BRIAN KENNY: Having been around since 1908, Harvard Business School has had a role in shaping and sharing some of history’s most consequential management concepts. Such was the case in 1958 when a burgeoning Harvard Business Review published an article by Harold J. Leavitt and Thomas Whisler titled, “Management in the Eighties” that featured the first use of the term information technology. Their foresight anticipated the vast influence that information technology would have on business and society in the future, describing IT as a field that would fundamentally alter how organizations processed and managed information. Today it’s almost impossible to enumerate the ways in which digital technology has transformed our approach to managing business sometimes with mixed results. And as we’ve seen throughout history, change is hard.
Today on Cold Call, we welcome professors Iav Bojinov and Edward McFowland III to discuss the case, “Pernod Ricard: Uncorking Digital Transformation.” I’m your host, Brian Kenny and you’re listening to Cold Call on the HBR podcast network. Iav Bojinov’s research focuses on developing novel statistical methodologies to make business experimentation more rigorous, safer, and efficient. Edward McFowland’s research interests lie at the intersection of machine learning, information systems, and management. You are both new to Cold Call. It’s great to have you both here. Welcome.
IAV BOJINOV: Thank you so much.
EDWARD MCFOWLAND: Thank you for having us.
BRIAN KENNY: When I was writing the introduction today, I looked at business fads over time, things like TQM and business process re-engineering, and there’s all these things that if you’ve been around as long as I have, you’ve kind of lived through all these things and they seem like fads. They sort of come and go. And digital transformation is one that has been in the lexicon now for a few years and it doesn’t seem to be slowing down. In fact, it seems to be picking up and it feels like it has staying power. So, anybody who is working anywhere is probably dealing with some aspect of digital transformation. So, thank you for writing the case and for being here to talk about it. I think it’ll be a great conversation.
Edward, let me ask you to start, if you could, by telling us what the central issue is in the case and what your cold call is to start the discussion.
EDWARD MCFOWLAND: The main issue really is at a high level is thinking about how do you take a more traditional organization that has thrived and done well in previous times and bring it to this new century with digital transformation as being one key facet, but also to mention the fact that AI and data science are built on top of that and how do you take that organization and figure out what made it great but not get locked into that and be able to be nimble enough to become a company that can exploit and thrive with AI and data science. To answer your second question, the cold call really, we’re still working through all the kinks of it, but really the main focus, I think how I’m thinking about positioning it is taking students and saying, you’re in the seat now of Christian Porta. You’re thinking about this question and you’ve got something working in one of your key areas and you have other key areas where it’s not working so well.
And so, the key question becomes: why is it not working in this particular area and what do you do to resolve that problem? And I expect students will have all kinds of questions up front, all kinds of thoughts up front, which will be fun. I would get them on the board and get them all down. And then we’re going to get realized as we go through the case that that’s maybe not the central problem. The problem I thinking is technology, it’s some people. It’s like you’re going to really come to realize that it’s really a combination of all these things and that’s what you need in order for this kind of transformation to actually take success.
BRIAN KENNY: Yeah, that’s great. Iav, let me ask you, I’m always curious about why faculty decide to write about a particular organization or a particular firm. Why did you think this was an important case to write about?
IAV BOJINOV: For me, what was really interesting is you’ve seen all these companies focusing on digital transformations over many years. They’ve invested very, very heavily in it. They’ve started to see some returns on it. But for me, what was really interesting was to find a company that was really focused on trying to operationalize AI. And what I mean by that is not just do the digital transformation where you go from having an analog system to having a digital system, which is what a transformation is, but it was really about building AI capabilities and deploying those capabilities and getting their employees to use those. So, it wasn’t just about digital, it was actually about AI. And that was the piece that was really interesting for me. And I’ve written a number of cases that are all really focused on either startups or really digital native companies, companies like LinkedIn, Yelp, Etc. And I wanted to write a case that was more on a traditional company that was both trying to do the digital transformation but then was also thinking about AI.
BRIAN KENNY: And by the way, I appreciated the use of Uncorking Digital Transformation in the title of the case. Tell, for our listeners who don’t know, about Pernod Ricard, can you describe a little bit about their business and where they sit in the landscape?
IAV BOJINOV: Yeah, so you may not have heard of Pernod Ricard, but you have seen their products. They are one of the biggest alcohol producers in the world. So things like Jameson whiskey, Absolut Vodka, a number of really high quality champagne wines, et cetera, they produce all of that. And so when you go into a liquor store, about half of the shelf is made from them, one of their products. And they’re a really traditional company. They’ve been around for over 100 years. It basically was started by producing this sort of aperitif style, very herbally drink that’s very popular in France. And then that was the original version of it. And then over the years it had many acquisitions, et cetera, and it’s grown from there.
BRIAN KENNY: It’s funny, if I think back, we’ve done a lot of cases it seems that are focused on the alcohol industry. I’m not sure what that says about us.
IAV BOJINOV: Look, when they take us into these amazing buildings, their headquarters in France, you have a nice long lunch and, they’re able to pull out a couple of different wine bottles.
BRIAN KENNY: Edward, let me come back to you for a second and ask you. I mentioned the term digital transformation and how it’s been used quite a bit. Can you tell us about what the difference is between digital transformation and AI transformation?
EDWARD MCFOWLAND: I’m going to give you my sense of it because I think we use a lot of terms in practice and in literature, in theory that become synonymous with each other but really I think originally had different meanings and they get muddled up. So to me, as Iav pointed out, the digital transformation part, it’s really about taking your systems from a place where we used to write things down by hand, track things by hand, much more manual and taking the data and information, digitizing it so that it can be stored in systems and processed by systems. That is the transformational aspect of it. The AI and data science facet I think is kind of typically required the digital part, but it’s taken it a step further because now with the digital part, the humans were still responsible for encoding the laws, the logic, the rules into the system, and it would just be processed more automatically by the system. It would execute on those laws, those rules.
Now AI is saying perhaps the way you think you should do this thing or think about this thing or analyze this thing, the human things, you may not be the optimal way or the right way. So now you tell us the objective, not the rules, not the process. Tell us the objective function, and we will then find the AI in the system then figures out what it thinks is the best.
BRIAN KENNY: And this gets a little bit to the change is hard thing that I mentioned in the introduction, because we’re all grappling with what AI means for us in our various roles. And a lot of those people have concerns really about those things, and they know it’s going to make it more difficult. At a place like Pernod Ricard, which has been around for so long, I would imagine this change can be even more complicated because they have grown up in a different way than a digital native firm might have. Can you talk a little bit about how they’re thinking about this?
EDWARD MCFOWLAND: I’ll speak to it from what I was able to glean from our conversations and writing the case. I’m sure they may also have evolved over the time as well. They started with we’re going to try to compete in this market and acquire lots of different brands, as Iav pointed out some of those brands. And at some point as it gets pervasive and the proliferation of brands, it becomes hard for you or I to manage all these things and deliver value for all of them.
And so it became a necessity to have systems, digital systems that could help gather data, collect data and help these humans make better decisions about the brands. And then it became a necessity I think in doing that of well what we think is the best way to market or what we think is the best way to sell may not be the best way to market or sell. And if you have a bunch of different brands that all require attention, it becomes a question of, okay, we need to now extend beyond digital to the AI transformation. Especially in the compilation in a company like Pernod Ricard where they are very spread out. There’s different organizations within the company that have their own P&Ls, their own power. And so you have leadership both at the high level of the organization as well as the individual locations. And that becomes really tricky when you’re decentralized in that way.
BRIAN KENNY: Yeah, you’ve got geographical dispersion, you’ve got cultural differences.
EDWARD MCFOWLAND: Exactly.
BRIAN KENNY: So that’s super interesting. We’ll talk more about that. Iav, let me come back to you and let’s talk about some of the specific tools that they use that are brought up in the case so D-STAR and Matrix. What are those?
IAV BOJINOV: So let me actually pull us back a little bit.
BRIAN KENNY: Okay.
IAV BOJINOV: Because I wanted to speak a little bit about this transformation piece. The thing that’s really interesting and different for me about a digital transformation in an AI transformation is that digital transformation, as Edward explained, you really go from A to B. You go from analog to digital. And presumably once you get to digital then you’re in steady state, and you can go about your life. Now you just use a laptop instead of writing things by hand.
With an AI transformation, what I’ve seen after this case, and especially with some of the other stuff we’re doing now, especially around generative AI, is that’s not true. So no longer do you go from A to B. So if you think about an AI transformation, it almost implies that you go from no AI to AI. But because AI is continuously changing and continuously evolving, there’s no longer a steady state. And so I think one of the big lessons for me from this case is that the notion of an AI transformation, sure we may use that word, but that is no longer a transformation. That is a completely new way of doing work that is continuously evolving, and it’s no longer going to be a stagnated thing. So my hypothesis is that really transformation is kind of a term that should die out because that implies you go from A to B. That is not the case. It’s continuous and ongoing and you’re always having to reimagine what your work is going to look like, especially with some of the latest technologies. And again, you see this in the Pernod Ricard case with D-STAR and with Matrix. D-STAR is a tool that is around figuring out where your sales professionals should focus and where they should visit, which shops, which bars. And then when you get there, it gives you next action recommendations. So they’ll say maybe Absolut Vodka is selling really well, why don’t you offer them our premium rum brand? But that’s version one of it. And then the next iteration of that is saying, actually, your portfolio in this store should actually be 20% of the vodka, 5% of the whiskey. So it’s continuously changing and evolving and expanding. That’s D-STAR.
And then Matrix is more about marketing spend. So traditionally, you had the chief marketing officer of a region like France would say, I’m going to put 10 million on Absolut, 5 million on this, 2 million on this, and then we’ll figure out how we’re going to mix that between TV, online, et cetera. And Matrix basically said, well, actually we can be more data-driven, so we can take all of that information and give you a recommendation on what would be the optimal way of designing and distributing that money. So maybe you need to decrease over here and increase over here, and that’s going to increase sales. And the point is it’s meant to be the scientific data-driven approach. So that’s the second tool. And again, similar to D-STAR, this is something that is continuously being improved, enhanced, and again changing how people are doing their jobs. So it’s no longer this A or B, non-digital, digital. It’s AI continuously evolving and changing and transforming.
BRIAN KENNY: So as a marketing guy, which I am, I can see how that data-driven decision making would really be attractive to me. The sales guy in me, and I’ve done sales before too, says, I don’t want somebody telling me what I should be ordering next for my customers. I think I know my customer better than you do, and I’ve built a relationship with this customer over time. How do you deal with that kind of human resistance that you’re likely to encounter?
EDWARD MCFOWLAND: I want to answer that question by piggybacking off of Iav. We do this all the time. So I think what Iav is capturing beautifully is this idea that AI transformation or changes really is a cultural change. It’s not just like I change the system, but I have to change my mindset as an organization to support that because it’s always evolving. You have to be adaptive along with it. You cannot be stagnant with what it was even a year ago. And so to that point though, you bring up a really great point. You said your hypothesis, right? Your position was that you could see marketing being super excited by this optimizing things. And sales being like, I don’t know, I know my customer better. But you’d be interested to find out that actually it may be the opposite in their case of who was accepting it and who wasn’t. And it was fascinating because marketing, you have this view of this is the brand I’ve been supporting for five years, I put all this effort into it. What is this system going to tell me about what people want and how much I should do this for? And it’s my identity, I know how to get you to buy the thing. And so what do you know from data? What is data? I know the human psychology of it. Sales is like, well, I know what’s better to sell. And you say, well actually you can make 20% more sales. Like, oh.
BRIAN KENNY: There you go.
EDWARD MCFOWLAND: Okay, I’m listening now. And if you do it and you make 20% more sales or more, it becomes a fact of like, well, I don’t know why you’re right, but if you’re right and I got 20% more sales, that’s more money in my pocket. There’s more everything I want. My identity might not be so much wrapped up into this idea as long as I’m making the sales. I think that’s in fact the assumption is that it’ll be the one way. But in fact we found in our conversation that was very different.
IAV BOJINOV: And just to add to this, there’s a really interesting thing that happened in the France market in particular. Pernod and Ricard are actually two separate brands that joined. I can’t remember how many years ago, 20, I forgot how many.
EDWARD MCFOWLAND: Yeah, 20 or so.
IAV BOJINOV: It was a little while ago. But in the France market, they kept them as separate entities and they only recently merged operations. So the salespeople, like Edward was saying earlier, they didn’t know all of the brands from the other part of the business. And so for them it wasn’t the, oh, I know how to sell. It was, I don’t know how to sell these because I don’t even know what they are. And so if you can tell me that they should be picking up these five other products that I tangentially have heard about, then yes, I’m going to do that. But the marketing person exactly as Edward was saying, what was it, 1975 actually. Sorry. Thank you, Edward. 1975 is when they merged, but they didn’t actually do the integration in France for quite some time.
The marketing person very similarly had, the chief marketing officer there, had this idea that she had been spending her time focused on one part of the business, and there was a few brands that she really liked and wanted to make sure that they get the money. And when the algorithm was saying actually that money, sure you can spend that money, that’s not increasing sales, that brand is well-known, everyone knows about it. It’s already being sold at sort of the maximum you’re going to get for this. You can activate these other brands that you don’t spend any money on and increase sales massively. And that was certainly going against her identity because in the past, that’s what she did. She picked the five products that they were going to invest in and that got her results. That got her to the position where she was. And now she was delegating authority and decision-making responsibilities to an algorithm that she wasn’t even involved in developing.
BRIAN KENNY: So nothing succeeds like success though. So if you’re getting the results, it’s hard to argue against the information that you’re being given.
IAV BOJINOV: But that requires you to adopt this and to start using this. And if you just blanket say, I’m not using this…
BRIAN KENNY: So let’s talk about that a little bit because you mentioned earlier, Edward, that the geographic dispersion, the decentralized management approach probably makes this harder in some ways. What did they encounter and how did they deal with it?
EDWARD MCFOWLAND: I think it’s a great question and a great challenge. And I think Iav’s point sort of crystallizes a bit as well because you can have situations where in different locales, different cultures, different countries, different structures will adopt and support like this is great. Or someone will be like, I don’t like this as much. Now I can have a conversation with you and push you in ways I couldn’t before based off of data. And so I think that does create some challenge. So I think one of the major questions for the organization as a whole was how do we do transformation not only structurally but also culturally? How do we bring people along? How do we convince them that this is not a replacement, but support and how do we get them, to Iav’s point, adopt. Because I think at the end of the day, these algorithms are fantastic if they can reach, support you with the potential you want. But that requires you to adopt them.
IAV BOJINOV: And just building off of that, one of the things that came out in this case is that really adoption at the heart of it is about trust. And I’m sure you’ve had some of our colleagues talk about trust between people. But now there’s a new framework that we’ve been thinking a lot about, which is trust between AI and humans. And really when you look at it, there’s basically three pillars. You have trust in the AI, which is answering questions like, how accurate is it? Is it free from bias? Does it have hallucinations? Basically does the algorithm work, right? That’s trust in the AI. That’s one pillar of it. The second pillar is really trust in the development team and do I trust that they have my best intentions at heart? Are they just building this so that in six months down the line they’re going to completely replace me? Do they listen to me? Do they understand the things that I’m struggling with? Do they take that into account? Did they take my feedback into account? So that’s really about the development team and the development process. And then the third pillar is really about the overall process that the organization puts in place. And what I mean by that is if something goes wrong, who is responsible? Or if something goes wrong and I want to overrule it, how do I do that? What are my incentives? How are the incentives of data? Edward was talking about if I get 20% more sales, do I get the commission or is that going to go back to headquarters because they built the AI that’s driving that, right? How is that going to be distributed? How’s that accounting going to be done? So really if you want to drive adoption, and we’ve now started looking more systematically at this question, it really comes down to trust in the AI, trust in the development team, and trust in the process. And if you don’t have adoption, you can almost always map it to one of those three key questions. And once you map it, then you can start to understand how to change that. And in the case that impacts, we hide it a little bit so that in class, we can bring up these frameworks, but it shows which of the pillars in the various places where there were challenges to adoption, which pillars do you need to go after to fix that?
BRIAN KENNY: Yeah. Those make perfect sense. The case also talks about nobody does these things alone. So they worked with a partner, Boston Consulting Group, and I want to talk a little bit about the global digital acceleration and the role that played.
EDWARD MCFOWLAND: It was key for them because at the beginning, they’re an organization who had done things a certain way. They’re totally not digitally native, but even digitally sophisticated, they’re trying to get there. And what BCG brings to them is expertise to help them figure out how to start building this capability inside the company. And so for them, it felt like a smart play up front. And so they help them structure, they help them come up with what they need to do, the plans, what things to build, how to build the team, a transformation team, in each locale. They had a consulting with each internal team that was being deployed, really helping throughout and also doing some of the analysis. What was interesting is a choice that Pernod Ricard makes is that they don’t want to do that forever though. And they don’t want to outsource to anyone else. They decide to build an internal data science team at central. And that’s a really fascinating choice because it is another question of culture and who has power. Because if algorithms sit centrally and they dictate decision-making, now what used to be powers even to the head of the locale is abdicated to some degree to central. And central has tended to be a very much hands-off. They let you do your own thing, you produce results better. And so the organization BCG was a great fit up front, but they were very clear that they wanted to internalize it and build the capability themselves, and they did it. They went from never having data science team not doing this kind of analytics to having a full-fledged data science architectural team that builds algorithms, deploys them out, and all that within a few years. And to me, I think that’s one of the biggest things they’ve done. That was remarkable for me.
BRIAN KENNY: Yeah, that sounds like a mammoth undertaking to build an organization like that within your organization. How did they ensure that the data that they were getting was the data that they needed? When you’ve got that many different players around the world, how do you ensure that you’re getting good data integrity?
IAV BOJINOV: Let me take that simple question. For people listening, Brian had a look at me expecting me to have a good answer for this. So here’s the reality of it. We made it seem in the case as if there are two products that were built centrally and then pushed to everyone else. But the reality is because each market is so fundamentally different, they have different regulations, they have different data requirements, the granularity of the data that they have is very, very different. For D-STAR, I think something like 80% of the code base is actually built tailored for that specific market.
BRIAN KENNY: Oh, okay.
IAV BOJINOV: And only about 20 of it is central. Matrix is a little bit more different. It’s about 80 central and about 20% different. So there isn’t one D-STAR, it’s one D-STAR per market. And so the reality is that for some markets where they have good data, they’re able to do this and they’re able to do this really, really well. And the data, actually most of it, a lot of it is third-party data. So they’re buying it externally from other companies that are sort of data curators. And so that can help ensure that high quality of data is the fact that this company is selling that, and they give them guarantees. But for the data for some other markets where they had very, very poor data, they essentially couldn’t achieve the types of results that they can for the larger markets. And they do their best, but it’s unclear how good that data actually is. So I don’t have a good answer for you.
BRIAN KENNY: No, but I’m sure this sounds really, really familiar to anybody who’s listening, who’s tried to go through the same exercise.
IAV BOJINOV: Absolutely, absolutely. And they did some smart things. They tried to go in the cloud, they tried to have a single source of truth that was continuously updated. They tried to build automatic checks for it and all the usual things that companies do. But that’s the reality. When you depend on third-party data and you don’t have your own data, it’s really hard to guarantee that peak quality. And like all the other companies, they’re trying to explore ways of getting that first party data.
BRIAN KENNY: We haven’t talked at all about the protagonist, the CEO, Alexander Ricard. Can you tell us a little bit about the role that he played as a senior leader? I would imagine this has to be top-down driven, and I’m sure his role has to be really incredibly important.
EDWARD MCFOWLAND: We spent a lot of our time with Christian and Pierre Yves in conversations about the case and things of that nature. But we had a chance to understand from them how Alexander set the tone and the structure for this. Because what they were very, very clear about, and it became very obvious to us when we were there, this was a top-down mandate. You were going to do this. And so again, I think looking back, it seems like the smart thing, especially in today’s world. But looking back in that moment it was not an easy decision to make to take a company, again, a family company, a family brand, his name’s on the door, and what they have built and take a risk/gamble in a way on this in ways their other competitors had not. And so I think from my perspective, what really happened was he said with his advisor and thinking about things and the board, we’re going to do this, we’re going to invest massively in it. And it’s going to be a risk, but I believe it is going to work out. Now. I think there are other forces as well pushing him toward that, but he could have resisted those as others have. What the company got was a very clear vision from their leaders saying, we’re going to do this. It’s going to work out, I believe, and we’re going to invest in it therefore.
IAV BOJINOV: And maybe just to add to the pressures and forces, Edward, you were just mentioning, there were two sources that were really big. You had the internal pressures, which is the number of products that they had was just getting, there was so many, it was impossible for a single person to keep track of that. So those internal pressures, the sales folks, the marketing people were saying we need help to be able to manage this vast portfolio. And at the same time, there were also external pressures. The competition, Diageo actually had publicly talked about some of their AI initiatives back in 2017. So they were already sharing these. They were saying, look at these major wins we have. A lot of external pressures, lots of internal pressures, but of course you can have those without leadership support to say, we are going to do this, we’re going to write these giant checks and have no notion of exactly when this is fully going to pay off. We can guess maybe one or two years, but it’s never certain.
BRIAN KENNY: How’s it going? Is it paying off?
EDWARD MCFOWLAND: From when we were there, it had already clearly paid off from their perspective. They said, we’re going to actually build our homegrown from start our own data, we’re going to do it and build it our way. And I think that was, again, a further step. And it seemed that they have been extremely excited by success that they’ve seen, and therefore they have been actually ramping up.
IAV BOJINOV: And just to add to that, I actually bumped into Pierre Yves a couple of months ago. So he was here, we had a conference for the Harvard data science initiative that was, “From Vine to Mind,” which was doing data science on the wine industry. They came in.
BRIAN KENNY: That’s a clever title. I don’t know who made that up. I like that.
IAV BOJINOV: One of our colleagues, Xiao-li Meng came up with it. And so he was there and actually brought a bunch of their wines. So I had a chance to catch up with him. And we need to write a B case basically, because they’re heavily, heavily invested in not just the capabilities that we talked about, but they’re really re-imagining what the future salesperson looks like, what the future marketer looks like. And they’re heavily investing in generative AI to really get integrated in basically everything. And again, this comes back to that point I was saying, which is this isn’t a transformation. This is an evolution, a way of life that is just going to continue. And that’s the direction they’ve been going. So I think from him, they’ve been tremendously successful and they have a lot of backing and a lot more resources are going into this right now.
BRIAN KENNY: This has been a great conversation as I knew it would be. I’ve learned a ton. What lessons do you think our listeners can take from here and apply to other industries?
IAV BOJINOV: We talk about digital, we talk about AI. It’s all about people. It’s all about culture. It’s all about change management. And so I think the big takeaway is the technology is there, people are not. And so what you really need to focus on is your people.
BRIAN KENNY: Yeah, that’s great. Edward, you get the last word here. I’m sure Iav could jump on if he wants to.
EDWARD MCFOWLAND: He will, I’m sure.
BRIAN KENNY: But I always end by asking if there’s one thing you want our listeners to remember about the Pernod Ricard case, what would it be?
EDWARD MCFOWLAND: That’s a great question. I think it’s what I learned from the case as well, and it is essentially what Iav mentioned, but I want to take it a step further in the context of this organization and say, I think we often think, and I’ve looked at companies all the time that are more traditional companies that go, we want to do AI, we want to do this transformation. And I think the key thing is they realize that it is the people. The organization is a collection of people working together toward a common mission and goal supported by resources and structure. And so if the people have the will, and that often requires people who have the ability to support will and to push resources a certain way to support this. If they have it, you can do it. And it was just a remarkable opportunity to see and witness and talk to people and see how well they did this. So to me, it is, you can build all the fancy algorithms you want. We do that. I’m writing theory as we speak on different cool tools and technologies, and it’s great and it can do great things. But if it’s not integrated into the workplace of people helping them thrive and they don’t feel that and they don’t adopt it, the organization will not thrive as a function of it. And it can be done.
BRIAN KENNY: Or Harvard Business School, we’re going through our own process right now.
EDWARD MCFOWLAND: I was not going to mention that, but I thought about mentioning it for a second and then I said maybe not. But yes, that’s exactly right.
BRIAN KENNY: Edward, Iav, thank you so much for joining me on Cold Call.
EDWARD MCFOWLAND: It’s been a pleasure. Thank you so much for having us.
IAV BOJINOV: Thank you.
BRIAN KENNY: If you enjoy Cold Call, you might like our other podcasts: After Hours, Climate Rising, Deep Purpose, IdeaCast, Managing the Future of Work, Skydeck, Think Big, Buy Small, and Women at Work. Find them on Apple, Spotify, or wherever you listen. And if you could take a minute to rate and review us, we’d be grateful. If you have any suggestions or just want to say hello, we want to hear from you, email us at coldcall@hbs.edu. Thanks again for joining us, I’m your host Brian Kenny, and you’ve been listening to Cold Call, an official podcast of Harvard Business School and part of the HBR Podcast Network.
