The CX Leader Podcast: "Don't Make it Evolutionary, Make it Revolutionary": Audio automatically transcribed by Sonix

The CX Leader Podcast: "Don't Make it Evolutionary, Make it Revolutionary": this wav audio file was automatically transcribed by Sonix with the best speech-to-text algorithms. This transcript may contain errors.

Pat:
You've heard of the old expression, if it ain't broke, don't fix it? Well, as CX leaders, I think we can do better than that.

Luis:
You know, when I came in, I came into a well-developed program, had processes in place. It had standards. I didn't have to fight for attention in the organization. Everyone knew what I did in Dallas, but I also found a program that hadn't changed a lot over those years. And so I thought, hey, might as well take advantage of being new in this space and see what's out there.

Pat:
We'll take a look at a company that decided that they should build on their success by changing some of the ways they do things, on this episode of The CX Leader Podcast.

Announcer:
The CX Leader Podcast with Steve Walker is produced by Walker, an experience management firm that helps our clients accelerate their XM success. You can find out more at walkerinfo.com.

Pat:
Hello, everyone, I'm Pat Gibbons sitting in for Steve Walker as host of The CX Leader Podcast, and thank you for listening. As we say each week, it's never been a better time to be a CX leader, and this podcast explores topics and themes to help leaders like you leverage all the benefits of customer experience and help your customers and prospects want to do more business with you. We've all seen that graphic common in business lectures and theory. You know, the one, the one that has a circle of arrows demonstrating a life cycle of development, testing, implementation, evaluation – there are many versions of it. It's relevant to many, many aspects of any organization, including CX. They stress the importance of regularly taking a critical look at the processes and methods to make certain you're delivering the best possible experience. And today we're going to look at one company that did just that, making major changes to how they gather customer sentiment and help better understand their customers. My guest today is Luis Angel-LaLanne, vice president of customer voice at American Express, the payment card services company whose brand is known worldwide. And Luis, thanks for being on the CX Leader podcast.

Luis:
Oh, thank you. Looking forward to it.

Pat:
Well, you know, it's great, you know, to meet you and to talk about your program. Obviously, the brand is very recognizable, and I know you've been at American Express a long time, and I'm guessing you haven't been in customer experience that whole time. So nobody, you know, studies this in school. At least very few people do. Tell us how you got into customer experience.

Luis:
Yeah. So it started the way most of us do with naval architecture, of course.

Pat:
Of course. [laughing]

Luis:
My undergraduate was a naval… Was in naval architecture from the University of Michigan, so I was a yacht designer for a few years. And I wanted to make a career change, wanted to get into a field that was a little more dynamic. Even though yacht design still sounds the best at cocktail parties when you say, yeah,

Pat:
Absolutely, absolutely.

Luis:
But so I went to business school, you know, to kind of broaden my horizons, change what I was doing and learned about risk management at American Express. And I felt like, Oh, what a great place to go to combine the analytics I love from being an engineer with all the great business topics I'm learning in business school. So I ended up getting an internship in risk management with Amex and went back full time after I graduated and I spent my first 10 years with the company in risk management, working in different disciplines, different customer sets, and I thought that was a terrific place to start. But from day one, I always knew I wanted to build a well-rounded career American Express and get to experience all the different parts of the company. So after 10 years in risk, I moved over to the servicing or the operations group. And my first role there was to build and lead our compliance monitoring program. So I did that for about five years and that was terrific. It was my first global role and my first role, getting to build something from scratch. At a company is big and old as Amex, it was really exciting to join as employee number one and when I left we were two hundred twenty five people. So it was a really terrific experience. And then from after doing that for five years, it came time to think like, Well, what's next? And this role became available. So I came in, I moved over to the customer experience or customer voice role, and it was, you know, what I would say about that transition was I had the same partners, you know, supporting the operations and the call centers, but vastly different engagement once I moved to this role. You know, I was just grab compliance monitoring as medicine. You know, you have to take it, but you don't want to. Whereas this role is central to our culture and American Express and wanting to know what customers are saying, so it was really fun transition. And yeah, it's been about six years now, just six years in this role.

Pat:
Yeah. So, you know, normally I would be wanting to right away dig into what you're doing today.

Luis:
Yeah.

Pat:
But again, I you know, you've had, you know, some articles and publications and things. And what I'd like you to do is maybe cover the first five years…

Luis:
OK.

Pat:
…because you made some, I think, really good strides in that. And I just want you to tell that story.

Luis:
Yeah. So, you know, when I came in, I came into a well-developed program. You know, I think we'd been surveying customers since at least 2007. So, you know, I stepped into a program that that had processes in place. It had standards. I didn't have to fight for attention in the organization. Everyone knew what I did in it. So that was really, really nice to walk into. But I also found a program that hadn't changed a lot over those years. We were doing a great job, we had a great trajectory, a great baseline. And so I thought, hey, might as well take advantage of being new in this space and see what's out there. And so the first place we started was actually with our technology, our survey platform. We've been with our previous vendor partner for, I think, since the inception of our program. And obviously, technology changed. And so we decided, let's go see what's out there and went out with an RFP and started to explore new technologies and new vendor platforms. And that turned into kind of a public change, which was which was really nice. We were able to engage our colleagues in all the different dashboard users we had using the current platform and say, Hey, what do you want in a new platform? What do you want to hear? We let them join some of the calls with different potential partners.

Luis:
And it was a good way to signal also to the organization that we're now revisiting and reconsidering and reevaluating what we do as a team. And so we took that technology change, which sometimes is the back office thing and made it a public process to signal, like I said to the organization, that we're changing. And so that one, well, we made a switch. Everyone was really happy. And then for there, the next step was to actually look at the survey itself. You know, our survey wasn't really long, but it wasn't short, either. And again, it hadn't changed in a while. The survey invitation hasn't changed in a while. We thought there's an opportunity to uplift both of these things. So let's make the invite more modern, sleeker, more inviting. Let's shorten the survey. Let's pull out a few questions here or there where we either should note we either know the answer today or we should know the answer today, and that helped dramatically that that after we made that change, we were able to increase response rates two and a half to three times.

Pat:
Wow, that's that's impressive. That's not easy to do.

Luis:
Yeah, no. I was really, really happy about that. And it's funny because the idea of that, that change came from one of my partners who challenged us to to make it better. And he said, like not just evolutionary, make it revolutionary. And that became our motto. This is our revolution. And you know, we had fun with that too, in that we tested a couple of different survey invites. We'd send it out to all the leaders in the organization. Let them pick well, which one do you like the best? And then when we got the results back, we'd say, Oh, which one actually won? So the way to engage people and signal that we were driving this change. And then finally, the third thing we did was to change our internal technology or uplift our internal technology infrastructure. You know, when I joined, it was all batch files on the mainframe to get out to the vendor partner, and we moved the survey, triggering to our big data environment. So this one, we didn't really make much of a show of no, but we loved it because it made our program much more nimble, much faster and more transparent. So we were able to keep track of the health of our program. If something went wrong, we could identify it and track it down really quickly. So even though that one definitely got less airtime in the company, it was something that was equally as important as the other two. So, yeah, the first five years was, you know, trying to overhaul every part of the survey experience and just program management.

Pat:
Well, and I love the way you went about it because obviously one of the biggest benefits was getting people engaged and getting attention and notability. And you know, I think in many cases we have we see a lot of programs where one of the biggest challenges is getting people to notice or getting the acceptance within the organization. It sounds like you had a degree of that, but you built upon it very nicely.

Luis:
Yeah, because even one of the other things that we did early on in my tenure was kind of recommit to our beacon metric. You know, we use a version of NPS in our survey and, at the time, customer effort score was very popular. We had a lot of internal executives, internal customers of mine asking, Hey, should we measuring something else? Is it still the right measure? Our scores were really good. Know asking the question like, Hey, what's the point of chasing ever higher scores? And so, you know, I had a strong opinion that there is no perfect measure out there, like, let's make sure we understand the drivers and let's commit to it. So what I was able to do then is get the executives around a table and we had that conversation. I laid out pros and cons of different measures like, Oh, this really focuses on that aspect. So the one focused on a different aspect, and we got everyone to recommit to the one we actually had all along. But what we recommitted to it.

Pat:
Yeah.

Luis:
That was another really nice way to get everyone to understand the theory of what we're trying to do here. It's not just a number that we're picking up, like there's a reason why behind it. And so that actually ended up also being another really effective way to get people committed to what we're doing and not just the metric.

Pat:
So, you know, given that scenario in that background, there are many that would say, Wow, you've got a great foundation, just you could just make some minor refinements and things from there. Or maybe it's time to move on to another department or whatever, but that's not what you did. You have advanced to make some what I thought to be some pretty significant changes. Why don't you tell us about what's going on today.

Luis:
Yeah. So one of the biggest projects we're working on now is rolling out a model sentiment score off of the phone call. You know, the idea is today we're using a survey for our for lots of different reasons. You know, we're using it for our front line incentive. We're using it for to identify process improvement opportunities and then for cross-functional measurement, you know, like how's the call center doing all the scores? And we get that. So we're using the one survey for three different, different use cases. And I think the original idea behind us was we could probably find a better tool for each of those individual use cases because obviously trying to do one use one tool for many things. Some of these are going to be suboptimal. Yeah, we've been exploring NLP and using like transcripts with Amex to answer other questions. And that got us thinking like, Oh, what if we could use the transcript call metadata and other NLP tools to really understand and measure customer sentiment? And if we could do that, we could potentially replace the incentive in the front line, which then allows us to do lots of different things with the survey because one of the things I'd felt over the years is since we use our survey in our front line incentive, I can't change it that often. If someone comes to me with a particular question that they, Oh, I'd love to do this research with our customers.

Luis:
I can't pull a survey out and replace it with something else because now I'm going to disrupt the incentives and it's going to be unfair to certain groups. So we've got to keep that machine running really well and equitably. But if I could move the incentive to sentiment, then I could free up the survey to do much more customer research. So it was a lot of these ideas coming together. And so my team started exploring building a sentiment model literally like two years ago, think it was like 2020 and no one had asked us to do it. So we started kind of on our own. And by the end of 2020, like, Hey, we're on to something here like this could really work. And so last year, the goal was, let's make this inevitable. So we started socializing it and sharing it with people and importantly, doing call listening sessions where I said, Oh, here's where the model matches a survey, and here's where it's different. Let's go listen to where it's different. And so we started that socialization and it works. So now we're at the point where everyone agrees that this is the right thing to do. We're now refining the model to make sure it's it's kind of industrial strength ready. And then we're going to start rolling it out to different portions of our front line with a hope of rolling out to the full network next year.

Pat:
Yeah.

Announcer:
Are you looking for a little recognition for your hard work as a CX leader? Well, here's just the opportunity. Applications are now being accepted for the U.S. Customer Experience Awards. Finalists and winners will be named in 19 different CX categories, and you could submit an entry in multiple categories. This could be the chance for your team to finally get the recognition it deserves. To find out more and submit your entry, go to usacxa.com.

Pat:
So, interesting, let me let me back up and maybe ask you to go into a little more detail on NLP or we have a wide variety of listeners, so just to clarify, it's natural language processing. Maybe talk a little bit more about kind of how that works and really more appropriately, how it works and how you see it working in your organization.

Luis:
Yeah, we've actually internally, we've developed like, you know, we've got a couple of slides that we share with people to help explain, like artificial intelligence, machine learning, natural language processing.

Pat:
Yeah, it all gets kind of confusing, and it's all still very new to CX. And yeah, I think as you showed in your last five years or so, there's a lot of people that they're still really just relying on surveys a lot and you're going beyond that.

Luis:
Yeah, so and I definitely it's I'm glad you asked. It's important to cover because, you know, I sit in meetings about taking care of people like they want to sprinkle a little NLP on and do some AI and it's just like magic.

Pat:
Right. [laughing]

Luis:
And so the way so the way we describe it is kind of really like not a not a scientific definition, but natural language processing is really using computers to understand and understand and then like turn into data, human language. And not just the words, but but go beyond the words into getting that sentiment, you know, so it's kind of a bit of linguistics, a bit of computer science. And so that's that's natural language processing. And so the way we use it is we get from our call recording system, we get the transcript of the phone call, so they use natural language processing to turn it into the word. So we get the transcript. We also get a sentiment score from our call recording system and it is their internal model of what happened on the call. And that's using things like laughter, tone, speed, you know, dead air time, et cetera. So we take the day, we take the actual transcript, we take that sentiment score from the call recording system. We also take call metadata so like length of the call. Again, some of the things like cross talk, dead air time, and we use all of that to build our own model.

Luis:
And part of what's different between the sentiment score that comes out of the call recording system that we've built is that sentiment score coming out of call recording systems just kind of the average sentiment over the length of the call. But when we build our model, we model against the survey, which is the customer's perception of what they like they're feeling after they walk away from that call. And that's what we want to measure. Not the average.

Pat:
Right.

Luis:
You know, like I think of a credit call where a customer might call in being really angry and they might be angry for seventy five percent of the call. But at the end, we get them to understand and they agree and they make a payment and they walk away feeling really good. And so if you take the average, the average would be negative, but the customer could easily walk away from that interaction, feeling good about it. And so that's part of why we build our own model is we really want to link back to the survey to measure what was the customer's perception after that experience.

Pat:
Yeah, it's a great example. Tell me if see if this makes sense. I'm curious what it means to that front line employee that is handling the calls. And you know, what's how is their role going to change?

Luis:
Yeah. So it's it's really going into this. You know, my team was all excited about the model and the results and correlations. And then, I guess, had the ability to potentially use this in the front line to free up the survey to do other things. So internally, like my team, I were all excited. And then we thought like, Wow, how's this going to land on the folks in the front lines? And we got very, very, very nervous thinking like, Geez, how are people going to feel good about their monthly incentive being partially driven by some black box math equation? And it turns out that so far the reception has been really positive and it's been great. And it's funny. This was kind of mentioned to us. We had an early conversation with some, some analysts at Forrester, and they mentioned to me like, you know, sometimes front lines will accept it. They understand because they understand the model is fair and that's what we found. So our front lines, folks so far who have seen it and experience like listen to the model, they understand that the outcome is fair. The model is being treating everyone the same way. So there's that fairness thing that they feel good about. And then the other thing that that helps, you know, our operations, front lines and team leaders and all that feel good about this is the dramatic volume increase. So instead of getting a survey back on every like what, one out of 12 calls or something like that, you're now getting a score for every single call. So that volume makes the makes this score a better measure of your actual performance. So when we when we go share it with different teams, we'll give them. Here's a stack rank of your people last month. Here's how the stack rank would have looked if we'd used our model sentiment. And then does that feel right to you? And the team leader who knows the team really well, he or she will say, Yeah, no, that feels really good. The people at the top should be at the top. The people at the bottom should be at the bottom. So so far, the reception has been have been very positive. And like I said, I did not go into this expecting that, but it's been a very pleasant surprise. Like I said, because of the inherent like fairness of the model and the fact that it's lining up with their expectations and the fact that that getting so much more scale feedback helps people feel good about the accuracy of it.

Pat:
Yeah. Well, and I'm guessing again, just from, you know, the the work you've done over a number of years that the culture is such that there's a value for customer input and so forth, and that, you know, it's not like the first time you're coming to them saying, Hey, we're going to let the customer say something here. And yeah, you know, so and it sounds like you're you're getting action on the customer by customer measure, your managers are getting information they need and you're also looking at some overall systemic changes that might take place. Is that kind of how you divide it up on your team to try to take action on this?

Luis:
Yeah. I mean, I think it's we've got to like the way you just laid it out. There is kind of how we've got to make sure we're all pieces of this are going to be successful. Like we've got to have a really good, high quality model from a math science academic point of view. We got to make sure that the folks who are going to be scored by this feel really good about the change management and they understand it. And we're giving them not just a score, but we're we're helping them see these are the major drivers of the score and make sure they have access to dashboards so they can see how those drivers change with the score. And then finally understand what's the macro impact of all of this and having questions about like, well, what's the future of the survey that if we're doing this? And so these are great discussions to be having at this point to help people understand, like, no, the survey doesn't go away. We still want to ask, still want to use it for customer research, you know, we just don't need to drive the incredible scale we're driving now. So these are, you know, great discussions to have. But we are trying to make sure we're hitting all different aspects of this as we're moving forward.

Pat:
You know, it sounds like you and your team have just been really good about, again, not being satisfied with where you're at, but always looking kind of forward. I'm just curious, what's next? You know, do you you've been focused. I imagine that this whole emphasis on natural language processing and how you're getting additional feedback and tying it together, that's been a major focus. Yeah. But I also get the sense that maybe you're looking at what's next, what else could we be doing?

Luis:
Yeah, I think there's there's two other things one kind of in parallel, and the second is probably a future evolution, but in parallel, we are trying to expand our use of journey surveys. You know, like I said, we're using our survey today in front line. And because of that, most of our surveys are very transactional focused. You know, so you call up your Amex, you'll likely get a survey the next day asking about that phone call, which is terrific, but but sometimes makes it harder for us to understand the experience across a multi-step journey. And again, if we free up the survey because we have the centric model, we now can do more journey surveying. So we're really starting to really lean into that idea more than we have been able to in the past to think like, Hey, maybe we could slow down some of these transactional surveys, wait till the end and do a journey survey at the end. So we're trying to right now go through the five or seven or eight big, you know, experiences that you have with the servicing group at American Express. And just think through, how do we lay out a good journey survey for each of those? So that's the thing that's kind of happening in parallel right now. The next step evolution for sentiment is something we started getting a lot of questions from our partners across the servicing network as we roll this out like, Oh, can we build a centric model that measures the sentiment of the care professional, you know, and things like that because they're thinking like, Oh, that could be really powerful for coaching.

Luis:
It's like, Oh, here's like, does your sentiment match the customer sentiment? You know, like, does your tone match their tone? So I think one of the things I'm very I'm looking forward to and I'm excited about with the concept of modeled sentiment is, I think in the future, it's not just one model. Hopefully it's just like a body of work with like multiple different models and multiple different options. So the team is already looking at how do we get more granular within our centric model, like using the transcript, tracking the words? And can we find the key moments in a phone call where the sentiment changes and really highlight like, right here is where that happened? So I think using sentiment, getting more precise with it instead of saying, oh, here's the score for the call now we can get into nail here. Exactly here is where it transitioned and we're some of the team leaders who we've been piloting with are already doing this. So we give them the sentiment score overall, but then we give them the centrist score for each quarter of the call. And some of them are already literally like using that to coach. Here's where the customer came in angry and you did a great job of helping them understand and getting them to a good place. Here's where a customer came in angry and you were unable to change that. So I think sentiment is really going to hopefully open up like a whole body of work that helps us improve the tools we give to our team leaders for feedback, which is really, really exciting.

Pat:
Oh yeah, I mean, identifying the patterns and things that can really help each individual improve and help your your whole department improve. So well, Louis, we've gotten to that point where we ask every guest for some take home value, and this is kind of the one tip that they can take home today and maybe put to put to use next week or something. So, Luis, what is your take home value?

Luis:
Yes, I think what I was thinking about this and thinking about my team and what we've been through over the years and been through all the time, even recently, the thing that I think is important to me is making sure that you're evaluating your program, your survey program, your voice of customer program for the impact it's having on the business and not just for optimizing your internal survey program. Because that happens to us like we love our program, we want to tweak it and make it the best it can be. We want to make it really slick and have terrific response rates and we can you can get lost in that, you know, and at the end of the year, it's like, Oh my God, we had a great year. Look at all the stuff we did, but maybe you didn't have a great impact on the business around you. So I think something for us that we try to be mindful of, even though we get enamored with our survey program is be sure we're really have an eye on the impact to the organization around you. And maybe that'll lead you to like, step away from the survey program and work on a model sentiment instead. But that's probably the thing that's that's important to me to be thinking about and something that we can all be thinking about at any time in our careers as CX professionals.

Pat:
It's great advice for CX pros out there. Next week, take a step back for a little bit and really look at the impact that you're making and see how it might change your perspective. That's great. That's great. So Luis Angel-Lalanne, he's vice president of Customer Voice at American Express. Luis, thanks for being on The CX Leader Podcast. I really enjoyed the discussion.

Luis:
Now this is really fun. I appreciate it. Thank you.

Pat:
And Luis, if people wanted to continue the discussion, can they find you on LinkedIn? Is that a good way to kind of message? All right. Great. Great. And if you want to talk about anything you heard on this podcast or how Walker can help you with your business customer experience, feel free to email us at podcast at walkerinfo.com. Be sure to check out our website cxleaderpodcast.com to subscribe to the show. Find all our previous episodes, over 200 episodes, podcast series and contact information so you can let us know how we're doing well. The CX Leader Podcast is a production of Walker. We're an experience management firm that helps companies accelerate their XM success. You can read more about us at walkerinfo.com. Thank you for listening, and remember, it's a great time to be a CX leader. We'll see you next time.

Sonix is the world’s most advanced automated transcription, translation, and subtitling platform. Fast, accurate, and affordable.

Automatically convert your wav files to text (txt file), Microsoft Word (docx file), and SubRip Subtitle (srt file) in minutes.

Sonix has many features that you’d love including automated translation, world-class support, collaboration tools, secure transcription and file storage, and easily transcribe your Zoom meetings. Try Sonix for free today.