Buy or build? The decision you can't afford to get wrong

The technology was never the hard part. I'll give you an example. When one of our first use cases was deploying a software engineering productivity tool, GitHub Copilot, it was very clear. The use cases was clear, the technology was sound. But when we first deployed it, even after a month or two months, we thought that everyone would adopt it. And the adoption was actually quite low. We really tried to understand why wasn't people, why weren't the engineers really adopting it. When we really looked under the covers, we realized that nobody had the best practices. They didn't have the tool. It wasn't really applied to their everyday work. The tool worked. We just, as an organization, really wasn't ready to receive it. So that's really a pattern I've seen everywhere. That innovation doesn't stall because the teams aren't capable. It stalls when we treat actually the launch as the destination. And we rebuilt the approach entirely. We put together a change management plan, we brought in speakers, a champions program, and we started to see that the slack channels and people sharing use cases started to fill up unprompted. The tool stopped being just something that we deployed and something that people actually used every day. Welcome to In Other Words, the podcast from Frase, where we speak with leaders shaping how global businesses grow, scale and operate. Today's guest is Elaine Barsoom, former Nike tech innovation and partnerships leader and now venture partner at Silicon Foundry. At Nike, Elaine led AI Centers of excellence and strategic partnerships across marketing, service, engineering and product teams. Her experience offers a massively rare inside view of what it takes to scale innovation inside one of the most visible global brands. And in this episode, we're going to examine what changes when AI becomes part of everyday operations, how partnerships and ecosystems make it easier for customers to engage, and why build and buy decisions matter more today probably than ever. So, Elaine, hi. It's great to have you on the show. Hi, how are you? I'm great and thanks for joining us. So let's get straight into the first question. And you've worked in global leadership roles across many enterprise brands and digital ventures and various different roles and partnerships and ecosystems. But for people listening to the podcast today, who may not know your background, where did you start and how did you get to where you are today? I didn't start in AI. I started watching things break. My father was an entrepreneur in multiple companies, multiple careers, and I watched him build things from nothing. And I also watched things fall apart as well. So what always stayed with me wasn't the ones that succeeded. It was understanding why the others didn't succeed. And it was never failed for lack of a good idea. It failed somewhere in the translation, I would say between what he saw and what the people actually around him could actually do. And that pattern really followed me everywhere. I went to college in grad school thinking I'd be working in international affairs and international development. And I was really drawn to complexity, different actors, problems that really didn't have clean answers. And somehow my career really followed that same shape, just in different rooms. And I kept getting brought in when things were hard, whether to build a new company like 11 Games or New Market. And what I understood, what it actually was, that I was the translator between what the CEO meant and what the team could actually do, between the vision on the slide and what someone has to actually make on that Tuesday afternoon. And that was really the work, not the abstract strategy, but translation into action. And I've learned that builders don't fail because they lack ideas. It's when the translation breaks and when that strategy really never becomes action. I love that phrase that you said earlier, problems without clean answers. I think that's a really good way of thinking about that role, of interpreting business outcomes that people want. And then how do we actually get there using what we have? So let's look a little bit at your leadership time in your role at Nike. I think I say in a very British way. I apologize to any international listeners, but Nike, let's call it. What did you find most challenging about making that kind of innovation work consistently inside what is essentially a global brand with all its process and all its systems? So what was the most challenging aspect? It's a great question. I'll first start by saying the technology was never the hard part. I'll give you an example. When one of our first use cases was deploying a software engineering productivity tool, GitHub Copilot, it was very clear. The use cases was clear, the technology was sound. But when we first deployed it, even after a month or two months, we thought that everyone would adopt it. And the adoption was actually quite low. And we really tried to understand why wasn't people, why weren't the engineers really adopting it. So when we really looked under the covers, we realized that nobody had the best practices. They didn't have the tools, they wasn't really applied to their everyday work. The tool worked. We just, as an organization really wasn't ready to receive it. So that's really a pattern I've seen everywhere. That innovation doesn't stall because the teams aren't capable. It stalls when we treat actually the launch as the destination. And we rebuilt the approach entirely, we put together a change management plan, we brought in speakers champions program and we started to see that the slack channels and people sharing use cases started to fill up unprompted. The tool stopped being just something that we deployed and something that people actually used every day. That's fantastic. That's when you know you've cracked it. If people are just. It's just no one talks about it too much. It's just happening and going in the background. One of the things that I realized this change is really uncomfortable. It's really hard for organizations. The technology is rarely the hard part. And I know I've said that, but getting people to move through that discomfort together, that's really the work there. So I think it's the people part of the. What is the old triangle was always technology. People process those kind of things. You've got to have all those kind of working in order for something to be adopted. But moving into when you're innovating and as brands grow in that global sense. You talked about one specific example, but there's an expectation, I think even more so today with AI tools and everything that each market or each function should be building its solutions and innovating. How do you avoid them all overlapping each other with the same problem? Especially, especially when you want to create experiences for customers and consumers that are all consistent and don't fall over each other. Oh, I'm going to help the customer this way. And on off that ad. A proven model still needs translation across markets. And I'll just give an example even we launched a joint venture when I was at American Express. It was. We brought in a $2 billion proven model from France and Europe to the U.S. so the playbook existed, the brand relationships were there. We thought that this would immediately we have traction here. But what we realized is that the US was just entirely different market, more sales driven, noisier. Brand relationships that were in Europe required a completely different approach here. And so it wasn't translated for the right market. And so the relevance of that is that there's just a real tension in that question because you want consistency across the board and local but you also want the local relevance. And when no leader defines what the brand standard is, what's unique to each of the local markets and what's brand standard, you end up in fragmentation. And so fragmentation happens when no one defines which is which. And you always find out when the customer does right and the customer is not buying. So I have found that the best leaders actually define what's the brand standard, what's uniform, what's that trust and governance boundary. And then they give freedom on everything else, what can be unique to each of the markets. And it's about clarity. These are very different things. That's a very valid point, right? Is that you need to have some kind of central way of talking about the business or the brand or what you do and your purpose and the expectations. But you've got to leave a little room for flex in the regions because certain things just won't work. We all know this, we all have great examples where different things don't work culturally. I don't just mean messages, I mean systems can actually not work because people don't want to use it in that way or can't use that way or don't have the infrastructure to use it. So it's quite interesting when you do it the reverse way. I also thought it was interesting that it was American Express that were applying things that have been developed in Europe back into America. That's quite interesting because that's not the way you might expect it, right? No, it's not. And Europe is not a sales market. It was a market, it was a model that grew very quickly with all the E commerce and sales driven culture here in the US it's very different. Apply a flash sales model here. So definitely a learning experience. You've worked as well closely with internal teams, startups, platforms. What do you think when large organizations, what do they often misunderstand about the value of working with lots of different partners across ecosystems and the globe and whatever else? That's a great question that I invest a lot of time over my career investing in. Most companies think that partnerships are about access. It's a new capability, new territory, new promise. And the first part is actually true. When Nike acquired Artifact during the peak of digital collectibles moments and it was really successful on its own terms, brand expanded into a space that it didn't know at a pivotal cultural moment. But what the experience taught me about partnership and transformation is when you bring in something that operates on a completely different model, different culture, different cadence, at some point you're not operating one company anymore, you're actually operating two companies and manage those scenes is a full time job. The partnership was sound, it made sense during the time, but the coherence was really the tougher problem. And so we often tend to underestimate what does it take to maintain that coherence when the excitement fades? No, seems it's the operating model that has to function right after the announcement. So I always say that the real work starts after the deal assigned, every time, without the exception. So partnerships fail when they're treated like transactions, when they're treated like operating systems. I think that's a really important point, isn't it? The partnerships. It's almost like partnerships aren't just one and done. They are an investment that continues across that entire life cycle of a business. When you think about that as A, and you're building ecosystems with various partners, M and A, whatever else, and it's much more a strategic capability set you're building, how should brands think about it as a strategic capability rather than this collections of individual partnerships that people have? Yes, the shift is actually in the question you start with. Most companies ask, who can we partner with? And that just produces a list. They get managed, they get reviewed, they get renewed after time, but they don't compound. And so strategic capability, I believe, starts with a very different questions. Where do we need help to win? And this single shift changes everything. Where do we need help to maintain or build our competitive advantage? So you stop evaluating partners in this isolation and you start actually designing a capability and how they fit together, where one partner's output becomes another's input, where those scenes completely disappear before the customer even sees the system. So the test I use is actually pretty simple. Can your team navigate the ecosystem without a map? If they need of math, it's not a capability. It's still a list with a story. So when it actually works, teams stop thinking about the entire ecosystem and they just start moving faster. It's about the workflows. That's invisible infrastructure and that's really the goal that you want to have. I love that sort of invisible idea. It's very good. So imagine, you know, we talked a little bit about you're building an infrastructure, but I imagine that only gets harder as you're starting to operate across regions, functions and all these different languages that crop up in your experience to leaders ensure that those partnerships make the business easier to engage with as a customer rather than just like, oh God, like you said, it's got to be invisible. But how does that manifest itself to what leaders need to do? Yeah, absolutely. Well, customers don't experience our ecosystem. They experience the moment. Think about the last time that you had a genuinely seamless experience with a brand. You didn't think about what vendor handles the logistics, or what partner ran the local compliance, or what platform process the translation. It just felt like it worked. It was Seamless. And when it's working, it disappears. That's the invisibility. We've seen this publicly, healthcare.gov, it was a classic example of multiple capable vendors throughout our public, each responsible for something different. But there was no one responsible for the full ecosystem. And when it launched, the scenes were very visible to the customer. So customers don't care about which contractor, they care about the friction. And when that complexity leaks to the customer, when they can feel those scenes, then you've done a disservice to the customer. And that's not technology, that's leadership. It's great ecosystems just disappear in use. That's not a metaphor. That should be the standard for customers. And then from your perspective as a person leading that charge, as it were, how did you kind of navigate your way around the business leaders that you had to influence? How did you work that? That's a great question. I always believe in diagnosing the problem first and strong alignment early on. Don't initiate a partnership down the line and then have to get the sale in. So transparency, alignment, start with the diagnostic, right? Do the work early on. Because if you do the work early on, you get that alignment. It'll be much more seamless down the line to your customer and across the board, easier to implement as navigate. Love it. We're what we call our mid show moment where I ask you a question that's not quite so. So worky and we think about what's one task, either professional or personal, doesn't matter which. You wish you could automate. It seems simple because we're almost there. But I'd automate my travel, the complexity of my travel. And when I say my travel, I'm incredibly detailed. Timing, connections, what flights, what airlines. And so I want, I would love an AI that just knows me, that's a travel companion that actually paid attention. When do I need to rest? When time. How much friction can I tolerate? When do I need margin? And not just a booking tool, not just one that can help me book everything, but one that really knows me and inside out. I love that. It's like a travel advice person. You need basically a travel companion, let's call it exactly. Someone tell me to rest now. You need rest now. You're good. Yeah. You're getting grouchy, Too tired. You need to rest. I love that. Thanks for that. Let's move into sort of a bit more of a topic of the day. You know, AI. And in many ways I think AI has made, you know, it feel almost deceptively. Simple to people on the surface level to build new things and new capabilities. And from your experience, where does that sense of simplicity break? Once organizations try to run things themselves at scale, AI feels simple until it meets your infrastructure. Internally, we evaluated an Israeli Martech company and it was an incredible technology. Dynamically optimized and accelerated product videos, e commerce sites, really exciting, reducing load times, had a number of different use cases. Technology was really impressive. But when we started mapping need and creation, we just found that between our, our tech stack and our existing cdn, it was just adding complexity to an infrastructure that was already carrying weight and we decided not to move forward. But the technology, it was just a great example of the technology being ready, but the workflow underneath that just wasn't ready. And so that's the pattern that I see is that AI and technology, they just feel simple when you look across the board, but when you look at them in isolation, they feel simple and they break when you try to weave them into what already exists. Old infrastructure, legacy systems, existing commitments, workflows that just have been running for years without really taking a close look of how to redesign those end to end ecosystems. And so teams sometimes build something without rethinking what's underneath. All the processes, the new tools, the same habits. If AI actually doesn't really change how you actually work, it won't scale, it'll just stall. It might be running, but it isn't working. So that's really what I have found. So do you think that leadership teams often underestimate the infrastructure that's already running? And what about the idea of one of the things I'm sort of looking at is that you don't understand the long term ownership that comes with building something yourself, let alone buying technology and putting it into your infrastructure. But when you go, okay, we'll develop this. Especially when you've got things like governance, quality, risk underneath the hood. Have you seen that? That they underestimate this idea of ownership across the lifecycle of whatever you've built? Yeah, well, leadership teams underestimates that. Ownership compounds when you build often comes later. Now, not against building, but I sat with teams that are maintaining systems that nobody fully understood anymore. The person who originally had the build decision had moved on years. The architect, the team keeps it running without the context of why don't we build this? And regulation shifted the vendors. The system technically worked. It actually didn't work for what we needed to do now. And so that debt wasn't just in the code, it's actually in the Organization. So when you build, you're not just taking off software, you're taking on every decision that the software touches. You talked about government quality control, the risk management compliance is with AI changing in every markets and regulation. Each of those requires people and expertise and ongoing attention. And here's what's really uncomfortable is that weight doesn't really land on the leaders. It lands on the people who have to maintain it years later. Often I've seen without the context of why was it built that way or in the first place anyway. So when you build, you own that learning curve and forever. And most leadership teams just. We don't budget. They don't budget for forever. Yeah, that's funny, isn't it? The mostly look, you know, a year or two years max out. But yeah, I love that. So when somebody's buying a kind of platform rather than building it all yourself, is that the problem they're trying to solve that isn't necessarily obvious at the start? You're buying for longevity. You're buying because you don't have to maintain, you don't have to almost innovate because someone else is doing that, someone else taking that risk and putting all of their resources into it. Most leaders think that they're vying automation. That coordination we brought on a platform at Nike early on, and it was designed to scale innovation and ideation programs, measuring innovation, roi. And the capability was incredible, but nobody used it. And when we looked, we found that we bought a solution to a problem that, that we didn't actually map out how our innovation decisions got made, who owned them, and how did ideas move from submission to action. We brought automation to a workflow that actually wasn't there. So when leaders think that they're buying automation faster outputs, what they're actually trying to buy is that coordination. How do we get teams to work from the same system instead of stitching them together, less manual work, what they're actually buying. And so the hidden problem is never the capability, it's the fragmentation. A platform just doesn't solve that task. It solves the coherence underneath. Technology won't change organization, it's the design does, the leadership does. Platform is only the beginning of the work. Training does all of these other things that we don't talk about that really is what changes organizations, the workflows. I suppose it can be a catalyst to get those things done. But you've got to be careful about putting the cart before the horse. So implementing the tech before you've got the right process workflow and all of those. So looking back, you've given a couple of examples, but not just in your career, but at other companies. Have you seen examples of where getting that wrong has a high price tag? Getting that decision of building ourselves versus buying something or implementing too early, it's debt. And that's just not technical debt. It's organizational debt. The highest price isn't the money, it's the years. I've watched Organizations spend months, 18 months unwinding Bill decisions that just, it felt obvious at the time are pulling apart technology that had grown into the walls of the business or rebuilding workflows that adopted around a broken system. And that's just not transformation, that's archaeology. So technology stack on technology costs rising, workflows getting harder, not easier. In large companies, unwinding those choices requires real leadership and across multiple functions, a long Runway that most teams don't have right now. And that if you don't envision how the system could operate end to end before you build, you're not really getting transformation, you're just getting an expensive version of what you already were. And that's the real work that needs to be done right now. Especially when integrating AI into systems from decades or from years ago. It's how do you rebuild those systems, how do you redo those workflows, how do you think about the designs from one function to another function to another function? It's a lot about process design as well as people, I think, isn't it? And you've led these initiatives. And so let's say you're putting in, you're getting it, you've got all those things, how do you kind of then go? Because people are now asked much more than ever, well, what's the value of this? What's the ROI of these things when it becomes part of the daily operation, how do you assess whether it's delivering value or how have you done that in the past? What I have learned is that value actually doesn't show up in the dashboards. It shows up in the behavior. I think we were talking earlier about a champions program and GitHub copilot. And when we implemented this Champions program, at a certain point I stopped watching, just like the adoption metrics and I started watching and reading the Slack channels. And when those channels started real filling up with use cases and people sharing and champions program and evangelists instead of trainers, and when people stopped acting, how do I use this to hey, can this solve a particular problem? That was real, real shift. And this tool stopped being just something that we deployed and started being part of actually how work was getting done and how workflows were talking. And people started feeling differently, they started feeling more confident and decisions happened differently when they did. Before the handoffs were cleaner, work felt lighter instead of heavier. That's real adoption. That's really the only signal that matters. It's that behavior. If AI is not changing how that work is getting done and it's not delivering value, might be running, but it's not really working to its fullest capacity. So that's the real work that needs to be done today. I love the idea that it's so pervasive that you don't even need to ask the question about roi. Many years ago somebody said to me, don't ask the ROI of your running shoe. And I think that's a really good analogy because it's something you need and you're going to go barefoot in the street and run. So it's there, it's pervasive. And that's quite a good, I think, way of thinking about it. People stop thinking this is AI and they start, wow, I've freed up my time to do something real high valuable work and whatnot. So just coupling this with one of the themes that I think we're all seeing in the market at the moment. There's a lot of debate about this. Human versus automation versus AI or with AI or where do you think AI helps the most and where should the human kind of stay in the mix from your perspective? And I appreciate things are changing very rapidly at the moment. So these things are not. Not fixed? No. This is actually a question that I'm super passionate about in the human and AI question that I. That's a lot of time, especially spending right now with leaders. I really don't see this as an automation versus human question. The framing is actually the wrong question. What we actually have is the judgment allocation. AI is extraordinary at the pattern recognition, at the scale to speed removing the friction. And it surfaces signals that humans would miss. It handles volume that would exhaust any team. That's real, that's what matters. But where the customer facing work that lives in the unpredictable, the context, the emotion, the moment when someone's situation really just doesn't fit into any category or any pattern. And that's where human judgment isn't optional, it's the product actually. So when we try to automate judgment instead of supporting it, that trust really erodes. And so I really believe that AI should inform decisions, but humans should own the consequences. And the organizations that get this right don't think about where does AI end and where humans begin. They think about what each does best. Where do humans really do best? And they design around that. It's a whole design principle around the humans, around the AI. Then that's fundamentally a change. It's a shift in how a lot of organizations operate today. Are we going to have humans that are managing a set of agents rather than workforce? And what does that look like? And questions around how dependent are we going to be on this technology? But we have to maintain our judgment as humans. And so I really believe that, that it could elevate human skills but not replace. And I think that key thing there that you said is that the accountability doesn't go away. There's still got to be someone who's accountable for what that's delivering. Let's think of a world where it becomes part of everyday operations. And your advice to leaders that are building an ecosystem, they're building their business that delivers results. What's the one bit of advice you would give them when thinking about ecosystems, AI and that kind of technology, innovation, space, Bringing it back to, you know, the beginning that my father built things. And the ones that lasted weren't the ones like with the best opening or the companies weren't. They're the ones that designed for what comes after. The businesses that could really adapt and that got better with use. And so stop thinking in launches and start thinking in waves. I developed a framework called waves, called waves woven into workflows, adaptive value through learning, empowering humans and launches peak. They get attention, they get the resources for that particular moment and then they fade because no one really decides for what comes after. Waves are different. They build, they adapt, they carry things forward. The ecosystems that actually deliver are woven into how work happens every day. And partners embedded into workflows, not boiled on, not just stacked learning that compounds over time, people who feel more capable and not more dependent, as we talked about. So is this ecosystem building a wave? I would ask can it adjust? Is it carrying people forward or are they just swimming around? If the answer is no, you don't really have a strategic ecosystem. You just have a launch that just already peaked. So that's the work I'm really passionate about and focused is helping leaders build in waves, turning experiments into living systems and how that can carry organizations forward, that can turn it into a real strategic capability and not for just this corner, but for what comes after, for thinking about the future of Hearns. And so if anyone's more interested in any of this work, you can find me on LinkedIn. But that's what I would really ask leaders today. I'm not the advice I would give leaders today. Yeah, I love thinking in waves. That's a very, very interesting thought which I'm going to go away and think about for our business as well, which is brilliant. So thank you for that. And look, I think, you know, we're nearing the end of our conversation now and it's been a fascinating one and we're just going to give you a couple of quick fire questions to finish up with. So if you had to describe global growth in one word, what would it be? CO intelligence. Anticipatory. CO intelligence growth that used to reward scale, now rewards humans and systems that are working together before disruption does. Love it. I think co intelligence is one word, I'll give you that. And then a partnership that you admire for how that's been governed, a partnership that you've seen really work really well across business. Yeah. There was a recent one around Christmas. It was targeted in OpenAI. And I say this because the most established company and one of the most forward thinking AI company, they're redesigning the experience from inside out and that's rare and that's really hard to see. So I would say that that particular partnership. Okay, interesting. And then final question, then we can let you go. What's one book that every exec should read in your. In your point of view, from your point of view? I think every exec should read Hidden Potential by Adam Grant. Not just because I went to Wharton, but it's such a clear metaphor. It's about the group that's not obvious yet. The potential that systems and leaders keep walking about the past. And that's the work that I find myself and I think a lot of leaders find themselves doing every day. The non obvious work. Yeah, I love that. The non obvious work. Often the hardest stuff to get your head into, but the character development, the human stuff. Interesting. It's been a fascinating conversation and thank you for sharing your experience with us today. Oh, thank you. This has been such a pleasure. I really enjoyed it. Elaine, thanks again for a great conversation. You shared a very clear and grounded view of how innovation outcomes are shaped by those partnership choices, the execution discipline and what leadership decisions are needed as organizations scale. And that's it for another episode of In Other Words, a podcast from Frase. I've been your host, Jason Hemingway. And a massive thank you again to Elaine Barsoom for joining us today. If this episode made you rethink how your organization approaches ecosystems and execution at scale. Be sure to subscribe on Spotify, Apple Podcasts or your favorite podcast platform.