Name: It's An Add to Cart World: Operating the Modern Ecommerce Machine
Uploaded: 2026-06-02
Duration: 17 min 15 s

The modern e-commerce machine never sleeps—and neither do the operators running it. Managing a global shopping platform today means juggling AI integration, cybersecurity threats, partnerships, and more—all while satisfying customers' increasingly high standards for the speed of commerce. The decisions COOs make cascade across millions of transactions. Adrienne Down Coulson, COO of Rakuten International, and Tom Maguire, Head of Operations at Instacart, will get candid about how they architect resilient operations, make high-stakes decisions at scale, and build the foundation to fulfill what consumers want. 

Adrienne Down Coulson, Chief Operating Officer, Rakuten International
Tom Maguire, Vice President, Head of Operations, Instacart
Moderator: Kristin Stoller, Editorial Director, Fortune; Co-Chair, Fortune COO Summit

It's An Add to Cart World: Operating the Modern Ecommerce Machine

Conferences

Adrian, Tom, thank you so much for being here with me. Thank you for having me. So I wanna start with a, a, uh, bit of a, a big question so you can take a second to think, but what is something that you used to believe about running an e-commerce operation or a global operation that AI has just completely changed your mind about and changed the way you think? I need whoever wants to answer first. I, I'm, I'm happy to go first. Go ahead, Adrian. We have talked so much about AI and we're gonna keep doing it. Yeah, we hold on to your hat, um, so I think. You know, for me, um, I have been thinking a lot about this idea of like deep organizational real transformational change as opposed to these pockets of small bits of improvements and efficiency that we all see in our organizations, um, and I think the thing that really stands out to me is the role of the C-suite. That and that's not something I think I had really thought about before. We've lived for what decades, I don't know how long with the operating system that we have today at the C-suite level and it was built in a time for a for a different time in a time that required a different pace of change, a different pace of business, and, uh, and our leader. have spent, we all in this room have spent decades our entire careers building, I think, uh, you know, a career based on deep subject matter expertise in our functional areas and that is not what this moment requires, right? AI doesn't discriminate between one function and another and one silo and another it's across the organization horizontal. And uh and so I think that's what this moment calls for is an executive in a C-suite that is structured a little bit differently. Those are hard questions to ask ourselves, I think, um, but structured differently and, uh, and, and requires a sort of, I think, cross enterprise judgment and opportunity to think across the enterprise that we haven't all trained ourselves to do. Yeah, structure is definitely one we've talked about a lot these past 24 hours, Tom, is there. Anything from you as well. I, I, I love that. I think for what I've observed most recently is AI is only so powerful as the context around it. As an operations individual and and and proprietor, I think one of the biggest things I've been encouraging my team to do is not only look at AI as the abstraction, you know, as a layer of looking at averages of pulling up the data, it's also about understanding what's happening in the real world. Um, this is an old operations philosophy around walking the Gembba. I'm sure many of you are, are familiar with that, but it is a, it is a principle that, that we hold true in, in my function, and what it means is being in the real place, right, the place of value. And so we encourage all of our teams to be shopping all the time, acting as Instacart shoppers, placing real orders. We listen or take real. Customer shopper, retailer calls um just to understand not only what the data is telling us but living that real world because it's only so powerful as if you're actually playing in that sandbox every day. Makes sense. Now I wanna talk about some of those real world examples we love a use case here we love an example. Adrian, let's start with you. What is the problem that you are most obsessed with solving right now that has just been. Grinding your gears and you cannot fully solve it and how are you working to do that? Yeah, I think it does go back to that idea of all of those, all of our silos and where they bump up against each other and where they sort of blend at the seams, um. I, I can give you one example of something that one of our business units did in the last few months that I think really got at solving some of that problem, not maybe necessarily enterprise-wide, but, um, but definitely impacted multiple parts of the organization. So Racketton has like more than 70 businesses around the world in many different industries, but the one that most people in the US know the best is a cashback shopping. Platform that we call you thank you a lot of cash back over here so we call it Rackets and Rewards internally and that team built a conversational agent called Annelise that sits it lives in Slack and sits on top of all of our data sources across all of our 17 million customers and members and, uh, and that landscape across that organization was super fragmented not just from. Tool set perspective but also from an organizational structure perspective so we had a business intelligence team inside marketing we had analysts inside engineering we had analysts inside commercial we had them in finance and they all had a different idea of the source of you know their their one single truth and maybe that's maybe that's something you guys can relate to too but super fragmented. And it required each of the leaders of those teams to actually come together and redesign a different experience. So now we have every single employee in the company can go to the Slack channel and ask a question. We can somebody could ask like Instacart is on our platform. So if you see, um, sales trends moving in a different direction for this week year over year, and you could wonder why. Now you can ask that question instead of sort of getting in a queue with an. Analyst and waiting for them to write custom code and coming back to you with a back and forth and that used to take weeks. Now it takes 20 seconds so it's that that um efficiency you get but you also are getting that like cross department kind of, um, collaboration and redesign of how work actually gets done and how the service gets delivered to employees. Makes sense great examples, Tom, you got one for me. Yeah, I think you know one of the, the toughest sort of challenges we have is the element of how the physical world plays in the digital world. Some of the most difficult operational problems now are that at that intersection where, you know, we are living in both places. We're living in, in you're on your phone ordering uh a real order whether you're ordering, you know, a birthday cake for your, your kid's birthday or you're ordering a Christmas tree around Christmas time. You're living in that digital world, but the reality is the. Operational complexity underneath that is one of our hardest challenges because AI at currently is, is, uh, raising the bar. It's making it, uh, customers have a set of expectations around speed around uh getting personalization even better and and stronger and so for us a huge focus of my role is making sure that the customer doesn't see that operational complex that real world. Complexity even in times of sort of acute severe issues you can take something like weather. Um, it's no secret that our demand spikes in weather moments, uh, hopefully with no weather moments currently, uh, in, in the Scottsdale era, although like actually a high heat is a real thing for us like when people are in like right now people will stock up when it's cooler in the morning. Um, but what we have to do is make sure that we're building the orchestration behind that, navigating, helping shoppers navigate in the store, you know, we're right now we've been working through deploying different solutions for shoppers around, making sure they can make the right decision points in the store. Dozens of times during a normal shop they're having to make decisions around your order, whether it is an out of stock replacement. We have to use, we use AI, uh, different systems in order to surface to that chopper what the best replacement will be. And it's not just the thing next to it in the aisle. It is using your personalized what you really want in order to give it to them. And that's a way more complex problem that it seems at the surface and going layer by layer is a is a huge focus for us. That makes sense. I'm going to turn to the room in just a second for a question, so please raise your hand if you have one. Someone will bring a mic over to you, but Tom, follow up question for you, because one of the things I've been, you know, interested about from the Instacart perspective is you've really been experimenting with AI in a lot of different ways. Sometimes it works, sometimes it fails, and you're, you talk about that a lot. I know the most recent news is Instacart is now available in cloud, um, allowing customers to build a grocery cart right from the conversation. Very cool new tech. Last year, another. Instacart experiment with um AI pricing experience with it with dynamic pricing of grocery retailers didn't go as well as planned. Can you talk about what you've learned from, you know, past experiments that you've brought into this, this next wave of experimenting with AI? Yeah, thank you for the question. On, on the latter part, I just want to, uh, clarify one point. Instacart actually doesn't set prices, retailers set prices. So, um, just want to clarify that that piece. On, you know what things go well, I will say a lot of things I think we've all experienced the things that don't go well, right? It's like, oh my gosh, we're pushing for an adoption of a tool but actually haven't sold to our teams that actually that there's a real reason why that adoption is important. It's not just using the tool, it's about actually deploying it in a, in, in a way that's effective for the internal teams. I think the thing that has worked really well on the operational side has been. Um, things that are net new versus only trying to solve for efficiency, I think we all can focus on how do we make things faster, how do we use less resource time to do something, but the things that are net new are really the game changers for us. And while I have a ton of examples, one real world. Example is we, um, you know, we have service hours for our operations, our live operations, you know, we're doing millions of orders a day. Something is going to go wrong. Law, law of large numbers, there's going to be something wrong with millions of orders a day. And so we have live operations in the background and making sure things happen. And we needed to make sure that we're actually extending the ability of those teams. It wasn't about being more efficient. It was actually growing the pie as we call it, and the teams launched that, the teams that were closest to the problem, those live operations teams who are junior team members who are like, how do I solve this? And the reason why. Worked was because they were the closest to the problem. They also had pride in doing it. It wasn't something someone told them to do. They said, I want to do this. I want to make it so that my function can do more versus I want to do, do things more efficiently necessarily. Yeah, that makes a lot of sense. Any questions from the room? Sarah. Over here, we have a mic running its way to you. Hey, uh, Sarah Walker, CEO at Slack. I wanna pick on something you talked about at the beginning around the C-suite role is changing and kind of reimagining. We talk a lot about how we're reimagining our team's workflow. I would love to hear a little bit of how you're reimagining kind of the makeup of your C-suite and how you run the business. If any examples or how you're tackling it and approaching it, yeah, so, um, I'll admit there's no solve yet. I have ideas and I'm thinking about it a lot. I have in my role. I have the opportunity to, I oversee multiple business units, so I get the opportunity to kind of work in and around different C-suites. And uh and I it's so fascinating to see how watch how decisions get made where they get stalled where decisions have to get like relitigated and and I think a lot of that connects to trust so I think you know if I think about what's the C-suite. What's missing I think is that enterprise view that enterprise judgment and one of the previous sessions today there's a lot of talk about how COOs are maybe particularly well positioned for that because we already do have that enterprise view in a way that's different maybe than some of the others around the table and I'm not necessarily saying that that's the case, um, but it, it's an observation that is interesting but I think regardless each person around the table I think. We need to start building that muscle and uh and then I think what connects to that is this sort of missing piece around trust and because when trust is thin at that table you get uh you know it's maybe for a couple of reasons is it around competence like you don't trust that the person across the table from you is able to do the job and deliver like like they promised or is it around, you know, intent. Do they really have the interests of the organization in mind over their own and, and what's going on behind the scenes? And when that happens, you get the whole, all the gears start to, uh, slow down, right? And we can't move at the speed that's required. And you get that kind of backchanneling, you get the redecision. And so on, so I'm observing that and thinking about, uh, it, you know, it takes like a serious amount of introspection, right? And it's uncomfortable like when I think about my own role and how I think for me as a COO I have been an absorber of friction. And that's removing it from the other people around the table, but it doesn't get rid of it. It doesn't fix it. And so that's one of the observations I've made about myself anyway and maybe what might need to change. It takes, I think, some courage from all of us to really be that introspective about it. Can I ask a follow up on that? And Tom, I'd, I'd love your take on it too, because Adrian, I know you've talked about, um, reinvent. Mention exhaustion is what you, you've called it and my colleague Nick Lichtenberg last night was telling us about the conversation at his table about, you know, what would you do if you could do anything right now? And the answer was we can't because we're so exhausted from, from all the other things in our daily life. But you know, how do you audit, you know, the organization's capacity for change when everyone is just so bogged down with everything. That they're already doing, how do you reinvent in that scenario and Tommy you can answer this too. Yeah, I think it's, um, one of the things that I've been trying to sort of formulate, it's imperfect right now is the concept of not focusing on iteration but actually focusing on full transformation. It's, you know, we have worked to, you know, quickly iterate on things to solve them. So to your point, like. There's only so many things we can solve in a day, but how do we skip to the end? How do we, instead of encouraging our teams to do one or two steps, let's actually just do all the steps, and I know they can be daunting to your point on the exhaustion piece, but actually it gets way more inspiring. I found for teams to say, Oh, I'm not moving that metric 1 point, I'm moving it 40, and I think, and I will, it will get, I will get that there over time, but I, I think that sort of transformation versus iteration mentality has helped a little bit. Yeah, Adrian, any, any additional thoughts on that? Yeah. I mean we, we talked a little bit about COVID earlier in one of the earlier sessions. I think the same leaders who are being tasked with this reinvention and total transformation today are the same ones in this room and the same ones that led companies through massive restructurings multiple over the last several years, COVID-19. Um, all the geopolitical stress and strife, all of the stress and challenges around DEI, like the list goes on, right, and not a lot of time in between to breathe or since then, so I agree, complete exhaustion, um, and, uh, and I think what the moment calls for and how you audit really starts at the top. It's that introspection. I think that, you know, the companies that win at the end of the day. They're not going to be those that picked one technology over another or smart with those decisions, but really the ones that made the hard calls at that leadership level. Makes sense. And, and, in our last minute together, I'd like to ask you both, yesterday we talked a lot about whether AI should be a colleague. There seems to be a lot of debate about it, and I would love both of your take, you know, 5 years, 10 years in the future, how do you each foresee AI and humans working in tandem or not at each of your companies? Tom, oh, it's a great question. I wish I had like a, a really sharp answer. I think the, the way I think about it now is it's going to be an augment to humans because we for Instacart, uh, food is personal and everyone sort of has their own personal journey with food and so that's what we believe and we will always have this sort of human connection with the AI sort of element elevating that and make it better. Yeah, excellent answer, Adrian. OK, so I think a lot about the back office functions because I've spent the last several years pulling together a centralized set of services across multiple business units, so everything you could imagine all the HR, IT, security, facilities, um, legal, finance, all those things are one central organization now, and I think that that set of services is gonna look completely different in 5 years. And I don't necessarily think that's about displacing jobs, but I do. I'm seeing a little bit of that now like we're, where we're not backfilling for some things as we have natural attrition, but I think the jobs will be different, and it goes back to that idea of kind of those silos across the central services as they bump up against. To each other and those scenes start to blend and we will reinvent work flows across all of those things um I have if you have time for one example I know 20 seconds or less OK no OK we'll we'll ask her after we'll ask her at the lunch because she is going to lunch but Adrian, Tom, thank you so much for joining me and sharing your stories.