Mars' Successful Venture into Data Products

The data products approach is a fresh way for CPG companies looking to level up innovation, R&D, consumer engagement, supply chain, inventory planning, and more. This approach also sits squarely in the comfort zone of consumer goods executives who are familiar with the processes behind product development, iteration, and winning in the market by design. As one of the world’s largest consumer goods companies, Mars is making strides in unlocking the value of data products in several areas, and not shying away from experimentation and innovation underpinned by a robust analytics foundation. In this webinar transcript, Deepak Jose (global head and senior director of ODDA solutions, Mars), Michael Connor (global head of consumer goods, AWS), Justin Honaman (head, worldwide retail & consumer goods GTM, head, worldwide food & beverage, AWS), and Albert Guffanti (VP, retail technology group) discuss the keys to success around data products, touching upon everything from creating a shared vision for your company, to how to handle change management around these innovations, to leveraging artificial intelligence in a responsible, value-creating way. Albert Guffanti: Hello everyone, I'm Albert Guffanti, the VP of the Retail Technology Group here at EnsembleIQ. Welcome to today's webinar, "How Mars is Introducing Data Products in Consumer Packaged Goods." Becoming more data-driven has been a conversation we've been having as an industry for many, many years. It was very pronounced, recently, at our May Analytics Unite event where everyone in attendance was leaning in on how to bring data to life, how to embed it into the business, and how to move from a siloed data and analytics strategy to a more democratized automated strategy that drives business results. In today's webinar, we'll talk a lot about data products: What are data products and how do you develop them? It's an approach that is taking up a lot of steam, particularly because consumer brand executives understand how to bring products to life. This process of creating data products enables organizations to operationalize the data and analytics strategy. Today, we’ll dive in and look at it from a lot of different angles — technology, consumer goods practitioners, and more. Let me introduce the panelists. We have a great group of speakers with years of expertise and knowledge, and are very passionate about this topic. First, is Deepak Jose, the global head and senior director of ODDA solutions at Mars. Next is Justin Honaman, head of worldwide retail and consumer goods go-to-market food and beverage worldwide at AWS. Michael Connors is the global head of consumer goods at AWS. I’d like to hand it over to Justin Honaman, who will set the stage and provide some lead in discussion to Michael and Deepak. Justin Honaman: Data and analytics in the CPG world has been important for years. It's only gotten more important, and more of a focus in the last 8-10 months as we've gone from focusing on core data and data products to all the AI, machine learning, GenAI, etc. This area is blowing up right now. Again, I’m Justin Honaman, I lead worldwide retailing, consumer goods go-to-market. If you don't know Amazon's vision, it's to be Earth's most customer-centric company within CPG — whether it be food and beverage or any other sub-vertical. We’re focused on three areas: 1) Helping to build great brands consumers love; 2) Innovating beyond the product. As you look at outlet execution, the customer experience, the end consumer experience for them with your product; 3) Centered on operational efficiency, supply chain, upstream, digital manufacturing, all the way through supply chain and that part of the process. If you've been in CPG for a couple of years, you know the last two years, that has been a huge priority. When you think about customers in the consumer product space worldwide, these are some of the big priorities. When we talk about data and then products from data, they underpin every aspect — think about privacy ties to data security, routes to market with direct-to-consumer and e-commerce marketplaces tie back to data, and information around product and inventory locations and meeting demand needs for orders, which then links together and does a better job of planning and forecasting. All of that is tied back to data. Think about sustainability, one of the biggest challenges in CPG right now, is rolling up data into a data product to then talk about the impact the business is seeing around ESG and other parts of sustainability. Smart factory, data and information across manufacturing locations, and doing benchmarking to be able to look at process efficiency. Of course, Generative AI has been the talk over the last couple of months with customers, but all of our talks link back to large language models and foundational models, which then link back to your data and the data pile that those models are running. As we think about data products, there's many different aspects of where to potentially see value from focusing on this space. This is why we chose to do this webinar, we saw the demand at the Analytics Unite event, then put some structure around that, and Deepak brought the message together. There's many areas, whether it be innovation, R & D, consumer engagement, supply chain, in terms of inventory planning and reduction. Each of these ties to a data product. With that, let me hand it to Michael Connor to provide a better definition of what we're talking about when we say data product. Michael Connor: I led a data science innovation team at Coca-Cola and it was interesting to see how we approach data science and try to use data to help the business. At AWS, I work with a lot of the top brands around the world and data is the number one thing people ask us about. Sometimes it's around Consumer 360 and CDP, or overall equipment effectiveness and manufacturing, or return on ad spend advertising. It masquerades as different names, but it's all the same thing. People are trying to figure out how to use data to improve their business, to grow revenue, to improve margins. The other interesting thing is there's so much complexity. Looking at the CPG industry, in the last 100 years, there was not nearly as much change as there has been in the last 10. Over the last 10 years, with all the things Justin mentioned, the pandemic, and supply chain disruptions, CPGs are now moving to direct-to-consumer. They're trying to modernize with smart factories, everything's different. E-commerce has had a big impact, but they still have a lot of the same resources, the same team. They're trying to figure out how to tackle the complexity, new channels, new ways of doing business, and do it efficiently. They're all looking to data to help them figure out how to approach this. There’s an analysis that Deloitte did which I like because everyone intuitively knows data is the new oil. There's been a million analogies. Everyone sees the opportunity, there's no shortage of stories about how data helps companies. Time and time again, when customers come and talk to us, they ask how Amazon approaches data and how other CPG customers approach data in an effective way. Often, they're frustrated. They hear all this excitement, they hear about AI, they're fired up, but at the same time, they're struggling internally to actually get this done. What is a data product? This is interesting. Look at CPGs and where data came from, it tended to come from IT. It was people that were using SAP to close the books, and it was backwards looking — how did we do last quarter? As the business has accelerated, we need more data faster, more real-time things like product recommendations. This older way of approaching data and analytics, like centralized IT teams creating dashboards that tell you how you did in the past, aren't as effective. We talk a lot about data products with customers — that's how we approach things in Amazon. It's a new thing for CPGs, but it's interesting. CPGs do a great job at developing consumer products and have brand managers with accountability. They have small teams, they're scrappy, they're agile, and it's part of the DNA of CPG companies. They know how to do this, but at the same time, don't apply these same approaches to data products. It's not about technology, it's about agile teams, people with accountability, people that are more at the edge of the business that are focused on advertising or supply chain. It's not the centralized teams. Just like a CPG brand, there's SLAs: Look at metrics, hold people to goals, data analysis, APIs, and application of the four components to a data product. How to look at this, how to identify opportunities for data products, how to fund it and operate it is fundamentally different from what’s been done in the past. Some of the common issues when we talk to customers is that they seem resigned, frustrated because our meetings devolve into who’s spreadsheet is this, or we have more data than ever, but it doesn't translate into meaningful business action. More and more people have tremendous numbers of data sets. There's third party companies, second party — they're getting it from retailers and people are still unable to use that to make a significant impact to the business. A lot of times, even if they can get access to it, they can't make sense of what it means. These common failure modes are what we want to talk about and where we're focused. In just a moment, Deepak will share Mars’ story. Mars is a customer we work with regularly that has done a great job in this space and inspired us. Back to the DNA and culture of data and analytics in CPGs. It came from closing the books for SAP and serving the finance teams (and to some degree the sales teams). Most CPGs think of data and analytics as spreadsheets, Tableau, Power BI, or QuickSight and its dashboards. It's very difficult for someone to proactively go into a dashboard, look at it, make any sense of it, and take some business action. It's rare when somebody can do that. When I came to Amazon, it was a bit of a culture shock. We have very few dashboards. Instead of using dashboards, we push everything into automation, figure out how to use the data to make a decision or take some business action, optimize that, and turn it into something that's autonomous. I'm encouraging customers to start with visibility. It's important to get there, but then move to alerts. With alerts, you can tell whoever needs to hear that message what's going on and give them a link to the system where they can take action. It’s level 3 decision support — don't just tell them there's a problem, but provide them with decision support and a recommendation on what to do about it. For example, if you have an inventory imbalance in the supply chain, we recommend you move inventory here. Level 5 is, of course, autonomous. It offers a level of comfort that the system knows and is going to be accurate, makes a good recommendation, and that's when you turn things over to be autonomous. At Amazon, we found that some of our progressive customers are automating all of these things with data products, and then have visibility. They have dashboards that actually show them how the automation is doing. There’s a supply chain control tower that includes every supply chain metric around the world — we can see it in Amazon, but it doesn't necessarily translate into a meaningful business action. Data products in Amazon include product search, product recommendation, our “fulfilled a promise” engine for how to get products to customers for Prime members in under two days, things of that nature. We build teams around it. A lot of this is built around how to organize accountability — there are team founders, who are agile, etc. For the most part, technology has been something that people focus on. It's not about the technology, it's much more about the people, the process, and the relationship to the business. It's the perfect example of working with customers to try and think through what the business goals are for the next few years. There’s been many times we've seen IT do something brilliant with data, but the business isn't willing to take the steps, change management, or train the sales team to actually take advantage of it. And so this is where we see it's important for IT — the data teams, data product teams in the business — to work hand-in-hand. We're able to present the goals for the next year, or the two years, and then they can look at the classic consultant matrix to see which things will make a meaningful difference for the business that are also low complexity and easy to deliver. It’s an example for supply chain, of course, but there are similar use cases for marketing and everywhere else. It's more about the alignment with the business, focus on the teams, and accountability more than the technology. Guffanti: We've equated data products with consumer products, product development. When speed-to-market is so important — we know how long it takes for consumer products to get developed from innovation to rollout — does approaching it like a consumer product slow down the speed-to-market for a data product? Connor: It actually speeds it up. Everyone that is accountable and cares about that data product is on one team that can move quickly. The way that CPGs tend to be organized is by function and process — IT, finance, manufacturing, supply chain, lines of business, etc. In the past, when you try to do something, it’s across all those different groups. Having a data product, it's more like an agile team where they wake up every day, and they live and breathe that business problem to solve, they live and breathe the data, so they can move a bit faster. Instead of relying on a center of excellence, you'll have to put a team together. Maybe there's some hiring, a little bit of startup costs that are going to take a second, but it's well worth it because those centralized teams that are trying to balance different priorities from different lines of business to figure out who to serve first have not been effective. Companies like Mars are building data products with the teams and accountability around it. That's where it's been effective, especially when they can borrow from everything they know about consumer products. These are things that they've known and they just haven't applied to this area of the business. Guffanti: Excellent, and we'll get to the topic of change management as well as who leads it. Connor: Let me turn it over to Deepak, the global head and senior director of data and analytics for Mars, I’m excited to have him join us today. Deepak Jose: Thank you for the introduction, and Albert, thanks for giving me the opportunity to talk about some of the data and analytics journey at Mars. It is always exciting to present something with Justin and Michael, two of my ex-Coca-Cola colleagues. Thanks to Coca-Cola for uniting this group, I've had the pleasure of working with them. Now, I am representing Mars Wrigley and the One Demand data and analytics team, which is the work of many very smart people. When you work for Mars, you can call yourself a Martian, which not many companies can offer. At Mars we believe that the world we want tomorrow starts with how we do business today. Mars is a purpose-driven company that is more than 100 years old, with more than 100,000 associates and 10 billion brands. The Mars petcare division has a fantastic theme of building a better world for pets, including pet nutrition, pet health, — it is our biggest division. I am part of the Mars Wrigley division. If this was in person, I would've definitely tried to share M&M's or Snickers with you to make the presentation a bit smoother. Then, we have the Mars Food Division and the Mars Edge Division. Mars is one of the largest family-owned companies, and as a Martian, I'm proud of our five principles and the principle of decision-making when it comes to the day-to-day of every associate. Here are the key five themes that I'm going to talk about: 1. A brief introduction into One Demand data analytics, the why behind this data and analytics function. 2. What is our strategic framework and an example. 3. How are we building, breaking the silos, leveraging data and analytics, and how are we driving value to the organization? 4. I'll briefly touch on responsible usage of AI, which has been our principle for a period of time. 5. How can you leverage it when it comes to some of your Generative AI experiments and some of my personal learnings? Now, let me start with this: Several decades back, the founder of Sam's Club and Walmart, Sam Walton, said, "There is only one boss, the customer." The goal as a company is to have customer service that is not just the best, but legendary. This is something which holds true today as well. The objective of any large organization, especially in the CPG and retail, should be about customer service. The world has changed quite a bit, but the complexity is also significantly high. What does this mean? Building an integrated consumer journey or giving that unique brand experience. While it is important to succeed in the omnichannel world, the way we interact with consumers is through different platforms. It is going to be a combination of off-platform and on-platform. Let’s begin with the platform. As a consumer you might hear about our products via search, through social media, or by reading our contents and reviews posted on social media. You will primarily interact with us through our own platforms or Amazon.com, our biggest e-commerce customer. You may buy some of our products from brick-and-clicks like Walmart, Kroger, Tesco, Carrefour, etc. There is a new group of customers tracking through our on-demand delivery aggregators, such as Instacart or Delivery Hero. You can also visit one of the M&M's stores, which is another channel to touch and feel our M&M products, characters, etc. There are multiple ways a consumer goes through interacting with our brands. To offer our integrated brand experience, the connectivity between all of this is important, but what happens in a large organization? Many of you might have heard the example of four blind people explaining an elephant, where somebody is touching the trunk of the elephant and calling it a snake, somebody is touching the leg and calling it a stump, somebody is touching the tail and calling it the furry mouse, it goes on. It's a great example of a large organization where a sales function (the elephant in the room here) is the consumer — a sales function might explain the consumer a certain way, a marketing function might explain the consumer in a slightly different way, supply chain function slightly different again. This happens because there are different sources of truth. How can you build a single source of truth? This is why we need a connected decision. This is the fundamental thing that we need to keep in mind when we are building a data product: how can we have a single source of truth. That is the foundation of our belief system, of One Demand data and analytics. We believe that the secret sauce while building a data product is to think about holistic and integrated analytics, which is supported by domain expertise. It can be a revenue management function, it can be a perfect store function, it can be a category function, or a marketing function. It is important to think about these capabilities not in silos, but as a connected ecosystem that is going to be very important. As part of this initiative, One Demand — we call it Order One Demand data and analytics at Mars — we are very excited to help Mars go through its transformational journey. What are some of our key principles? We start with the business problem first, and a data and tech second mindset. We believe if you have a decision map mindset, you will be able to drive the biggest value for the organization. If you are in the data and tech space for a period of time, we used to have a data warehouse, then it became a data lake, then it was data lake houses, and it might be data mesh in the future. It always started with the data foundation, which generated some knowledge and insights to generate some action and value of a specific business problem. At Mars, we want to invert this curve. This is something that we are really proud of. We always start with the business problem first mindset. When you understand and prioritize the business problem, you can prioritize the actions which will generate some value. To drive that, we can clearly understand what the knowledge is, insights needed, and the data and tech comes later. Having a business problem first, and a data and tech second mindset has helped when it comes to building data products for us. This is something a data second mindset, or a data and analytics practitioner should not say, but the business problem first is the key message that I want to drive. What is the key business problem that we are trying to solve to give an integrated brand experience at Mars? We call it the integrated brand experience journey. You might be a consumer of Mars, or any product. Let's say a brand would need to have a purpose to inspire you. You might be thinking and planning Halloween, but then you might be thinking about shopping for a moment, then you might be at a store to buy, then you might go and consume. There is an opportunity to engage with you as a consumer. The interesting part is each part of these brand experiences is owned by different parts of the organization. This is not a connected journey, sometimes it can be a very exponential, or non-linear journey as well. You might get inspired, you might go and decide to buy. You might be consumed, then you might decide to go back and buy. You might be planning and directly going to buy. The complexity of a non-linear consumer journey is very important for us to understand, and that is the business problem that we are trying to solve. To solve that business problem with connected decision-making, we need to have a connected data foundation, which is very important. Various organizations might have multiple different connected decision ecosystems, but it always starts with the decision to what insight is needed for the decision that AI and analytics capability will solve for. There are data features that are going to enable those analytics features; it always starts with a connected data foundation. If there is one key message you leave here with today, it is the importance of building a connected data foundation, which is the single source of tools for the organization to make you successful. There’s an analogy I heard from Brett Munster where he shared the story of a Honduras bridge. In the 1940s, the Central American country Honduras was prone to earthquakes, hurricanes, and whatnot. The Honduran government collaborated with the U.S. Government to build a bridge that could withstand hurricanes and natural disasters — the Choluteca Bridge. This became a tourist attraction. The bridge was built to last and decades have passed since. Then Hurricane Mitch hit the country, causing a lot of flooding, and the river changed its course. Now you have a bridge with no water under the bridge. This is a perfect example of the importance of disruption. We used to build capabilities to last, but now, we need to build to adapt, which is very important. When we build out the capabilities for data products, our ability to make it interoperable, or to add new data sources has helped tremendously. The Pandemic, which was quite a disaster, changed consumer preferences dramatically. E-commerce channels started accelerating, so we were able to leverage a lot of changes in consumer behavior and mindset. That data was then used to drive better decision-making for the organization — building to adapt is very important. The capabilities that we have built fall into three key areas: Strategize, plan, and optimize. We build capabilities to help strategize; we build capabilities to help plan better; and there are capabilities to help optimize better. The data product we built has helped one of our customer teams in making decisions faster. This capability was started as a proof of concept in one of the geographies, we scaled it in that specific geography, and then we scaled it globally. Here is a video from one of my colleagues who leverages the data mindset, the data product, and is a chief evangelization officer when it comes to driving adoption of data products within the organization. Kelly Gehring [via Video]: Hi, I'm Kelly Gehring. I lead one of the sales businesses in the U.S. market, New Transactions. Winning on Amazon requires us to decipher and win the very complex algorithm, oftentimes making real-time decisions around areas like supply category, digital shelf, and portfolio. We get more than one million different data points coming at us, and oftentimes we have many disconnected tools to be able to collect that data. This leads to very slow, siloed, and misguided actions for decision making. The days of piecing them together for a comprehensive story are not tenable, nor sustainable, in the dynamic changing world of digital commerce. Enter Voyager, a data backed solutions tool and one single source of truth for digital commerce. Voyager provides performance and insight tracking across multiple work streams, namely product and portfolio, market, supply chain, investment optimization, and search and category. Voyager is a comprehensive AI powerhouse that drives data backed solutions to fuel a more efficient and effective digital commerce business. Voyager has made an immediate impact on the U.S. business in four areas: We will see a 3-million Mac enablement in 2023, we've freed up over 20-plus hours for our associates per week, there is interconnectivity between our One Demand partners, and most exciting, is the scalability of Voyager. We've been able to take this to our top five customers in the U.S. and scale it across different Amazon markets globally. Jose: This was the beginning of the journey. How did this help us? Last year, we had one of the best Halloweens in the history of Halloweens, as I would call it. Mars Wrigley was recognized as one of the Most Featured CPGs in spotlight because of the best-in-class content and retailer marketing experience that we are able to provide to consumers. This is one of the many success stories. A data product can add significant value to the business from a day-to-day standpoint. What are the key values that a data product can unlock? I’d like to come back to three of the five principles for Mars. 1) Freedom. We need freedom to define our future. We need profits to remain free. The top and bottom line growth impact a data product can make is the most important thing. 2) Mutuality. We don't want to be in the business of creating data products that nobody uses. It has to be mutual to the people that they're using. Our data products have been able to help associates make decisions with a lot of speed and save a lot of time. 3) Efficiency. The data product should be flexible, and more importantly, it needs to keep on improving the accuracy of the decision making. These three principles have helped us launch several data products. With this, I’ll touch on a few aspects of Generative AI and the responsible usage of it. This is a framework that we are using when it comes to Generative AI. We use Generative AI to do things — to do things better, and do better things. There are, depending on the level of disruption and complexity, different capabilities that we are actively working on. The important thing for Mars, is the fact that responsible AI is non-negotiable. We have a corporate AI governance board writing an AI Code of Conduct to ensure there is fairness to the AI modes. This is where it comes to ethical collection of data, we have standards. When it comes to explainability, there is a lot of work that the team is doing already, which we are proud of. When it comes to building a culture of innovation, it should always start from clear and prioritized business problems. It should always be a business problem first, data and technology second. If you have to build a future leveraging any kind of AI or data product, we have to invest in the connected data foundation. Foundation is really important. Lastly, the agile mindset is something we learned from Amazon day in and day out. The idea of two pizza teams, how can we have empowered teams who can take decisions faster? We should not be in a business where it takes six months of alignment to do six weeks of work — it should move fast. The idea of prototyping should be faster, so we can build futuristic capabilities. The most important aspect when it comes to capability is not about tools or data products, it’s about people. There is a honeycomb structure which Mars is using, where tools are one part of the equation. But to drive systemic change, it is a combination of organization, skills and competencies, the processes, and more importantly, the culture and mindset should start from the top. When the leaders start asking data-driven questions, the associates will start giving data-driven answers. I'll get into my second example of the elephant analogy. The organizational culture and change management is critical to the success of a data product. A data and analytics team, or data product team, generally thinks of themself as a rational rider, and the organization is an emotional animal. Unless we cater to the emotional, the empathy, or unless you are able to put yourself in the shoes of a business user, you won't be able to make a transformational change. At Mars and Coca-Cola I've learned that for every dollar we spend on data products, if we can spend an equal additional dollar on change management and organizational upskilling, we will be able to drive transformational change. I am always inspired by Mother Teresa, who said that not all of us can do great things, but we can do small things with great love. I encourage all of you to go back, find those small things that you are really passionate about, and figure out how you can do it with a lot of love to truly transform your organization. Honaman: Thanks. I love it, Deepak, that was great. I have a question for Connor and for Deepak, which is a great segue. Deepak you talked about the keys to success around data products and the things that have to be in place. Connor, we've talked a lot about this with some of our customers. Where have you seen it not work, when it doesn't work, which of those areas are dropped off the SE segments of making a data product successful? When it doesn't work, it's because, what? Let’s fill in the blank. Connor: Are you asking where data products don't work, or where data in general doesn't work? Honaman: Either way. Data vs. data products, you could imply it either way. Connor: Deepak talked a lot about starting with the business problem and working backwards, that's always the big one because of the change management. You've got to have the business on board, it has to be important to them, and they've got to be willing to do the change management. I've never seen a data product that didn't require someone to think differently, operate differently, or change some business process. It has to be driven by the business. If it comes from some data group that wants to try to influence the business — I've never seen that work. Jose: Three things that come to mind when things have not worked. First, when we have a project mindset, not a program mindset. We have seen graveyards of minimum viable products within the organization. Having a programmatic mindset is very important. Second, Michael mentioned the importance of business. We call it the business product owners. The reason for our success is the business product owners. The data and analytics team is one part of it. Business product owners are equally important, and more important, for success. Lastly, is the ability to embed these digital products into organizational processes. The faster we can change the existing organizational processes into a digital one, you'll be able to embed it into the standard organizational processes. Those are my top three. Guffanti: Michael, you referred to it earlier on change management, but who is leading the project? Who is leading the development maintenance rollout of the data product? If we're talking about needing to be business first, is it in the case of say a supply chain issue? Is it the supply chain executive leading it? What have you found from both your side as a solution provider, but also Deepak as a practitioner? Jose: From a practitioner's standpoint, I strongly believe in a business led and analytics enabled mindset. The business problem is always going to be prioritized by the business teams; if it's an e-commerce problem, it's the general manager for e-commerce; if it's revenue management, the head of revenue management — it's always going to be the business. As a data and analytics practice, there are a few things that we can do, including evaluate the feasibility of it using something called a DVF (desirability, viability, feasibility) framework. Data practitioners can do a huge job in making sure that the business problems can be prioritized depending on the feasibility of it, but I believe in a business led framework. Connor: Many ask how we innovate. The way we run every single thing in Amazon is with a two pizza team, which means teams of 8-12 people. There's always a product owner, someone who's leading that, and that person has everything that comes with an agile team — stakeholders, a product roadmap, operational plans, requests for funding — it's that single threaded ownership, which is so important. That's where CPGs fell down on innovation. Innovation often requires 5-6 different groups, including finance, supply chain, you got to go to these different groups to innovate. When it can be that one product owner that can have full accountability, they can move fast. Gufftanti: Got it. Thank you. Well, we actually have a bunch of other questions, but we're unfortunately out of time for the discussion today, so we will certainly address those directly. Parting thoughts from the panelists, maybe some words of wisdom as to how to get started in terms of data products mentality. Connor, why don't I start with you. Connor: It's really exciting to go to your business leaders and say: What is our vision? Where are we headed in the next 2-3 years? Sometimes I'll talk to executives and I'll ask five different executives to write it down. I'll get five different answers. To think as a company, where are you going to go? How will you differentiate? What makes you different from anyone else?

Related Topics