From Ai4: Coca-Cola FEMSA's Jose Martinez on balancing continuous improvement and CX consistency

Speaker A: Acast powers the world's best podcasts.

Speaker B: Here's a show that we recommend.

Speaker C: If you've ever dreamed of quitting your job to take your side hustle full time, listen up. This is Nikayla Matthews Akome, host of side Hustle Pro, a podcast that helps you build and grow from passion project to profitable business. Every week you'll hear from guests just like you who wanted to start a business on the side. If you can't run a side hustle, you can't run a business. They share real tips and so I started connecting with all these people on LinkedIn and I saw Target supplier diversity was having office hours. Real advice.

Speaker D: Procrastination is the easiest form of resistance

Speaker C: and the actual strategies they use to turn their side hustle into their main hustle. Getting back in touch with your tangible cash and sitting down and learning to give your money a job like it changes something. Check outside Hustle Pro every week on your favorite podcast app and YouTube.

Speaker B: Acast helps creators launch, grow and monetize their podcasts everywhere.

Speaker D: Acast.com guessing is for game shows, not your business. Workday is the enterprise AI platform for HR finance and it. With AI grounded in your context to get work done and done right, it's a new workday.

Speaker E: Hi, I'm Greg Kilstrom, host of the Agile Brand, and here's a question for you. When an AI system gets something wrong about a customer, the customer doesn't experience it as a data problem. They experience it as a company that doesn't seem to know them. Agility isn't really about moving faster. It's about being able to change repeatedly without the company starting to contradict itself. That depends almost entirely on whether there's one usable version of the truth underneath everything. Today we're recording live at AI4 here at the Venetian in Las Vegas, and we're going to talk about what it takes to build that. We're going to be covering a few areas relevant to any enterprise leader, what separates a data foundation that AI systems can actually act on from one that only produces reports, how a company operating across many markets and millions of points of sale keeps its decisions consistent as more of them get automated, and how to make the case for foundational data work when the payoff doesn't look like a campaign result.

Speaker F: Welcome to season eight of the Agile Brand Podcast.

Speaker B: This season we're going all in on

Speaker F: Expert Mode, MarTech, AI and Customer Experience, talking with the people and platforms behind

Speaker B: the brands you know and love.

Speaker F: Again, I'm your host Greg Kilstrom and I help Fortune 1000 companies make sense of martech, AI and marketing ops. Hit, subscribe or follow to make sure you always get the latest episodes and leave us a rating so others can find us as well. And make sure you check out our sponsor TechSystems, an industry leader in full stack technology services, talent services and real world adoption. For more information go to techsystems.com now let's dive in.

Speaker E: To help me discuss this topic, I'd like to welcome Jose Martinez, Chief Data Officer at Coca Cola femsa. Jose, welcome to the show.

Speaker A: Thank you, thank you for having me.

Speaker E: Yeah, looking forward to talking with you about this and definitely a lot of talk about AI in the air here at AI4. Um, before we dive in though, why don't you give a little background on yourself and your role at Coca Cola femsa?

Speaker A: Yeah, sure. Um, my background is a little bit more technical background. I was previously was CIO for Nissan in Mexico six, seven years ago and I started thinking about kind of like putting a baseline in Nissan for to have a great uh, platform for managing data analytics and stuff like that. So I created the data strategy for Mexico but it works. So they told me that I could continue moving forward with the rest of the countries, US and Canada and stuff. This is how I enroll entirely on the data landscape I will say. Uh, but before that I was doing enterprise architecture and working with so many other type of technologies. So I will say it helped me understanding where we're heading to because at that time AI was not a thing. It was the dashboards, uh, era, the analytics era. So when we jump into AI, I already knew what was coming. And uh, now in Coca Cola femsa, I'm the chief data officer in charge of all the data landscape for 15 countries across also America and Mexico. And um, basically what I need to do is to have reliable uh, data ah, that we can use for different purposes so we can mine that data and work with that data for different areas to take decisions based on that.

Speaker E: Yeah. And maybe for the, for the listeners, can you explain a little bit about what exactly does Coca Cola FEMSA do?

Speaker A: Yeah, Coca Cola FEMSA is, is a bottler, is the biggest bottler in the world. Is a, is a bottler that sells the most Coca Cola around the world. So we, we have the production of the Coca Cola different, different type of uh, presentations of Coca Cola and other products as well from Coca Cola, uh, like Powerade, uh, Coca Cola obviously the other type of flavors as well. And we, we sell Coca Cola in cans in bottles, in different type of presentations as well. The bottle safe itself is operating in different countries, uh, being the biggest ones Mexico and Brazil, but it's operating across uh, Uruguay, Argentina, Brazil, Colombia and so on.

Speaker E: Great, great, wonderful. So let's, let's dive in here and I want to start with talking about data at real scale because you know, as the, as the largest bottler, you know, you're certainly dealing with some scale here. So let's start with the physical reality. So you're operating across multiple countries and serving an enormous number of points of sale, many of them small independent retailers. What does that scale actually do to the data challenges that you have? You know what's the hardest part of getting to one usable version of the truth?

Speaker A: Uh, I will say, um, there are two things. Streaming the transactional data, uh, because we are generating millions of records every day, uh, because we are sailing in real time and we are also catching uh, different type of uh, data, financial data, uh, commercial data, operational data, retailers, uh, point of sales data and stuff. Everything at once in real time. And the second piece it would be the amount of data that ah, we generated. We're talking about billions of records, uh, we're talking about uh, petabytes of information, um, that we need to process, that we need to work with in, in a fast way, uh, in a reliable way, in an interconnected way because we have several data sources, several. So uh, right now we're, we are in the, in the um, challenge of trying to create this one single source of tool that can handle all the information at once with one single strategy. Uh, and uh, we are towards that. Uh, because business grows really fast. Technology was not able to catch up in time with all the growing that we were having. So right now we are kind of uh, covering some gaps from technical depth and at the same time moving forward with the new technologies that uh, we are using right now.

Speaker E: Yeah, and to talk about that moving forward, you uh, had a panel here at AI4 talking about turning data into a competitive advantage. So that certainly seems like the right theme there. So in a practical sense for your organization, what does that look like to use data as a competitive advantage? Is it about operational efficiency, better last mile delivery, more effective trade, marketing, all of the above or something else entirely?

Speaker A: It's all of the above and more. Um, you know, data, data is an asset. The data is a new oil, as everybody's saying. And uh, the way you use that data is, is what will differentiate you from the rest of the competitors. You can have your data organized, you can have your data well established, you know, in a, uh, in data lake or data warehouse or whatever, how you use that data, Is the data enabled available, ready to be used? Is the right data there? Because not all data matters, uh, the same. There are different weights of the data that you're using. And also, are you looking at the data as a whole or are you looking at the data as a single, in a single domain, how you interconnect the data? So are you able to find patterns between one of the main, let's uh, say operations and um, financial. Do you see the difference? Do you see the correlation? Do you see where they are getting mixed? That's where you get the competitive advantage. If you are able to see that. And it's not just the tools, it's also the mindset. We're talking about having a company that thinks about taking decisions with facts and data. So if, if everybody in the company is thinking that way, everybody is able to find those interconnections and then use them. Tools are for automation. Normally what you want to do with an AI solution with a, uh, dashboard or something else is to automate a result, to have a result. How you get the most out of that, those tools, knowing what you're looking for, if you don't know what you're looking for, you can ask any AI. If they don't know what you want, they can give you even something worse. They can give you hallucinations or stuff like that. So it's kind of like finding the mix. Uh, I will say that's how you find a competitive advantage.

Speaker E: Well, and I think that's a great way of framing why we're doing. It's um, as fun as this is, to your point, it's to build something that we can act on as an organization. Right. So, um, alongside the data silos and some of the tech data and all of that, there's all large organizations are going to have organizational silos and things as well, in addition to some of the legacy systems and things that you mentioned. To be successful in your approach so far, should you uh, prove things out first? Is it a matter of a lot of proofs of concept? Um, do you integrate broadly? Like what's the, what's the, what's the right approach to follow here?

Speaker A: Yeah, you need to be able to, to fail fast. I will say fail fast and try it again. Yeah, but you need to, you need to balance the cost of doing that because every, every, um, opportunity you have to test and to prove a concept is also coming with A cost associated with that. So how much are you willing to give in order to learn and in order to get the most out of it? That's something that you cannot estimate by uh, doing the calculation. You need to kind of uh, think about how you can get a result, how you can get value of what you are trying to achieve. If you're just testing by because you want to test, you're driving something that is not going anywhere. Probably you can find gold, right? Yeah. Wow. We nail it, right? But that's not always the case. You need to have kind of like a compass. Hey, I'm looking to see the data of this domain because I'm trying to see if I can sell more Cokes in this area. I don't know what's happening there. Probably I can increase my numbers there. Let's see what's happening there. So you have at least a uh, high level view of what you want to achieve and that's how you focus your efforts. And yeah, you need to have your data available so you can test and uh, you can prove. And that's how we are working today. We are trying to test fast, fail fast, do fast, um, so we don't lose time trying to fix and figure it out if we did something good or wrong.

Speaker B: If your team wants a website that looks and feels handcrafted but is still fast to ship, Framer is built for that. You design on a visual canvas with responsive layouts, hosting and a CMS built in so the work is production ready from day one. Agents work alongside you to draft pages and polish sections. Then you review and publish what goes live. Framer is the pro AI website builder for creators, teams and businesses that want a professional site and care enough to get every detail right. Framer is an enterprise level solution powered by agents that provides premium hosting, enterprise grade security and uh 99.99% uptime SLAs that give the world's leading brands like Perplexity and Miro the confidence they need to build their websites in Framer. Learn how you can get more out of your site from a framer specialist or get started building for free today@framer.com for 30% off a UH Framer Pro annual plan. That's framer.comagile for 30% off framer.comagile rules and restrictions may apply.

Speaker G: Before we had ATT business wireless coverage, our delivery GPS wasn't the most reliable. Once our driver had to do a 14 point turn to get back on route. A uh, 14 point turn. An influencer even livestreamed the whole thing not good for business. Now with AT&T business wireless routes are updating on the fly and deliveries are on time. And the influencer did get us 53 new followers though.

Speaker A: AT&T business wireless connecting changes everything.

Speaker E: We're here at AI4 and certainly the term AI as you mentioned, you know it if you would think that AI was invented a couple years ago when you know, it's been around for decades in various, various forms. But you know there's also, there's a lot of types of AI so you know everything from some of the, let's say less complex automation to giving insights, some of the things that you just mentioned to agentic systems that are doing a lot of things either autonomously or semi autonomously from a data quality standpoint. How do you think about that? You know, is there a different bar for quality when, when you're looking at some of these different types of.

Speaker A: Uh, yeah, for sure it's, it's even you know, data quality. Data governance has been there as well for a long time already. Uh, as you mentioned, AI as well in some kind of forms. Before the HYP was regressions, it was uh, naives and it was a lot of type of statistics, uh, you know, tests that normally people were working with data. Then machine learning came in, deep learning came in and that was the new AI. And then now everything is in one single package, automated. And you are asking questions in using neural networks. Yeah, yeah. So data quality, what's present in all that since data before. So you need to have data that is clean, that is curated, uh, works that has uh, the right value so you can run a statistical test, run machine learning or now use an LLM model to do something agentic if you want. If you have tons of data that is not normalized, that is duplicated, that probably is not the right data you need to have. You will not get any result. You can have the biggest brain entropy, kind of whatever you will not have the result you're looking for because uh, the data is not well managed. Um, and that's why it's important and it's even more important today because we are, everybody is using AI somehow right now and the usage of that has been increased. It's faster. So you are getting also bad information faster. Right. You can see Groog for example, if um, somebody in X is generating a lot of misinformation data in X, Brook takes that information as the knowledge race.

Speaker B: Yeah, yeah.

Speaker A: And you can ask a question to groove based on misinformation. That everybody's spreading. And Groov will answer you with misinformation. Right? So they are not intelligent. And, um, even Geoffrey Hinton tells that AI is not intelligence, it's logic, pure logic. So if you, if you are using bad data based on that logic, he will give you a result. The problem is not the right one.

Speaker E: Yeah, yeah. Well, and I want to talk a little bit about the, the people and the organizational part of that too, because I think there's a lot of, and not even, you know, speaking specifically about your organization, but, you know, at large enterprises with disconnected data. I think there's often, with good reason, there's a level of mistrust in some of the data because it's been, you know, coming from all different sources. There's not a single source of truth. And so, you know, as a, as a chief data officer, you know, how do you look at the task that many others like yourselves are in? Kind of rebuilding trust in the data because for lots of reasons, but also they need to be able to, teams need to be able to trust what the AI is going to do. Especially when you have a lot of leaders and others that have been in an organization for sometimes 10 or more years, they think they have a way that things got done and they trust themselves more than the data. So, you know, how do you, how do you look at that, that task?

Speaker A: You need to be a little bit of a seller, um, a little bit of a marketing person. You, uh, cannot be no longer a technical slash data scientist guy. You need to, you need to think about also business. And what I mean by business is you need to show the value of having the right data, of uh, involving the right teams to understand the data that we are working with. So you know, this because stewards, right, from different areas. So if you go with, uh, if you go to the board of directors and say, hey, I need this, this team to do governance, they will tell you what, why. Well, what is that? Was that, what's that for? I mean, it's going to slow me down. Right? Right. It sounds like, sounds like compliance. Right? But if you, instead of saying that, you say, hey, I think we can decrease costs of usage of data. I think we can get faster information. I think we can get more information that we don't have right now in the data like that we're using. I think we can, with that information, we can create more use cases like this, this, this, this that can give us, uh, increase of revenue, cost savings and stuff like that. If you start thinking about that you are adding the value to the data. When you add the value to the data, is, is more often that you can, you can involve more people because they want to do that, because everybody wants to add value. Everybody wants to, to put the mark in the company. So you need to be able to convince them that that's the right way to, to work with, with the data. And that's just data. Uh, everything you do, even compliance or even all those areas that doesn't seem like are giving any value, they are giving, they are doing that, but we are not able to sell that because we normally, we are engineers, we are people that are technical that we don't think about that, but we need to, we need to do that.

Speaker E: Yeah. Yeah, I love that. Love. Well, Jose, thanks so much for joining today. I have two last questions for you as we wrap up here. Uh, first one, you know, we're here at AI4 in Las Vegas. What's been a highlight of the conference for you so far?

Speaker A: Um, it's a really nice place. I mean, being here with all the people that, you know, talks about AI, talks about data in one single place. It's a really nice place to be, um, because you don't see so many people working in the same area at once in one single place. So you can talk about this kind of stuff, this nerdy stuff. Normally you don't do that very often. In other areas you can do that very well. You can look at, you can lose yourself a little bit. So that's great. And also look at new technologies. It's always nice because it gives you ideas. Yeah, everybody's using AI, but how they are using it is the right, is the right question. So if someone said to me, hey, you can ask a question to your data. Hey, I know, yeah. When someone says to me, hey, this is going to be working with you in the backline, I will organize all the information, I will do this, I will do, uh, that for you. You don't even notice. That's a new search that probably could make more sense, uh, for some companies to have because they have, uh, tons of people working in the backlines doing, doing that. So you start thinking about the value, right? Oh, okay. So I can do this and I can increase this cost and I can do that. So this place is a nice place to start, you know, thinking about that and, and find, um, out new ways of work.

Speaker E: Yeah, that's great. And last question for you. Uh, what do you do to stay agile in your role and how do you find a way to do it consistently.

Speaker A: For me, I will say it's automatic. Uh, I love to be learning every day, literally learning every day. Uh, every day I read something new. A new paper, a new technology, a new conversation, survey, whatever. Uh, not just because I want to be on, uh, top of the subjects, but because I like it. You know, I want to see things and ask and question things because I don't think anything for granted. No matter who says whatever they said, I always look at that and say, hey, is that right? Is that the right way to do it? And that's how I started thinking myself and motivating myself, um, on look at things in a new way, uh, or even try to implement something that I saw and see if that works. I have found out that something that is hyping right now, it's not new. It's kind of like they are just inventing a new term to name something that was already existing. So yeah, it's marketing at the end because everybody wants to sell something. But I try to kind of like see if that makes sense. Uh, and also how can I apply that to my real world?

Speaker E: Yeah, love it. Well, again I'd like to thank Jose Martinez, Chief Data Officer at ah, Coca Cola Femsa for joining the show. We can learn more about Jose and Coca Cola Femsa by following the links in the show notes.

Speaker F: This episode is brought to you by Tech Systems. They're leaders in full stack tech services, talent solutions and helping companies put it all in action. You can learn more@, uh techsystems.com that's Teksystems. And thanks again for listening to the Agile Brand podcast. If you like the episode, hit subscribe and drop a rating so others can find the show too. And if you're interested in consulting, advisory work, or if you need a speaker for your next event, feel free to reach out. Just visit GregKillstrom.com that's G R E G K I H L S T R o m m.com the Agile brand is produced by Missing Link, a Latina owned, strategy driven, creatively fueled production co op. From ideation to creation, they craft human connections through intelligent, engaging and informative content. Until next time, stay curious and stay agile.

Speaker G: Before we had AT&T business wireless coverage, our delivery GPS wasn't the most reliable. Once our driver had to do a 14 point turn to get back on route. A uh, 14 point turn, an influencer, even livestream the whole thing. Not good for business. Now with AT&T business, wireless routes are updating on the fly and deliveries are on time. And the influencer did get us 53 new followers, though.

Speaker A: AT&T business Wireless Connecting changes everything.