Albert Guffanti: Hello, everyone. Welcome to our webinar today, “How The Modern Milkman Delights Customers Through the Power of AI.” My name is Albert Guffanti, I'm the group publisher of RIS and CGT here at EnsembleIQ. I'm thrilled to welcome you to this webinar. We are in for a fascinating story about how a retailer achieved seamlessoperational execution with their customers by using AI and machine learning.
What I'm interested in, and excited to learn about, is how The Modern Milkman simplified the user experience. Many retailers are trying to figure out how to get products and delivery operations set up in a way that it's a seamless, delightful experience for customers, including right up to their doorstep. We're going to learn a lot from this incredible case study that we're going to dive into today.
I’d like to turn to our extremely experienced and expert panel here. Joining us for this discussion, we have John Hughes, the CIO of The Modern Milkman. We also have Tom Summerfield, who is retail director at Peak. Scott Landoc, the global lead of grocer, drug and convenience retail at AWS. Gentlemen, it's great to have you on this webinar, I'm looking forward to diving into this topic, witnessing, and listening to all your perspectives.
Before we go any further, I’ll let each of our panelists introduce themselves – tell us a bit about your experience, your roles at your companies, your company's mission, and what you do.
John Hughes: Hi, my name's John Hughes and I'm the chief information officer and co-founder at The Modern Milkman. We set up The Modern Milkman about 27 months ago to be perfectly accurate, I'm a data guy after all. Our vision was to reinvigorate a lost tradition in the U.K., which was doorstep deliveries of everyday goods, and taking the strain out of the weekly shop.
We started by bringing a milk round in the Northwest of England, a place many of you won't have heard of, but it's Colne to us. We started with one round, one van, and everything was written down on paper. We went round to collect the money after we'd done the rounds, and realized that we were duplicating the work in one respect. When we went to collect the money, expecting a thousand pounds, we got 50 quid, which is when we realized, "this doesn't work."
We set about building some technology. I started to get involved just after the technology tender because I want nothing to do with what was the first hash of our technology. The idea being Simon, our CEO, knew operationally what he wanted to do. He knew strategically what was important to him, and I came in with a vision to lead a data-driven business. I'd worked in previous businesses before, where data was used to mark homework rather than lead strategy. I was constantly leaving jobs when people were ignoring my advice because I was able to build a prediction model or highlight a type of analysis that said, "You are doing the wrong thing; you must do this."
Being okay with numbers and willing to back myself, I said, "This is pointless." I met Simon at the right time and took what he had. Everything he worked on was Google Sheets, colored cells, etc. I took it, analyzed it, and broke it down, to say, "This is your weekly spend. This is how much it costs to get your customers. These are your attention curves."
That was very early on before we started to introduce the app or anything like that. It was very much the building blocks of what would be our strategy going forward. We've always been tuned into the numbers effectively. Rather than saying, "Oh, I want to look good, tell me what these numbers are, and we'll tell a story around the numbers." It's literally, "If we do this, these numbers should change, and therefore we're doing a good job." We used the data to lead the strategy. Anyway, that's a brief intro for us.
Guffanti: Thank you, John. I'm looking forward to diving in further into that. Tom?
Tom Summerfield: Thanks Albert, great to be here. I'm Tom, the retail director at Peak, which means I have a view of everything we do in retail at Peak. We define retail as more of a consumer facing businesses, we have more traditional retailers within our customer base, and then exciting new consumer-facing businesses like The Modern Milkman and everything in between.
We are a decision intelligence business, which is a new category of software, actually. What it means is the commercial applications of AI. We're a tech business specializing in AI, but specifically specializing in how AI can be put to actual commercial use in businesses for revenue, top line stuff, and then profits, which impacts the bottom line. We play in the retail, CPG and manufacturing industries, but our most mature experience is in retail. We've got some cool customers among our ranks now, which we're proud to call The Modern Milkman one of.
Guffanti: Thank you very much, Tom. And Scott, why don't you give us your intro as well?
Scott Landoc: Glad to, Albert. Thanks again for the opportunity to participate. I'm Scott Landoc, I lead up three of the global retail segments — grocery, chain drug pharmacy, and convenience fuels — which is a fast-moving consumer goods area of retail. As part of Amazon Web Services (AWS), most know the story of Amazon, and that for the last 15 years we've focused on the technology opportunities with cloud and digital transformation, to help re-modernize business across a number of different initiatives.
One of the most important was the process and value of decision intelligence. I'm orienting our message around helping customers, retailers around the world with transforming that engagement and experience with customers, building brand, focusing on optimizing operations from one side of the business to the other. Then, the opportunity of improving insights, predictive and demand intelligence, and what role that can play in helping businesses differentiate in the market.
Guffanti: Thank you very much. How we're going to tackle this case study is by first talking about the business objectives, how that informed The Modern Milkman's strategy. We’ll continue on to discuss technology adoption and execution, and then switch over to outcomes, lessons learned, and a look to the future.
John, tell us a little bit about, from the customers' perspective, what the motivation is to use a service like The Modern Milkman? Clearly the convenience of it, but the idea of the milkman also has this nostalgic effect to it as well. From a business perspective, what are your customers telling you?
Hughes: Ultimately, we're a purpose-driven organization. Our purpose is to reduce the use of single use plastics in the food supply chain. The co-founders were inspired by David Attenborough's Blue Planet, when they began thinking, "Can we do something about this?"
What has happened in the last 30-40 years, as the supermarket shop has grown we've used a lot of plastic to preserve the food that sits within warehouses, and in supermarkets as well. We've reached a point where all that excess plastic takes many years to degrade — if it does at all. We've built up this technical death, effectively, with plastic that we use.
Essentially, we're trying to reverse that and take out as much plastic as we can. We're the only grocer in the U.K. that doesn't use plastic in the goods they provide to customers. That is what we're trying to do, eliminate plastic from the supply chain.
Guffanti: I love that purpose-built approach. It is a modern take on the grocery business. Of course, something as simple as delivering items to someone's doorstep is actually quite complex. There's a lot that goes into it from working with suppliers, the whole supply chain, the delivery, and everything in between. Can you share what the mechanics of that involves, what you're trying to deal with internally?
Hughes: We offer customers up to three drops a week, depending on where they're located. They will be served at a hub, which shouldn't be more than 30 kilometers away from where they live.
In that hub, stock is delivered on a daily basis, and it's disaggregated and loaded onto vans. Then, vans go off in petal formation to deliver the goods based on the drive route we have. We're doing this six nights per week, so it's heavily labor intensive.
Essentially, we are operating through the night. What we deliver is the stuff that you use in the morning, so there's a very small window of opportunity. If we fail, we let the customer down. I'm not saying that we never fail; we do. I have data sets to show you how many times we failed today, etc. It's really about minimizing failures, and that is where we come into decision intelligence.
Guffanti: Very interesting. I want to turn it over to Tom and Scott. We always say you can't adopt technology for technology's sake. Artificial intelligence has to have a business objective and a vision fueling it. It seems to me that John's approach with the business objectives first, the vision, the strategy, and then the technology adoption seems to be in-line with successful case studies. Can you comment on that? What's the chronology of technology adoption?
Summerfield: I just picked up on something that John said about minimizing failures, which is an interesting point. A data scientist from Peak, way more intelligent than me, once noted: What is the AI doing? It isn't creating flying robots or anything like that, it's actually if you can minimize it. Intelligent people in businesses are making thousands of decisions a week, and it's such a convenience-driven business.
The Modern Milkman takes a more pragmatic approach on how to never disappoint a customer. By leveraging AI and making that decision, making massive decisions across the granular details, is really fascinating.
For adopting the tech, a big emphasis is on adoption in the business, then with the users who need to have that explain-ability. You can demystify what AI is for people by mapping it to existing business challenges, processes, and opportunities. A lot of the time it's common sense that begins to kick in. It's a way of optimizing existing processes. What's fascinating with what John and the team have created, is that consumer first, never disappoint a customer approach. AI is a common sense enabler, hopefully along that path. It's a journey though, so you have to start small and grow it. Hopefully, you're bringing people along for said journey along the way.
Landoc: I agree with everything Tom said. From my CIO days through my CTO days, trying to help retailers on this path – and my own colleagues when I was a grocery CIO – think about the value proposition of data. Often I would talk about the fact that things are just increasingly complex. That was many years ago, today we're dealing with massive shifts in customer behaviors, customer wants and needs. We're dealing with unique challenges around managing supply chains, variability, especially in the world that we've all lived in over the last 18 months. Dealing with unique characteristics of fresh and perishable products.
How do we ensure we have all of that insight and granular-level detail organized and managed in a way that it can help, through prediction and insight, to deliver on behalf of the brand, the customer, that level of expectation 100% of the time? Never easy to do. The role that decision intelligence and demand data can play in that is a really strong and quantifiable role in the way that intelligence helps retailers execute.
Guffanti: We’ve mentioned a few times customer expectations, execution, and minimizing failure. It occurs to me that this is a recurring theme heard over and over with retailers. Demand forecasting certainly comes into play here as the glue that holds a lot of this together. John, was that one of the priorities that you felt you needed to get right? What was your journey like to optimize your demand forecasting business?
Hughes: It was double-headed. What I didn't want was to create loads of waste because you can delight a customer by having all the stock and over-ordering, but that can be wasteful. In a purpose-driven organization like ourselves, we didn't want to be doing that. We want to reduce food waste, reduce plastic waste, we're not about that. It was perfectly set with our sustainability roadmap to introduce. We're a growing business, so there are multiple applications for decision intelligence and for AI and machine learning that we could adopt. We could adopt it in our marketing algorithms, in our operational algorithms, in our rooting algorithms, etc.
For the moment, they're not as aligned with the simple principle of a purpose-driven organization. It was from that point that we wanted to tackle it and say, we can reduce food waste this way by ordering sensibly, we can hopefully reduce plastic usage by not disappointing customers and keeping them on that journey.
To point out, we do sometimes disappoint customers. You've used the word “never” a couple of times, but in a human process, you will always disappoint. We can't get AI to drive the trucks, unfortunately – not yet, anyway.
Guffanti: I love your statement of you're coming first and foremost as a purpose-driven organization, not producing way too much in order to achieve business objectives. How has that approach forced you, and has it benefited your business in the long-term? Becoming so tight and surgical in your approach, has that had, aside from loyalty from customers, has that approach benefited your business?
Hughes: It has. It's benefited us from the caliber of people that we've managed to hire. For a lot of people, the last 18 months have been tough, and it's made them reassess. Not everybody wants to be earning as much money as possible, the shift has been that I have an acceptable living standard that I've set myself, but what I want to do is give something back.
We get people coming in that are moving across the country to come and work for us because they want to reduce. They see a problem with plastic, or they see that we're a purpose-driven organization. One guy was in France, he basically packed up his bag and moved over to be with us because he was so enamored by that.
Maintaining that purpose, that drive, it's a huge badge of honor. I hope to turn around to my kids when they're old enough and say, "Your Daddy helped do that."
Guffanti: I'm going to ask Tom and Scott to comment on that, doing good by having a purpose-driven business. Are you seeing that across other retailers? For you, how gratifying is it to be able to provide technology solutions that net out into a positive for everyone?
Summerfield: It's really cool to hear The Modern Milkman's commitment to this because it feels like there is a little bit of a tide turning in the world right now. People want to take this a lot more seriously and actually try to do something meaningful – that might impact it. Classic retailers are impatient. Typically, business stuff needs to tie back to revenue because that's how people get paid. Maybe there's shareholders that need appeasing, that sort of thing. Where we are now, we've got a technology-enabler that can do both. It's not a trick, you can make money from it if that's one objective, but this whole sustainability conversation.
It is possible that there are some cool business outputs as well, which makes it easier. Over the past few years it's created some cynicism around the sustainability piece. It’s gone through: it's not here yet, it's not here yet…it's definitely, definitely here. Since we have a tech-enabler that can do both, businesses can do good and be prosperous commercially as well. Right now, that’s a necessary balance. The Modern Milkman is pioneering in the approach of its purpose first, others have not taken such a brave step yet because they need it to do both. From our perspective, it's the tech-enabler that allows both conversations to happen, which seems to be the balance that the wider world needs.
Landoc: We have seen a rapid growth in the belief systems behind sustainability, or as we often call it, ethical commerce. It’s gone from a lip service perspective to one in which it's increasingly foundational, often as it is at The Modern Milkman, cultural at its core. That doesn't relieve us, even when it is cultural, of the need to measure. You still need to understand and quantify whether you're delivering on that promise, that brand promise, that sustainability ethos. Again, data plays a role in that. It's not the panacea, it doesn't solve it, but it allows us to say, "We're not just saying it, we're doing it, we're measuring it, and we're improving it."
We're not going to execute 100% of the time. Decision intelligence is helpful in getting us to focus on what's not working. If we can get that insight quicker, that demand, or that business signal a little faster to be our cultural core, whether ethical commerce, customer and brand experience, then we're doing better.
We're differentiating ourselves, and we mean it when we say, “Sustainability and ethical commerce is at the core of what we want our brand to be.” We see that across a growing number of retailers. It's not universal, but it is definitely accelerating.
Hughes: To bring this full circle, we can't be a sustainable company if we don't have longevity. For that, we need financial sustainability as well, which is where the real applications of AI come in. It doesn’t do any good if we burn out in two years because we weren't financially sustainable. I've seen other companies in the U.K. in a similar space do that, so it's important to us that sustainability and financial sustainability are hand-in-hand.
Guffanti: That's a great point. It sounds like the stars aligned, or that the timing is right for you all to work together and make a difference. John, you mentioned earlier in your career, you were ahead of your time and not being heard by the wider organization. Finally, you're in a situation where your insights match with the larger organization, be it from the CEO all the way down. Scott and Tom, you're coming together at this time to help John on his journey. It seems like everything's aligning.
Scott, if you can comment on that sort of cultural, people, internal alignment. Is there a one size fits all, is there a roadmap to be able to work internally with retail organizations in order to have success?
Landoc: Well, there's certainly not a one size fits all. However, there is a prioritized need for there to be the classic top-down commitment to the role that data and intelligence is going to play in our role. I've been doing this long enough to have seen the early iterations of what we used to simply call algorithms or early predictive technologies. The information seemed solid, but the reception among operational colleagues and those that needed to make use of the data wasn't there.
We need to have these platforms that John is using, and Tom and Peak are building to have that. These are classic buzzwords, but that flexibility and agility to adapt to the operational processes without compromising the need for prediction and granular level insight, those are the proverbial knobs that can be turned on the business as that data is seen.
For us, the most common thread in success for projects involving AIML, is that commitment to execute, the willingness to test, experiment, and adjust as results are seen to make sure users and stakeholders are involved in the process. Again, it's not just customer experience, it's our colleagues' experience in using the technology as well. As long as you maintain the combination of commitment and flexibility, there is a greater level of success. You have to be willing to see that some things aren't working, but be quickly able to adapt as necessary.
Guffanti: Thank you. Tom, anything to add to that?
Summerfield: Scott summed up a lot of the key elements of it, but I always come back to the word “transformation.” At the heart of this, it is transformation and transformation can be tricky, it can be hard. All of us on the call today probably have a few battle scars from transformation efforts that didn't go quite as planned. There's hurdles in the way. These can be people, process, systems, budgets, you name it.
Starting small and with a purpose, then having leadership, like John and his peers are pushing and banging in the drum for it, then that can trump some little hurdles sometimes.
It's valuable to have leadership that gets it. There can be varying degrees of how to achieve that because it often needs value attached to it. Then, sometimes it can be fun seeing who wants to attach themselves to it, once there's a few numbers up on the board.
Again, The Modern Milkman, there's a healthy balance of, "No, no, no, there's enough intelligent people here who get this, and we're going over there." Then, there's a balance of commercials that underpin it – it feels like a good balance, but you need leadership to be driving it.
Guffanti: We always hear that it's so important when we're tackling digital transformation and AI initiatives to have some early wins. Early wins that can be measured and proven, and then get other people on board that want to participate. John, on your side, what were some of those early wins that you knew you were heading in the right direction, that kind of thing?
Hughes: The early win was actually a victory for me, in that I didn't get pushed on the time to implement. What you can do is go for the quick win or go for the right win. I pushed back to my exec team and said, "Look, this is why this is what we're going to tackle first, and this is why."
We started all of the forecasting, and actually it's easy to do on a product like milk because milk is highly subscribed to. Whereas some of the other products are more related to shopping, so they change during the week. A product like croissants sells a lot more on the weekend because people want to treat themselves essentially, whereas milk is a staple.
Predicting that is reasonably easy. We tagged the milk first, and they asked, "Why are you doing that? That's the easiest one to do." The reason that we did that is because at the minute, 18 different people ordered the milk. It's part of the daily role, so we have 18 different points of potential failure. By setting up the forecasting to focus on milk first, we can automate the output, and therefore, milk will be ordered from one place. We go from 18 potential points of failure, to one. That was key in that – from my point of view – the quick win was actually the long win. I don't believe in quick wins; doing that patches over some things.
Guffanti: Scott, do you see that being consistent across the retailers that you work with? The need to score some quick wins that point to a larger success?
Landoc: I don’t want to challenge John, but I do understand the point of quick wins, or rather quick wins with a purpose. Quick wins that help quantify that you're on the right path. It can be the smallest, most seemingly inconsequential area that can throw the desire for automation off kilter. Those can get exposed in smaller proof of concepts or early pilots. Again, cloud 101 is the ability to turn up capabilities very quickly, and quantify whether it's working based on our strategy, our plan, on technologies like Peak's. If it isn't, we move on to something else; if it is, we've created a framework on which we can scale it up.
As John mentioned, move rapidly toward automating from 18 buying sources to one. So again, I think quick wins may not be the right language, but quick quantification that the strategy that we have for demand intelligence is right, maybe with some adjustments. Those things that aren't working, we're turning off, things that are working, we're able to scale up.
Guffanti: Fantastic, thank you. John, I'm curious to hear your take. So, what's next for you? As you look at your business, you look where you are right now, where do you see this effort going forward for The Modern Milkman?
Hughes: The next four or five months will be about U.K. expansion, getting as far across the U.K. as we can, with a plot to come and see you guys in the U.S., hopefully. There’s no timeline on that just yet. Also, not to be confused with The Modern Milkman in Cincinnati, a guy called Seth, we're always clogging up his SEO.
Then, within the business itself, developing a bigger range and potentially separating, thinking about C2 emissions, are there any journeys other people are doing that we can potentially take over, consolidate that journey so that three different people aren't visiting a house on one day, and it's just one? That's an avenue that we consider.
In terms of AI application, it can be endless. There’s lots of places I want to go. I want to run Peak ragged, I've given the CEO many of my wants and desires. When I get a hold of him next week, I'll be asking him for more stuff.
Guffanti: Tom, similarly, you oversee all retail, where do you see next steps happening with your customers beyond the initial vision and strategy? Where is the expansion happening?
Summerfield: Like John touched on, quick wins – it's not a destination that you reach. That's some of the problem, people think, "All right, that works. Now we'll just stop there, that'll do." It could continue to evolve forever. The hottest chat in retail right now is, how are you, and that’s not isolated to retail. How are you connecting your supply chain up to your customers? That is really easy to say, but really hard to do.
If you've got a grasp of your demand forecasting across the entire stock file, inventory base, that’s great because that's better than most. Then, where do you take it and start to demand shape by connecting it to the customer piece? All of this, for The Modern Milkman, will be underpinned by what John touched on CO2 – there will always be that purpose to it. Connecting that in a more optimal way, which is the secret sauce of our businesses, we will win big time over the next decade and beyond. Potentially, how once you've got a grip of your demand, or what feels like a grip, then how do you start linking it to customers? That's where we move.
Guffanti: This wouldn't be a modern or a present day conversation without bringing up COVID and the pandemic. Retailers note that the one thing that COVID uncovered was that the supply chain is essentially broken. Scott, can you talk about using AI and machine learning for decision intelligence. There's a lot of opportunity for us to improve upon that.
Landoc: Sure, Tom would agree since we both work with retailers and CPG suppliers. The big parts of the overall supply chain were exposed as challenging through the pandemic. We've consistently seen a want and need to improve the sharing of insight and intelligence between the retailer side and the broader, increasingly large supplier side that was amplified without question, because of the supply chain challenges retailers have seen.
What we've seen is an increased willingness for what was traditionally the sharing of transactional sales information as part of a data collaboration strategy, becoming more intelligence-driven insights. Suppliers want to participate in a retailer's decision and demand intelligent strategy. Retailers want to get visibility and transparency into their supply networks, so that inventory planning, assortment and product mix planning can be better optimized.
That’s happening. The good news is, it's happening. Not as quickly as some might have thought it would, but the foundational role of demand intelligence and the machine learning capabilities that both retailers and suppliers can bring to that next level of collaboration we'll start to see the value being driven out of that. It would have happened anyway without COVID, COVID's just accelerating it.
Guffanti: John, as the retailer in the room, do you want to comment on that?
Hughes: Absolutely. For us, suppliers are partners. We've grown from being a small company, working with small farms, and we didn't want to lose that. Part of the tradition of the milk round is that the milk is from your local area. We've always wanted our partners to grow with us.
In some cases, it's not possible because they don't want to take the risk of putting in a new bottling facility, or they don't want to drive to Bolton three times a week – whatever it may be. We understand that, but the way that the supply chains were built before us, suppliers were there to be cannibalized. A lot of supermarket behaviors are around driving down price, basically pulling orders, etc.
We didn't want to do that. We paid a fairer price for our milk. It's a purpose-driven organization, I’m not just here to make money, I'm here to hopefully do some good. For us suppliers are partners, and part of that is being open with information, going back to them and saying, "we need you to go from this many pint of milk a week to this many." That's scary, isn't it? Especially if this guy is in a family business.
The first time we met one of our suppliers up in Yorkshire we were trading out of the back of a refrigerated van that was stationed on a caravan site. We had a tiny little fridge space, which was no more than five meters squared. The prime minister of the U.K. did visit that fridge, if you ever want to go and search for it, it was on election day in 2018. He actually visited that fridge, it was tiny, absolutely tiny.
In the partnership, they had to invest in us as a partner at that point. Since then we've repaid it, or we're trying to repay it anyway. I'm sure we still have difficult periods with suppliers – it’s human relationships at the end of the day, you have your ups and your downs. I hope that the big thing that comes out of this supply chain being messed up is that suppliers are partners.
Guffanti: That was a great insight. Tom, how important is it to get your data in order before you can embark on an AI or ML journey?
Summerfield: This one always comes up, and it's a fair question. I don't know if I could get the data to prove this, but nearly everyone apologizes for the state of their data before they show it to you. There's nothing we haven't seen, is the best way of putting it, at Peak. You can't let that be a roadblock because there will always be a reason not to try and do some transformation.
If you look hard enough, there's always a reason. We haven't found it to be a barrier with any of our customers so far. People have tons of data, some people have more, some people have less, some people have data they can't get access to. There's always ways around certain challenges. You can even consider some third party data sets sometimes – there's ways around it. I'm not satisfied with that as a reason to not start, basically. With limited data, there can be amazingly valuable outputs to be had.
Guffanti: Scott, anything to add to that?
Landoc: I couldn't agree with Tom more. Improvements in data visibility, data aggregation, and data quality will always make it easier. That state of data will paralyze the start of, or the mindset of, people wanting to move toward ML. The fear that if they don't get it all right, they won't see results. But there is data available, even if it's in 17 different places – which is the average number of data sources we see – retailers are trying to figure out how to get a handle on.
Within those disparate sources is valuable data that can quickly show the quantification and optimization that machine learning and demand intelligence can provide. It should not preclude starting. It doesn't mean that you need to disconnect from having a macro data strategy, from working on the improvement of your data foundation, but it should never preclude you from getting going and seeing those sort of early value deliverables.
Guffanti: John, outside of the customer experience and results, how has this impacted your operations?
Hughes: It's still being vetted, so the operation is always a moving feast. We're opening up new hubs all the time, learning new things, and bringing on new people. How it's impacted operations is by giving us a better idea of what is the ideal warehouse, how much space do we need?
Obviously, if we can optimize inventory management, we don't need to utilize as much space in a warehouse, and therefore we can stock a bigger range. It's opening that up for us within operations. It's also allowing us to eradicate the errors due to stock outs because we didn't order the right amount.
Now when we have a stock-out for a customer, it's likely that there is either a transport issue with the supplier, or we put the wrong stuff on the van. We can start to pinpoint those two things. We're working on the data validation around suppliers, using ERPs, etc. Then, we're also looking at how we operate the yards, making sure the exact amount of stock is on that van when it happens.
Guffanti: One closing question to leave everyone with. We have retailers on this webinar that are saying, "I have these business objectives, these challenges, I want to start my AI journey, I want to move the needle within my organization.” What's the first thing they need to do? How do they start this?"
Summerfield: There's a bit of misconception about what AI could be doing for people. We said, a bit tongue-in-cheek, earlier, "We could do some crazy flying robots, augmented mirrors in a shop, or something like that." No, no, it’s not that. As John mentioned, it’s fundamental existing processes, new business that everyone understands and gets, and can be optimized with AI. This is not criticism, but some people think we're going to need some big ideation sessions of where to start, and things.
There's three or four fundamental pillars of business that AI can improve, and it's about being ready to learn. I used to be on the customer side, that's how we got involved. It was as if we had a direction of travel, we were interested in personalization as it happens, but we were ready to learn about it as well. A little bit of bravery is involved in even reaching out. Fundamental business challenges are a good place to start, not creating new or interesting ones. Tackle the big stuff that already exists, it’ll show you where to think.
Guffanti: Scott, your perspective on where to start and how to find out more?
Landoc: Well, I had a very simple answer, which was just to call Tom. That is the right approach because the view that he just provided is very much aligned to the reason why Peak is such a good AWS partner. I've lived this journey through my old CIO lens, through a consultancy lens, through a vendor lens.
There are basic areas to start – the mileage will vary, the business priorities will vary from retailer to retailer, but there are always early quantifications. Again, the one example I've seen proliferate among groceries and fast-moving consumer goods retailers, is why do we still struggle so much with stock outs? Why is inventory optimization without over-buying still such a challenge?
There’s many examples like that – around pricing, assortment, fresh and perishable metrics. There are fundamental things that can quickly become part of the early stages of the journey. What has been consistent, not just in my role here at AWS or working with partners like Peak, but over many years I’ve seen early wins create energy, enthusiasm, and greater levels of commitment to see what else we can do. Meanwhile, machine learning is getting better, more accurate, and easier to use. Find things that will show quantifiable value for your business, start there and the rest will follow.
Guffanti: I want to thank you all for participating on this webinar, and taking time out of your day. It was a fascinating case study. John, I want to thank you very much for sharing your story with us and congratulations on not only how you evolved your business, but how you're doing good in the world in the process. Looking forward to seeing you scale that mission.
Tom and Scott, thank you so much for adding insight and expertise to this webinar. It was a fascinating discussion, and this discussion will continue because it is an evolving topic. Thank you, gentlemen.
That is all the time we have for today. I hope you found it productive and valuable. Have a great day, and be safe. Thank you very much.