going to be just so look through the quadrants of the call and find what's change is like. Now we're going to tell you the moments in the call where things went positive, when things but negative, what led up to that, what happened afterwards? So I think the future is going to be just I think a lot of the concepts were talking about now, but just much more precision. Yeah, and perhaps even going live. I mean I feel like I've probably be distracting in a call, but perhaps even the potential to do it in a much tighter window. Yeah, because we're doing it. I'll after the fact, you know, like the transcript gets process, comes in, goes into our big date environment, we run our model, get a score, push it back out to the agents dashboard. But yeah, you're right. I definitely think that's another piece of it is is making that time window such a like you know it's happening live on the call or you hang up the call and you see the score instantly. You know when you're taking you know what, like fifty calls a day, I think is is the average for our agents. You know, the ability to get them the feedback in real time is really meaningful. Like when they get the feedback a day or two later, they they don't remember that call. You know, they have to go look up that call, go listen to it again if they really want to remember. So that's definitely something we're spiring too, is to really tighten that window super the other question I had about the future, and I know you're doing some experimentation here and I know that it's early stage, but you know we've been talking about customer sentiment, customer analysis, customer voice, and I know you're doing some experimentation with employees and taking some of this stuff internal. How are you thinking about that? What was the motivation? Share anything that you're willing and able to about about that, because I think one of the in the reason I'm asking Lewis, is a constant theme in these conversations when I'm talking to people from variety of disciplines about the employee as the most important customer of the organization, and so share anything you can about about turning that a little bit inward. Yeah, so the first what the first part of my answer will be kind of a continuation of what we were just talking about in terms of sentiment. Is going to create the opportunity for better coaching and better understanding of what's happening on the call. So that's going to be a positive impact for our frontline colleagues, which I'm really excited about and you know, like I said, we're hoping to roll out in a few weeks with the first bats I've been at. I attended the first, one of the first training sessions the other day and you know, our frontline agents are recognizing it already. You know they were. They already said, oh well, this helped me with some of this, and we're like yes, it will, you know. So I think that's that's really exciting, that that, just like having better tools, more precision, will help them get better, because they obviously want to. They want to understand what's working what's not working. On the flip side, though, of like now using this to also understand the frontline per experience. We're we're really, really early in that. So I don't have a lot of best practices to share. I can tell you more about the conversations we're having, you know, because I think something like modeled sentiment, in my mind in the future, is not a single model. It's more like a platform at which you're doing a lot of things. And one of the things that we've already talked about is this modeled sentiment is modeled off of the customers a voice. You know what they're saying. Can we build a modeled sentiment off of the care professionals voice, and then now you can start to match those two up and you would expect. If a call is going really well, you've got a good match between the sentiment of the two. It's the calls going poorly, you know, that's where you might see a mismatch, you know, and so we're you know, we're thinking the concept of sentiment applies both directions and that's something we want to get to. All our energy, though, right now is on making sure that, as we're rolling this out, it is industrial strength, it is ready, it is robust, everyone believes that, everyone buys into it. But, like I said, off to the side my teams already starting to explore some of these ideas. Awesome. Thanks for sharing that and and good luck continuing down nothing path. I think there's a lot of exciting opportunity in that. And a little slightly related to that. You know, you've been it at mx for over twenty years now. Of course, you and I have seen lots of people...

...and lots of people we personally know, bouncing around, especially over the past year or two. What do you think keeps you anchored in there? Like what's something you think the company is doing right in terms of culture retention? I think just my own observations and spending a little bit of time with you before and during this this conversation. You know, obviously I feel like you have this, you know, freedom to pursue things that you're personally passionate and interested in. You've had the opportunity to bring some things together and take on some more responsibility. So I already see a lot of the pieces in play. But from your perspective, you know, what are some of the things that are going really well? They're from a cultural retention standpoint, personal or more broadly. Yeah, it's funny. I think a lot of the answer I'm going to give right now, having been here twenty one years now, would probably be the same answer I would have given after maybe like six or twelve months at the company. You know. So one of the first things that drew me to American Express out of business school was the ability to join a company with world class risk managements. I joined risk management. So my guard, I'm joining a company with the World Class Risk Management. I know. It's a company with World Class Marketing, world class operation. So I saw for myself the ability to build a career, a long career, and experience these different disciplines, you know, at a world class level within one company. That's always interested me, the the ability to, you know, like learn different things and experience different things. So I really appreciate that about Ammx, that, you know, you can move around and it's expected that you move around it. Like we want our leaders to have a broad understanding of how the company works. So that's a big part of it for me. And then you you mentioned the other big part of it. For me, it's like the ability to have impact. You know, I've never ever, even as a you know, someone who'd been here six months when I first joined, I never felt like I was just doing busy work. You know, I always felt like I was doing work that had mat like impact. You know, when I first joined I ran the risk management credit policy for internal acquisitions, and this is right after we launched the Blue Card. So we were a cross selling blue cards to folks with charge cards and like I ran that. Like that was really exciting that like people were getting these offers and signing up for cards and getting the Blue Card because of like the impact I had and and getting to feel like you have a real impact is really important. And the third thing I would say is about just culture. Like, you know, I'm working with smart people people I respect and that helps a lot. You know, like I sometimes hear Hor stories from from peer peers or friends, and I appreciate that. Like it's a company of smart people who care about the brand and about like our experience and then the customers experience and like what we're building, you know. So I think all of that comes together to like to make us feel like we're doing something bigger than ourselves. Like, you know, the company was here well before us, it will be here well after us. So I think it creates a perspective of it's not about what we can each extract for ourselves out of the companies, about like what can we build within the company. So it sounds crazy, but I don't believe you know, it's worked for me for twenty one years. That's awesome. It's something I think most people are still searching for in their career. I know I found it here. I've been at the same company for over a decade now and it's just been an absolute joined pleasure for a lot of the reasons you already to find quick fun one here, even at the risk of even at the risk of fronting log you and I were both at the University of Michigan for a few years together. We never met, you know, back in the s. But if the University of Michigan sent you a survey back asking about student experience today, what would you put in the quality to feedback box? It's it puts short response on the student experience you you had. Oh, I like just complete love of University of Michigan and all of it, all that it had to offer from you know, like I was an engineer Undergrad, I was a naval architect, so I was a like working to be a yacht designer, really small discipline part of the College of Engineering. So I loved that. Like I got to know my professor's I love that. I we have a really cool towing tank at the versity Michigan where they test ship models. When I was...

...there, the Americ one of the America's Cup teams, you know, like the highest level of yacht racing. They were testing their boats. I worked on that, like I just went down, knocked on the door, said Hey, can I help and they said yes. So, like I got to I got to do that as a student. And then just being part of like this huge campus with like fantastic sports, with the FAB five were there. I fat you know when I was there. So like totally that was like basketball was fantastic, football was fantasy, like all these like great experiences of getting to come together. So yeah, I'm a huge fan. We were back there the summer doing the college tour with my daughter, who's a junior in high school now, and yeah, we weren't shy about just telling her it's the greatest awesome. So good. I'm glad I asked. And then, and to tie those last two responses together, you can find intimate communities even inside very, very large organizations. You know them. University mision is not a small school and it's even bigger today than it was when we were there, but you can still find a lot of intimacy and value and impact in there. For folks who are listening, and I enjoyed this conversation so far with Lewis, I want to point you to two other ones. Back on episode sixteen of the Customer Experience Podcast, I hosted Lauren Culbertson, who's the cofounder and CEO of a company called loop, VOC VOC of course, standing for Voice of the customer, and we called that episode closing both loops with Voice of the customer and it's about closing those little loops on cervix service exchanges, but also rolling that up and closing the bigger loops and these themes that you're seeing as you do this Meta analysis of all of the small loops so that you can really tie things together and do some preventive work as well. So that's one hundred and sixteen with Lauren Culbertson and then a little bit more recently, episode one hundred and fifty five with Dr Roland Rust who's a professor at the University of Maryland and in their business school and Co author of the feeling economy, and they called that. We called that episode how artificial intelligence is driving the feeling economy. There were a lot of themes of machine learning, sentiment, customer sentiment and how our feelings and really customer empathy are being advanced because of the advances in AI and machine learning. So that's one hundred and fifty five with Dr Roland Russ Lewis. I've loved this. This has been a pleasure. I know I've it's been really fun. Kept you a little bit, but I've got a couple more things for you. First, because relationships are our number one core value here at bomb bomb. I'd love for you to give a thanks or a mention to someone who's had a positive impact on your life or your career. Yeah, that's a that's a really, really great question. It's fun to think back on. You know, I'm gonna go with personal and just say my sea scout skipper from high school. So I was, I was a sailor. I love to sail in high school and see scouts is like boy scouts, but we just had a boat. We went sailing and so my sea scout skipper, you know, I was part of that program for all four years of high school and he was great. He treated us like an adult all the time, you know, like if you're out sailing with him and you're like what's that, he's like, go get the chart and figure it out. When it came time to do the summer cruise, said, you know, we'd go where we going to go? It's like, get the chart, figure out where you want to go. How long is it going to take to get there? Like, you know, when we went, when we had to buy food, he'd like you guys are going to the grocery store. You're going to figure out how to buy food. So he was like a great influence and actually someone I'm going to go see, take the day off and go ski with on Monday. So yeah, I definitely want to shout out to Bill Austin, who was, you know, a terrific influence for me. Awesome enjoy. One last question for you. When you think about great experiences that you've enjoyed as a customer, what brands or companies come to my and do like maybe what's a good what's a company that always gives you a great experience? Yeah, it's funny, you know. I I want to say USAA. You know, my fatherin law was in Vietnam. So so we've got us a insurance and I know they're they're always held out as like great, you know, and PS and all that, and it's funny, like they're one of the first ones I want to say, but mostly because of again, there and PS. Maybe they're servicing experience isn't always great, but they've done an amazing job of making me and obviously lots of other people feel like their job is there to support us, like their job is...

...not to squeeze every last penny out of me. So when I've had interactions with them, it's always been good or fine. I wouldn't necessarily hold it up and say all that interaction was the best and it was so much better than every other interaction I've had, but the fact that, you know, it's coming from a company that I genuinely believe isn't out to screw me. They're not out to get me, they're really there to support me, and I think like that culture they've built is so inspirational for all, like all of us in the CX space, you know, to like really have your customers believe that you are there to have their back, and I know amerket expresses fires to be that for our customers. I hope people feel it about us. But yeah, I think I think USA like I think they do a great job of kind of living what they you know, kind of like with their marketing and branding. Is All about, yeah, really good car. I love that reference. We have our home and our cars ensured with them as well, and I say the dividend checks always helped to that. Like makes you feel like you're of the whole thing. And I would also say for folks listening, the Lippincott human era index if you search that thing. They were trying to figure out which brands were regarded as the most human, and this survey was like like tens of thousands of consumers of a variety of products and services, and they came out on top of the and not just in the financial services but overall, cross all categories, beating out companies like apple and some other, you know, category leaders as being the most human. Cited that in our most recent book, Human Center Communication. Anyway, Louis, this has been absolute pleasure. Of people have been enjoyed this time with you as much as I have. How can they maybe connect with you or learn more about your work or about American expresses? There anywhere you'd send people to learn more? Yeah, it's say, you know, I'm on Linkedin. It's just my name, you know, Louis Angel Alan. Feel free to reach out, you know, like you reach out with a question. You know, I try to have connections, keep my connections to people I've had interactions with. So so, yeah, if there's a question or like, well, someone wants to continue the conversation, like that's a great place to reach out to me and connect with me. Awesome, sounds good. I link all this stuff up in a post that we write for every single episode. That's a Bombombcom podcast. Louis, thank you so much for your time today and for you listening. Thank you for spending this time with Louis and me not thank for much. One of the most impactful things you can do to improve customer experience and employee experience is to include some video messages in your daily digital communication. Explain things more clearly, convey the at emotion and tone, save time by talking instead of typing, prevent those unnecessary meetings. There are so many benefits to using simple videos and screen recordings, and bombomb makes it easy in email, linkedin or slack messages from Gmail Outlook, sales force outreach or Zendesk. Learn how Bombom can help you and your team with clear communication, human connection and higher conversion. Visit Bombombcom today. Thanks for listening to the customer experience podcast. Remember the single most important thing you can do today is to create and deliver a better experience for your customers. Continue Learning the latest strategies and tactics by subscribing right now in your favorite podcast player or visit bombombcom. SLASH PODCASTS.

In-Stream Audio Search


Search across all episodes within this podcast

Episodes (234)