How Kraft Heinz and Instacart Are Using the Crystal Ball of Data

Visibility, and the ability to draw predictive insights, have been prominent challenges for supply chains across the board. What if, however, brands could tap into precise and robust data that allowed them to take a peek into the future? In a recent webinar, we talked with industry experts from Snowflake and Slalom to explore how leading brands can enhance their visibility in order to score long-term wins and thrive regardless of what the future holds. Read this edited transcript to learn how to embrace the power of visionary insights in order to evade obstacles across sourcing, inventory, logistics, and fulfillment. Lisa Johnston: Hello, welcome to “3 Ways to Build Resilience in Your Supply Chain with Data” webinar. My name is Lisa Johnston, editor-in-chief of Consumer Goods Technology (CGT). It's no surprise that retailers and consumer goods companies are navigating challenges in every link of the supply chain. Whether it's making the right planning or sourcing decisions, managing inventory, handling logistics, or fulfilling orders, companies are getting hit with roadblocks each step of the way. This is partly thanks to items such as inflation, shifting consumer demand, labor challenges, and concerns of a potential recession. We’re joined by two experts to explore the important role that timely data-driven insights plays in having future-forward operations. We'll discuss how this can help retailers and consumer goods companies receive full end-to-end visibility into the supply chain, as well as why that's important and how it can improve collaboration among partners. I’d like to welcome Nancy Dou, global head of retail and CPG industry at Snowflake, and Dan Wittstein, director of global supply chain at Slalom. Nancy and Dan, welcome, can you each introduce yourself to the audience and tell us a little bit about your background? Nancy Dou: Hi, my name is Nancy Dou, I'm responsible for building retail solutions with partners at Snowflake. My background has been in product and I've had the privilege of leaving product organizations across the consumer experience, internal tools and data products at retail organizations. I’m glad to be here to speak about ways that data can help make an impact for retailers and CPGs in the supply chain. Dan Wittstein: Hi, I’m Dan Wittstein, a leader within Slalom's global supply chain practice. I largely focus on supply chain planning and operations execution solutions for our clients globally. I've been here for a couple years with dozens of clients across industries, mainly in retail and CPG. Much of our work is around strategy, transformation, data and tech products, focusing on how to drive agility, visibility, and prescriptive insights for clients. Johnston: Thanks for taking the time to join us today. Many of us here know why visibility is important, but we're going to dig into the impact poor visibility can have across the supply chain. Right now, we know there are a lot of gaps. Dan, how did we get here? Why are things so difficult from a technology perspective? Wittstein: The answer to this question begins before the global pandemic, before the housing crisis, even before electricity in my mind. Bear with me, we're going for a little anthropological journey here. Let’s think really far back as a society to get to the origin of this question. About 800,000 years ago society began to form — early humans gathered around campfires, controlling them. Even then, the benefit was shared information, things around food, predators, etc. About 130,000 years ago, things started to advance. From research through the Smithsonian, we see evidence of these tribes, these social groups starting to connect up to 200 miles apart and exchanging resources. Social networks are continuing to expand and become more complex. This is the predecessor to today, where around the globe, we rely on information for goods, services, and supply chain. It all began to change around 11,000 years ago when farming became more mainstream in the world. The key change here is the advent of private property. Now, tribes that were used to sharing everything across the lands had individual boundaries and domains, and we start to see a little bit more control, neighbors sharing, but not across massive social networks. Why does this matter for the supply chain? We're not tapping into that first 99% of our social history where we had information sharing. We'll talk more about that, but I wanted to share this concept that we as humans, it's in our DNA to share. We've got to get back there. Johnston: I love that you took us back to the beginning, hearing the history of social sharing. Nancy, your thoughts on this? Dou: I think about it in a very similar way. One of the things that's incredible to wrap our heads around is that 80% of the world's data has been created in the last five years. Retailers, CPGs, and their organizations suddenly have informational building blocks that they are sitting on, and have to make a decision on what they want to do with. Taking a step back in retail, while organizations were able to manufacture and distribute goods to brick-and-mortar stores — since basically the industrial age — it wasn't until e-commerce and smartphones evolved in the 2000s that there were major shifts in the way that mail was leveraged as a way to conduct commerce. For instance, Shopify wasn't created until 2006. Walmart, which is a bellwether for the entire retail industry, didn't have the ability to let consumers shop their property until the early 2010s. Then, in terms of the gaps between the different components of sharing, it's over the last 10 years that retailers did whatever they had to to get a storefront online, to be able to have the ability to accept a payment, and then get goods into the hands of consumers. When industries emerge, most of the tools they need to be efficient and to optimize aren't necessarily available. The organizations need to make decisions that they have to for that moment, and as technology and strategy plays catch up — it's amazing to think that e-commerce, as an industry, is but a growing teenager — that's where we are in this evolution of retail and e-commerce. Johnston: If it's a growing teenager, then maybe the pandemic was its growth spurt. We talk about sharing and how it's inherent to us, but why aren't we sharing information in the supply chain? Wittstein: This is one of the largest challenges. When we think about what's happening and where we could potentially share within retail and consumer goods, there are a number of use cases, experiments with POCs, and pilots out there that we've seen. Where we are able to share again is where you are connecting different partners within the ecosystem and the supply chain. We may hear about digital twins, the metaverse, or any type of cross-channel labor planning. What we need to focus on is the data that drives that. What are the processes that enable that and encourage sharing between different nodes of the supply chain? Otherwise, we can put up a POC and start thinking about things on a small scale, but when it rolls up into a much larger organizational, or enterprise-level, initiative is where it comes back to the basics. Johnston: Nancy, what are your thoughts? Dou: What stands out to me is the complexity of all of it. When a retailer or consumer goods company thinks about supply chain, the goal is always to optimize that efficiency and lower the cost. As you look across use cases, it's a challenge to be able to articulate, track, measure, and ultimately, optimize across every component of the different use cases. One of the core reasons why it's hard to share the information across teams and functions of the supply chain is because, honestly, technology wasn't there. Technology has recently come from behind in a state where we can leverage it to start connecting the dots. I'm not sure how familiar everyone is with Moore's Law. In 1965, Moore's Law was a concept, at the beginning of the computer age, when Gordon Moore was the CEO of Intel, that stated he believed the complexity of transistors would double every year and the costs would halve — basically, technology would grow exponentially while the cost of accessing it would be reduced. When you look at every single use case, think about all of the nuances and differences for each different retailer or CPG. It's humongous, it's astronomical. For an industry to be in, basically, it’s infancy and for the technology to not be there to connect the dots, it becomes hard to articulate, measure, track, and optimize. We're finally getting to the stage we’re in today. It's no coincidence we're talking about how data can play a role now. The reason for that is because technology has become accessible for us to think about ways that we can innovate across the different challenges and use cases that the supply chain presents. Wittstein: Nancy, if we can get Moore's Law and Murphy's Law to play together, in the last few years especially, that's the winners in the supply chain. Those who can manage both the accelerated growth of technology and the disruption variability happening. Johnston: Why can't we share across the nodes of the supply chain? I suspect you've addressed some of this, but curious to dig in a little more. Wittstein: This is a hot topic and it's the most important one — it’s the why. We can talk all about the potential and the opportunities, but what's breaking? There's a couple of key elements, but the most important one is competing plans. Think about when you've got different elements of a value chain — perhaps they're all forecasting the same raw material, later stages of differentiation in the product of the life cycle. Those forecasts not being aligned could create massive inventory listing shifts, demand expectations that are never fulfilled, supply gaps, and if you've got a competing plan that you can't share — the basic form of Excel, a portal, some type of common ground system, or something more advanced — that's one of the key areas why things are not being shared today. We’ve definitely covered a little bit of the disconnected tools and data flow, but another one, from the non-technical lens, has unclear roles. What is a demand planner or a supply planner from one organization to the next? How are SOP and SOE viewed? That alone is a question that I get every week, and there's a different definition or multiple understandings. Understanding where the processes and people overlap within the supply chain, as well as what those roles and shared boundaries could look like, is a huge advantage to sharing. More importantly, if we look more toward the technical lens, lack of data visualization is key. It's not simply having a pile of data, putting it into a graph, and starting to see where you’re at today. It's more how do you look forward and think about predictive actions? What purchase orders can be delayed because that supplier can hold inventory longer? What type of transportation options do you have versus having to go to the spot market? These types of data inbounds, and at the point where the decision is made, are huge. Some of the other basics include overly complex workflows. For instance, making sure things in the supply chain, especially from an execution and planning standpoint, are capable of being accomplished in a straightforward, known process. If we could have more trust among partners and go from provider to partner, the learnings are unlimited. Having that openness and connectivity between the nodes will enable a lot of the above mentioned items. Dou: Also the strategy component of all of this. When you think about the supply chain and the hundreds of players that it takes to get an item, to create it and get it into the hands of a consumer — every retailer needs to take their own position on what that strategy looks like and how to optimize each component of all the processes. Dan, as you talk about being able to articulate that and build trust through that, it solidifies this concept that no man, retailer, or CPG is an island — it is so interconnected. Johnston: You’ve mentioned trust among partners. We've definitely heard about progress being made, but there's still a lot of work that needs to be done. It might be helpful to look at some of the challenges, dig in a little bit deeper, and talk about some of the challenges in the supply chain today. Dou: Let’s take a step back and think about what the supply chain constitutes. First, organizations have to plan and decide what they want to sell and what the things are that the consumers will want in that current environment. Then, retailers and/or CPGs have to source the items and ensure they're working with the right producers, or do it themselves, and get the right raw materials. Next, they have to make the item and execute production efficiently, focusing on quality. Once that product is made, it needs to get delivered into the hands of the customer. Then, depending on a multitude of distribution models, there's a lot of decisions that could be made. Ultimately, once the consumer gets it, maybe they decide that it's not a good fit and they return it to the retailer or the CPG — that retailer has to decide what to do with it in order to keep margins as high as possible. Those are all components of the supply chain. At every step of the way, we see a lot of different themes of challenges that have been prevailing. From a planning perspective, teams need to understand demand as it fluctuates with the economy, the times and trends. In sourcing and making, teams are dealing with inflationary challenges and whether it's the cost of raw material, labor shortages, or the cost of labor. For delivering, we see retailers and CPGs get customer-first and take omnichannel fulfillment methods in order to help consumers get goods the way they want to as quickly and easily as possible. For returns, there's a theme in the industry right now, that retailers are at this inflection point where they get all these goods back, but there's a lot of different ways to optimize operational costs with these returns, whether it's to have the consumer keep the item, while then balancing bad actors and being able to predict that. Ultimately, thinking about how to process and lower costs across returns. Those are all themes that we're seeing. Wittstein: One of my favorite stories: There was a CPG organization that we had worked with right around the beginning of the pandemic. If you remember, there were certain lockdowns in certain states, which means one state might be open at 25% capacity for retail stores while others were shut down completely. This created a heavily stranded inventory situation. The example is between Oregon and Washington. Oregon was the state that had closed down for several weeks, Washington was at marginal capacity. The challenge was solved through Excel and a knack in math of how to alleviate inventory that is locked up in retail locations because the stores are shut down per mandate, and get it 10 miles over the river into Washington where it can be sold, alleviating margin, reducing risk for markdown, and ultimately, getting the returns of the inventory that's been sitting there. In this case, we can forget things like the Jackbox games and virtual socials of the COVID pandemic, but we can't forget that story. Lockdowns and such, those were terrible, but they're learning lessons. We should think about them as we do with commonplace problems: A truck is late, there's a strike at a plant, we need to reallocate orders elsewhere. These are the types of questions that a resilient supply chain is able to answer with the right process and technology at-hand immediately versus a week later and several dollars short. Johnston: You mentioned resilience in the supply chain and the problems that can help solve. Dan, how does building resilience into the supply chain help CPGs especially prepare for consumer behavior changes? Wittstein: The idea of preparing for any behavior change is a very difficult job to do, but having a detailed understanding of things like production capacity and capabilities, when and where to make shifts in the cost of doing so, that baseline data will pay off in dividends when crisis hits. Knowing things is a common term, but knowing things like procurement contract terms and limitations, or the order of routing and logic because your vendor can potentially store more goods and scale faster depending on what the contracted terms are. Planning ahead today, thinking about the little parts of the supply chain that will pay off big in a downside situation, that's probably one of the best ways to start building resiliency. Johnston: Nancy, how can data further unlock value within the supply chain? Dou: Ultimately, it goes back to articulating, tracking, measuring, and optimizing. It sounds so simple, but it's those steps. Even though those steps are there and it's clear to everyone that they want to do that, it's actually hard because data is key as a part of this. Snowflake is a tool that allows supply chain teams to cohesively and cloud-agnostically be able to aggregate all data in one place, at one time, to gain this visibility. Dan, you mentioned knowing fulfillment terms, knowing how long it takes to get orders — you were talking about knowing what the buffer is between different CPGs, suppliers, the retailers, and themselves. All of that is captured in data. Being able to know shipment lead times, production schedules, demand forecast, and fulfillment lead times. All the different needs across the omnichannel of different methods, as well as regions. Being able to have visibility into that, to articulate and track that, that is at the core of centralizing all first party data in one place. That's the beginning because then, to help unlock value and build that resiliency, organizations need to think about the entire connected value chain, the partners, the different teams within your own organization that can give insight on how to manage the supply chain. For instance, second-party partner data, such as: supplier inventory data, the inventory data for drop shippers, fulfillment partners like UPS, USPS, FedEx's shipping lead time, the marketing data on how marketing teams have articulated goods and services to consumers that might change the demand, or different promises that have been made through organizations through marketing and messaging. These are a few examples. On Snowflake's Data Marketplace for instance, being able to access data like industry raw material data, CPI costs, foreign exchange rate data, as well as rate risk, commodity pricing, weather, event and geographical event data within certain regions (whether the store is closed in DC, but open in Oregon). All of these components impact the supply chain together. All of these pieces are ways that retailers and CPGs can have a forward-looking hypothesis and a forward-looking view to try to build that resiliency. Let’s take a more specific example of what this means for inflation, for instance as it's top of mind for everybody these days. For retailers and CPGs, as the costs of raw materials rise, CPG producers need to think about the costs of manufacturing goods: How much of it should be passing on to later components of the supply chain and their retailers? How do they incorporate these costs? Retailers then have to decide do they buy less items? Do they increase the price of products to be able to absorb some of the cost changes? Do they enter into new contracts with suppliers? What are the terms of contracts so that they're able to pay on time with inflation being a component of the cost? After that, retailers have to decide how much of the pricing they want to pass on to consumers. All of these are changes that happen month-to-month, week-to-week. For instance, data on the Snowflake Marketplace is one example of how that visibility across can help make these decisions at these critical points. As an example, we have raw material pricing for agriculture and food ingredients through Mintec, or NOMA provides producer price indexes for different industries and the commodities that are associated, and there's multiple data sets on foreign exchange rates. Even with some of these baseline economic data sets, retailers are better equipped to create demand models and pricing models on how much they should purchase and how much they should price their goods, to ultimately be able to see from that demand perspective, and how it will impact the supply chain. It's complex. At the base of all of it, the data is what can help guide retailers and CPGs to be able to figure these different components out. Johnston: It can be complex. We've been talking a lot about companies, what they should be doing, what they should know, what can happen. For the audience, what should companies know? What do they need to consider when getting started on this data journey, or those who still have a lot of heavy lifting to do? Wittstein: It could be tough to figure out where to begin, where the journey starts. The model Nancy provided around sharing data in the secondary/tertiary layer, bringing it back to what ultimately drives the customer decision to purchase, and what impacts the consumer is a great place to start. We also mentioned demand planning and thinking about the elements that could be leveraged for a machine learning forecast rather than a time series forecast — we're looking at third-party inputs from the marketplace. There's a CPG client in the Midwest that started the journey about a year ago. Every month it's taken a step, adding another source of data, formatting and improving the quality and hygiene of that input. Over the last year, we've seen a dramatic increase, around 9.3% forecast accuracy already. Generally, through education within the organization, people are excited about data — it's hard to say that, but it’s true. It's something we've seen across clients. In terms of customer-centric approaches to the use cases, we talked about resiliency and collaboration, but also about what the advanced analytics are driving and how they can be prescriptive actions. At the end of the day, it happens to be a demand plan, but it's up to the end user to make those execution decisions that enable that to happen downstream of the supply chain. One of the elements we hear when connecting with CEOs is that top-down messaging around supply chain initiatives can help amplify that speed. The key is to make sure that we're solving for 80% quickly. Looking at how we get there, whether it's an MVP, something to get moving, and then advancing from there so that there is value. This builds trust amongst the organization that this approach works and will continue to get better. Dou: Ultimately, when you take a data-centric point-of-view, there's three pieces to building resilience within the supply chain: Getting visibility across your data Understanding your demand and movement Upping your hypothesis and visibility across the whole ecosystem by connecting all of the different components in the chain Dan mentioned earlier that data is a journey. You can start from an MVP. What’s important is that it's okay if an organization doesn't go from zero to 100 in a month. There's so many small use cases along the way that can make a difference with data. With Snowflake, there’s a number of existing customers that chose to take bite-size chunks at a time. Instacart, for instance, a score is the availability of more than 500 million items every 30 minutes. By leveraging Snowflake, they're able to do that 40 times faster using a quarter of their own resources. Kraft Heinz decided to engage by getting all of their data centrally in one place first. They migrated half a trillion records onto Snowflake from on-premise to the cloud. They access different data sets from the marketplace, then have the ability to build machine learning algorithms to optimize safety stock. That was their use case. US Foods had item and supply chain data on Snowflake, then used data science to measure price elasticity of demand. These are individual use cases, but even with that, they were able to maximize profitability and gain visibility into their assortment. Once the data is there, the sky's the limit — it’s up to retailers and CPGs on how they want to leverage and drive benefits for their organization. Johnston: Thank you for sharing those use cases, they’re always very valuable for our audience. How far are we from getting to a connected demand and supply planning ecosystem? What are the barriers? Is store-level granularity needed for this to happen? Wittstein: In certain areas, in segments of supply chains, within our clients they are there. It's the connectivity across every store, every supplier, and bringing it back into a fully-connected network. It's a great outcome to be able to get a part of your network connected from top to bottom. Plus, it's where you interact with your customer — except for dotcoms, when it's the FedEx and UPS person coming to the door. You want that experience to be crisp. Having the supply chain work well up to that point is like the final exam, make sure that's the one you ace. Dou: In terms of how far we are getting to a connected demand and supply planning ecosystem, and the role that stores play in that, the way to think about that is as a prioritization of resources. Being able to track and measure things is a somewhat expensive technological function. As the cost of technologies decrease, today we are seeing things where stores are able to have tags, even at the item-level, have scannable barcodes in ways that are then to connect the entire aisle through beginning. Stores are able to measure foot traffic and measure when consumers take things off the shelves using computer vision or other technologies. Those are all things that are amazing in addition to their visions for the future, but it is dependent upon how a specific retailer thinks about how they want to prioritize getting that connected ecosystem. IoT is cheaper. In warehouses today, we're able to track things better, use QR codes, and leverage cameras to be able to scan, measure, and track items, pack size, and how efficiently an item gets through a warehouse or fulfillment center. For stores, it's a flow. It's measuring and planning for the flow of people moving across. As these technologies get cheaper, that'll be what helps determine how long it takes different retailers to have all of that data in one place. Johnston: What are the initial steps and approaches organizations should take from now to build the data centralization? Nancy, that sounds like a question for you. Dou: The first thing that is helpful is knowing where the data is in order to centralize data. Back to this analogy of retail being a teenager, there have been a lot of historical decisions that were made — sometimes in silos, sometimes without necessarily that forward-looking strategy — simply to quickly move and serve the customer. Because of that, data exists in many places across organizations today. The first thing that's helpful is identifying what's the one use case that you can have easy access to that core centralized first-party data. Then, once you're able to know where it is, it’s a lot easier for you to put it all into one place, then be able to think about what is in that secondary data ecosystem, and what's in that tertiary data ecosystem. Wittstein: We're working with a client right now — I won't share the name, of course — we came up with several use cases as to what the starting points are. Nancy, as you mentioned, there were some that may have been higher priority or bigger return, but for others we had the data available. We began with some of those use cases around capacity. Nonetheless, while we were building the first views of the control tower with those use cases, we split-off the team and were able to start mining and pulling in the missing data elements. We can work in parallel and ultimately get that priority one use case out of it. It comes down to the data at the end of the day, when you're looking at the starting point and what's available makes it a lot easier as well as the process is tied to that use case. Dou: There are a lot of solution accelerators that exist on the market to help organizations think about data strategies and data planning. Snowflake has a great partnership with Slalom. As you hear about all of Dan's amazing expertise working through some of the nitty-gritty challenges, it's important to note that if it's something you or your team want to handle alone, there are things that can help accelerate the time-to-market. However, if you don't want to handle it alone, there are teams that are truly experts in this space to help you think about the right way to approach it. Johnston: Dan, I’d like to tap into your broader expertise. We've been talking a lot about the CPG industry, but how are other industries navigating these challenges? Are you seeing or learning of similar themes across the board? Wittstein: I might sound like a broken record, but visibility and the ability to draw predictive insights is a challenge across the board. Several industries hit similar bottlenecks in the respect to supply chains. One recent example is the price of, for those palladium fans out there, rare earth metals and copper, which went up significantly in the last quarter. This may seem like it impacts aerospace and defense or industrial manufacturers, but it draws back into retail and CPG. Let's say there’s a CPG organization, a plant with several old lines. Those machinery require parts that are coming from the industrial manufacturers. Downtime is a huge element of any metric in a manufacturing setting, so the cost of repairs is going to go up. Other inflationary factors and the price of diesel, everything is going to start to drive margin and impact that severely. Another is labor competition. Frankly, it's everywhere. It can go from four weeks to four hours in certain cases if you're doing put-to-light. That enables faster uptime, more labor availability. Definitely those two, I'd say labor and the cost of materials. Johnston: We are nearing our end of time. I'm going to ask you to dust off your crystal ball, realizing that, of course, you’re experts but this might not be the perfect answer. Based on your conversations with your customers, what’s your sense on when we might see some relief in supply chain disruption, what are your customers preparing for? Wittstein: The crystal ball is very dusty. It's out of the camera view. In the most recent round of Slalom’s discussions with 250 CEOs this quarter, most of them are optimistic for the long-term. One actually said, "We fell asleep in 2019 and we woke up in 2030." The amount of acceleration that's happening is great. I believe that volatility is always going to be here. It always has been in the background, but that amplitude and frequency is going to continue to be elevated through 2023. Given that some planning cycles are 18 months for buying, inflation is still going to be a bit of a challenge over the next year. I’d say by the end of '23 we'll see some steady days. Johnston: Well, I love leaving on an optimistic note. Nancy, Dan, thank you for taking time with us today and sharing this insight with our audience. We really appreciate it. I'd like to thank Snowflake for sponsoring this webinar, and I'd like to thank our audience for taking time out of their day to join us

Related Topics