The Economist. Alright, I'm going to open my laptop and Google with my one finger typing. How much money does Google make from advertising? And what we see is that at the top of the screen we have something being generated called the AI overview. And it says Google generates the majority of its revenue from search advertising. In 2024, Google's revenue from search and other advertising was $198.1 billion. That's quite a lot of money. It's also the reason why its business model is under threat. Because if AI can generate an answer to my question and I don't have to click on all those sponsored links, how's it going to make money? Why should I go to Google at all? When I joined in New York, there were about ten people in engineering in the office there. Liz Reed ought to have an answer to my question. She started at Google in 2003. It was her first job after college where she'd studied computer science. And the company, not yet five years old, was already turning into a verb. It was an exciting time. Reed has a moment of that era on her desk. Sort of a print where people signed in and wrote messages, including one that has a picture of me when I was about four months into Google. Which is a great reminder about how Google has changed, how I've changed. I look like the dorky 23 year old at the time. Now I'm the dorky 43 year old. Dorky perhaps. But Reed has come a long way since her time as a coder. Last year, Google promoted her to head of search, making her one of the most influential leaders in the company. It's her job to help make the world's most popular website fit for the age of AI. I'm Andrew Palmer. This is Boss Class, season two. If you've listened to our episode on innovation, you'll already have heard Reed talking about developing AI overviews. Now, in the first of a series of Meet the Boss interviews that we've produced for season two, we'll be taking a closer look inside one of the world's most valuable firms. And we'll start with how it feels to look after a product with two billion daily users. I think it's both like a really large number of people, but at the end of the day, it's like people, right? And so it's people who come and they come to help find a job. It's people who, you know, they've been diagnosed with something medical or a family member has and they want to understand that. They also just want to like stay up to date with their favorite sports team, but the breath of what people come to for search continues to leave me in a sense of awe. Your career started as an engineer and then you've gradually taken on managerial leadership roles. I don't know whether you continue to do coding at the front line as well, but what does that look like, that journey from starting as an engineer to running very large teams? Yeah, I don't do very much coding these days, to be honest. It's definitely been a transformation and it's this question about like how do you have the most impact with your time and where is it? You know, I don't think I was the most amazing coder ever. And so I think what I found is that that opportunity to leverage enough understanding of tech, but with helping enable other people and really bring them in an environment where they can thrive and produce great things has been the right balance for me. So I try and stay close to the tech in the sense of what's the possibilities of the technology and what can enable, but I'm not actually doing a lot of coding these days. But it's different. You had to redefine impact when it was like coding, you would go and you would sit there and be like at the end of the week. Here's the feature I built. Like that's how you measure it. And this is a little bit different because you talk to a bunch of people, you coach people, you made some decisions. How do you measure impact on that? So you have to think about a little bit different. Do you have a way of thinking about that kind of multiplier effect? I mean, what is the best way to amplify? I think many, many different ways, right? So one is, can you give people the clarity of goals so that they understand what to do and they can figure out the how that really allows them to bring their ideas and generation into it? I do think it is a lot about figuring out how you have a team of smart people, people who are smarter than you, frankly, and who can activate great things. How can you help some coach and grow? Can you remove a roadblock along the way? Can you make good decisions because you have a broader purview of things? Can you be clear in communicating not just what you're trying to do but the why? I think that's really, really important because if you just communicate the what, people don't know why. And so if there's new information, it's like, oh, well, Liz said such and such. So now we're going to go do it, add infinite item. And you're like, whoa, it does not really what I meant. And so if you can teach people why, then they can come back and entertain a dialogue with you. Okay, I know you said this, but given this or since your goal is really this, I have a better idea on how to achieve what you're trying to do. So I wanted in this conversation to take us or have you take us through the work that you're doing on generative search and AI overviews in particular and kind of use that as a way of into a conversation around innovation at Google. Most people are now seeing AI overviews. I think either in labs or out there in the wild. But when did this story start for you? Oh, I think the story started several years ago at Google. Search has always tried to think about how can you use technology towards this mission of organizing all the world's information. And so we started using large language models a while ago. And then as the model is advanced, there were various different efforts within search on, okay, well, how does that allow you to ask new questions? Different teams sort of had demos or little projects together. And then at some point we sort of started to pull them together and we had more of a sense of what the product is. And I think that's often the case with innovation where people would be like, aha, there was this like what a hot moment, maybe. But like before that, it wasn't like there was nothing and then there was an immediate a hot moment, right? You're scratching the surface. You're finding out things. You're seeing what's possible. It's not. And then it starts to take shape and pull it together and build something that became what is AI overviews these days, which I think honestly is both an exciting step forward with search and just the beginning of what I think generative AI will be able to do for search in the coming years. And what we see now, were there many other or very different possibilities that you explored beyond the overview? Yeah. I think there were versions that were really focused on thinking about different forms of overviews. We had demos and we still have demos about thinking about different ways to organize the page. We had demos around thinking about how do you refine your query and ask the next question. But they were sort of like each individual team had some angle of expertise at which they were playing with it. And so it was kind of thinking about how do you pull them all together. And even what you think of AI overviews, the first version that we launched in labs looks pretty different than the product does today. And in the months that we were building it, it went through many iterations. You would figure out this felt great. This felt too hard for people. This was confusing. This was too long. This was too short. Search is really interesting because people have an expectation that they can get information very quickly. And so one of the things you're trying to do with AI overviews is figure out that right blend of ensuring that if you came in with a question, you can get to that answer as quickly as possible. And also people like learning a little bit more than they necessarily asked, right? And so what's that right balance, even within the response of the clarity, when should you elaborate, how do you help people have ideas of what to do next? And it's an evolution both based on understanding what users want and the tech keeps changing, right? So even if you better understand what users want, something used to be slower now it's fast. Something used to be confusing and now it's clear. And I've never seen a time where it's so complicated in the tech and that like I can talk to like two people on the same team working on the same thing. And they'll have very different ideas about what will be possible in three months, right? And so you can't sort of just plan and run. You have to like come up with a plan, but then you have to test it and then you have to revisit it and you have to learn and the level of agility required is a lot higher than when the tech wasn't changing as quickly. How did you conceptualize or articulate the problem that you were trying to solve with AI overviews? Yeah, I think if you go back to our mission of like organizing the world's information, you can go off and like stop right there, organize the world's information. But there's like the second part about making it universally accessible and useful, okay? And useful is a very different bar, right? You can have a medical book in a library that you can check out and it can be available to you, but you may not understand any of it with that hour. So great, it's accessible, but who cares, right? And so we constantly have this question about how can you make information more useful to people, okay? And so what we thought about a lot with AI overviews is what are the types of questions that are either not really possible or they're really hard for people to get a good understanding after and how do we change them? And so one class of those are cases where the information is scattered across the web. And so you and as a user have to go and figure out how do I translate that into a lot of separate questions. instead of each other.
being, "Okay, I'm going to search for dog friendly campsites, and I'm going to search for kid friendly campsites, and I'm going to search for affordable campsites." I'm like, "I need an affordable campsite that is available in this time and allows dogs and will be okay with my kid and a store." That's actually the question they had, but because nobody wrote a webpage specifically for all of their needs, they would break it apart. "Okay, well, can you use gender and a VAI to pull the relevant pieces from across the web and summarize it and then give you a link to the webpage that's great about talking about dog friendly and the webpage that's great about talking about kid friendly without requiring it to be across." This is even true in different languages. We take for granted a lot of users in English that like everything you can think of as on the web, that isn't true with a lot of other languages. The size of the corpus relative to the size of the speakers of that language can be dramatically different. Hindi is a great example. The size of the Hindi corpus on the web is much smaller. If your access to information is limited by, did somebody write in your language, then your access to information is gated. But the large language model allows us to say, "Okay, we learned this thing in English, but now we can communicate you into Hindi." Really thinking about this case of like, "What are the questions you couldn't ask?" Or they would have been hours and hours of work and how do we unlock them with this technology? That's still a sort of determining factor in when you use overviews or when you see links. The overarching way that we think about that is we want to provide an AI overview when is a net add to the search results page. That's a mix of both what's the quality of the AI overview itself, which changes over time based on the tech. What's the quality of the underlying search results page to begin with? The quality of the underlying search results page might be different in different languages because the web pages are not. What we see with queries that show a overviews is they show a greater diversity of sites. The reason they do that is because if your query was longer, there might have been very few pages that answered all of the parts of your queries. We can be like, "Aha, you want to know this piece and so I'm going to find a really interesting page about this part of your topic and a really interesting page about this part of your topic and surface that." We think of AI overviews as a place to get started and then continue to jump off as opposed to a complete end going forward. Is that how you measure success as well? It's right at the top of the page if it shows, so presumably it's bound to start to absorb an enormous amount of traffic. What's the way in which you define it is doing the job that you wanted it to do? You're right that a unit that should show up at the top will at some level get more clicks than if it were at the bottom. But you can ask the question, "Yeah, but is it like getting more user happiness or more user engagement than if a different unit was at the top?" That's the way you're thinking. You can look at this in different ways. We have, under saying, "Just one, will people use search more often?" If you put something bad at the top of the page, this will not cause people to use search more often. This will cause people to come to search less often. That can be a signal. How do they engage? If they're not happy with results, they often try again. They issue another query. If you're shrinking the number of times that a user has to issue a query and then try again and try again, we've seen years and years over. That's a sign that people are happier because you're basically saying you're making it easier for them to issue. We constantly ask this question, "What does it take to make search effortless?" Then we use a variety of metrics to understand it for doing a good job. We've seen this been very successful with AI reviews. This higher satisfaction from users when we measure it. People actually do actively more searches than they think about a question and decide to ask it than they wouldn't otherwise have done. They click on a greater diversity of websites. There's this myth that a lot of people think that if they are an active user of Google, all of the search queries they could imagine they come to Google. That's actually not true. You do a mental calculation that says, "If I ask this question, how much do I have to do I have to do much time and work? Is it going to be and is it going to get a good answer?" If you really want the answer, it doesn't matter. Even if your chance is 10% you go ask it. If you kind of care about the question, then you're not going to ask it unless your confidence is going to get a good response. What we see with AI overviews is people's confidence that they can get a good answer and they can get it quicker goes up with what they expect from Google search. They come and they ask new questions because now that bar of value has changed. I guess over time, initially I think overviews were spitting out some fairly odd results in response to some fairly odd queries like how many rocks should I eat a day, etc. Did you see that hit confidence or was that something that you were kind of just like we're bound to get through this kind of hallucinatory phase and it will get better? The examples you have absolutely were there. They were quite rare in practice, but they were real feedback and they kind of pointed out that people were asking us questions that we did envision. We envision some sense of like new questions people might ask. But to be honest, if you asked me before are people going to ask us how many rocks should I eat a day, I don't go like yeah, that seems like a question on top of people's minds. But then they ask that question. You're like oh, I need to go do something about that. This is something that's been true well before gendered AI in search. Nobody searched for opening hours 20 years ago on Google because they're believed that they would get opening hours was basically nil. They just didn't ask that question. Then you started to do it and then people asked more and then they were like yeah, but what's the opening hours for the breakfast menu? And you're like oh my gosh, okay, there's a new question. Now I need to go think about that. People constantly raise the bar for what they expect Google search can do. That's great. That's a great challenge for us to constantly meet to rising expectations. How do you handle that structurally? What I mean by that is like is there a different team that kind of looks at user behavior once a feature has been released or a product is out there from the initial team? Yeah, I don't think it's quite that sort of black and white. We have both data science teams that are constantly analyzing the usage and we have user research teams. So they go out and they will talk with people, they will run surveys and ask and get that. And so those groups are always working with the engineers and the product teams on how to improve. It is interesting to innovate because you both want to innovate on what people already come to you for and you're not doing a great job, but you also want to innovate on things that people aren't yet coming to you. And so our user research team will try to understand what are people's problems that they have even if they're not coming to search for them. Or what's difficult for them. And so they try and help us find the headroom that we don't even see and go after that. How do you identify that if they're not coming to search? You could do something like a diary study for instance in which you ask a set of people to go through and write down what were they trying to do that day? What were things that were difficult? How were they thinking about it? And you were like, oh, okay, they did all of these things or this was really hard for them. And this is what they really wanted to ask. And so it's a mix of qualitative and quantitative to understand sort of what is the area of future. Not just what we see today. And how do you handle within the company, though, sort of competing pressures? So a business model based on people clicking through links and worries that presumably are being expressed somewhere inside the company about what is this going to do to that kind of traffic? Yeah, I think maybe a couple of things I would highlight. One is I think helping users is always our North Star. And especially on the search team, that's what we think about. And Larry and Sergey talked about that right from the beginning of that focus on the user and sort of a relentless focus on the user. And so that's really what guides us. But I'd also say this question of the ecosystem sometimes has this myth that users want either an AI overview or a link. And that's actually not what we see. What we see that actually a lot of people like about AI reviews is the fact that they can get both. They can get started and then they can dig deeper and they can find a source that's useful. And search is also very broad. So if you're just trying to figure out the birthday of a peasant find. If you're trying to get fashion advice, do you really want to trust all your fashion advice to an AI generated response? No. And so I think of the way that we think about AI in search is AI is in a replacement for search. It's an enhancement. It's an opportunity. We talked about our first launch about super charging search. This ability to make search more capable. But I think people will still want to genuinely connect with other people for all sorts of use cases for years to come. Whether you believe that or not, Google has little choice but to press ahead. Whether it's holiday plans, fashion advice or rock-based diets, figuring out the ways in which search can be made more capable by AI requires extensive testing and refinements. So a couple of years ago, the firm unveiled a spruced up version of its Google Labs website, which allows enthusiasts from outside the company to play with experimental tools and features. That is one stage, a new idea passes through before it is properly let loose on members of the public. We will often do it by first building something within the team and then we'll do what we call a dog food in the sense of allowing a set of people internally to play with it a lot, give us feedback. We might run some user tests on that with the user research team and hear what trusted testers do. Then we bring it to life.
then we'll often do a live experiment in the sense of writing it on a percentage of the general population and then bringing it to the full population. Labs is something newer we launched a couple years ago. So previously we would have kind of gone from dog food to the live experiment on a set. The live experiment is really good when you think you have something that mostly works and you're trying to tune it. It's not as good when you're trying to figure it out because you have a set of people that are like really excited to try the new thing and work with you and you have other people that are just like, I don't have much time in my day. I just need the thing that works. Please don't give me anything that's beta form. And so labs allowed us to actually learn with those who are excited to learn with us and give us feedback before we impose it on a set of people who really rely on search for things as we go. I think one thing we learned over the course of that was around the sensitivity to time and latency. So if you can put out an AI overview that's pretty good and you put an AI overview that's a little bit better but a lot slower. Users will sometimes prefer the AI review that's faster but a little worse because it's good enough for them. So there's this precision that we have to work with so in search we can use models of different sizes between, okay, we want these ones. It's a pretty easy question. So just use the faster model to help you get it out. Okay, this is a much harder question. You want to help with math homework? You're willing to wait 200 milliseconds more to ensure that you get the right guidance. So sensitivity is very high because people are used to on search getting a page very quickly. And so if you slow them down and the page was good enough otherwise, they are unhappy with you. And so you're constantly paying attention with not just how do you improve quality but how do you do speed. And we knew this intuitively at some level from the years over but we did not realize how sensitive it was in some of our experience. Ooh, okay, I can't give that set the bigger model. I need to give them a smaller cheaper model because they didn't want to wait. They thought it was good enough without and they didn't want to wait. So that was one thing we learned. So you talked about how the tech is changing so fast that may therefore make it difficult for you to predict but very broad terms, where are your attention focused? So I think if we go back to both the mission, then a lot of the focus is on how do you truly make search feel effortless for any question you can ask. Okay, and so I think where a bunch of it is going is asking the question either where does it still feel very hard? So multimodality is one example where you know first you could only do text, then you could do image. Now you can do video or text plus image. Continuing on things like that evolution because that's how you talk to a person, right? Like if you were in a store and you saw something you'd be like, hey, I want to dress like this but in a different color or maybe shorter, like you would use this combination of different techniques or if you were pointing to a technician who came into your house, you'd be like, okay, see this dishwasher, it's these lights, that's how you would talk, right? So this question around effortless, I think to make it effortless, you need to allow people to ask more and more of the question that's actually at the heart. The question they want to solve, not the question they think the computer can answer. And so you're pushing the bar of what is possible in the types of questions people can ask. It doesn't necessarily mean the question is long by the way. So the question could be long, but the question could be like, how do you think the 49ers are going to do next year? This is a short question, but there's a lot behind that question, right? Okay, well, how do you think about the players? How do you think about the environment? Okay, so like to answer that question, you actually have to pull and look at a lot of different sources together and understand that question. So I think you'll see more and more of this push for is the tech coming to you. Okay, like right now you have to figure out how you take what you want to do and translate it in the way that the tech understands and the tech spit something back and then you have to translate it back to you. Okay, but like actually we want to break that down. So the tech comes to you and you get to talk about it however you want and you get to learn about it however you want. And then we will do all the translation back and forth to achieve that. And I think that will be really exciting to see. Do you need to like train us to ask different questions? Yes and no. I think part of the goal is not to train you how to ask a question. It's to untrain you from believing that you have to ask the question in a particular way because you don't think the technology will let you do that. And so I guess it's a form of training, but it's actually sort of loosening your expectation, right? I sometimes think about like three year olds were like they ask whatever question comes to mind because they don't worry about how hard it is to answer they assume you know everything. They don't have any sense of like time is precious sort of else. Okay, they just ask it and sometimes it makes no sense, but like they just ask can we actually activate that ability of curiosity and wonder that you had when you're three to everyone in the world, right? And then you can actually get a question on any topic they want and give them the confidence that they can actually get help with what they need. And that's really exciting to think about is sort of allowing the human curiosity that it's an innate in us to actually be like uncapped because the technology isn't holding you back from it. It's actually enabling. I like that unleashing our inner three year old as a goal is a good goal. Well, exactly that's right. Yeah, that's exactly right. So can we now sort of try and draw this out to kind of more generally how Google thinks about innovation. I think some of the themes have already come through, but also what your advice would be to people who are running businesses or aspire to run businesses in other industries. So one thing that's very clear is that you're very very data focused. What's the sort of room for instinct there in innovation and how do you conceptualize that? I do think you have to mix it with a set of instinct because you're charting new territories, right? So you can't just look at data before. I think one thing that stuck with me that one of my UX designer peers years ago told me is like make sure you're falling in love with the problem, not the solution. And so if you're really looking what is the thing you're trying to unlock, then you want to be obsessed with that. And so you don't want to just focus on here's your first idea to do it. You want to get focused on testing. Does the solution actually solve the problem for that? I think a lot of the instinct, the question I sometimes push people on with that is like what is the hypothesis behind the instinct? You can't always measure everything. So you can't always have data. But the instinct is like this is hard for people. Or watch somebody do this and see why it's difficult. Or if I can unlock this, this thing would happen. And so thinking about the hypothesis I think is good. Encouraging people sometimes to build demos and try things out. So if you're like I have this instinct, let me write six months of beautiful slides for that. That's actually less useful than actually trying to bring up a prototype, particularly because you're trying to see what's possible with the tech. I try and be careful, especially because Google has had a long tenure about like when somebody says we've tried that before. Most of the things that I think of as being really successful at Google, or at least in the search and mouth space where I've worked in, they were tried two or three times before and failed. Because the instinct was right, but the tech wasn't ready or the data wasn't ready or something wasn't ready. Like what? What do you have in mind that? Yeah, I'll give you a few examples. So right now on Google Maps, we have quite a lot of reviews from users and photos from users. We tried that for years before we could get any traction. We just couldn't figure out how to unlock it. And then two things changed. So one was, and they were mostly around mobile. And so they allowed the combination of notifications and understanding where a user was. So if I go to you, Andrew, I'm like, hey, Andrew, would you like to review all the restaurants that you've been to in the last two months? You're like, oh, now I have to think of the restaurants and I have to search for the restaurant one by one. Oh my gosh, there's a lot of work. So you're not going to be more relaxed and do that. But if you're happy to give us reviews because you love sharing your view. And we know that you went to the restaurant yesterday and we can alert you and say, hey, would you like to write a review for that restaurant? Oh, that's a lot less work, right? You just click the notification and you click five stars. Well, you've given a rating. And so the bar of friction was really high for us before and we couldn't do that. And then suddenly the bar of friction for the user got way lower. And we were able to interact with users. So I think that's an important part of fostering innovation. I think the other thing is just like really dreaming big enough. Okay. And so are you focusing on innovating on something that even if you solve it, nobody cares. Or you focusing on innovating on something where you don't know if it's possible. But if it's possible, it's going to be amazing, right? And so you really want to think about that question. And so like, this is not possible. I really wish it were possible. What are the things like you have a hunger for that you just don't think are possible historically. Can you innovate on those? And I think that's important is that level of ambition because you don't want to spend all your time working on something that you solve the problem that just doesn't matter. It's much better to spend a lot more effort and solve something that if you solve it.
it really actually changes people's lives. - But you're describing two different things there in a way, in the course of this conversation, right? I mean, the sort of iterative approach and kind of rolling out, rolling back, et cetera, et cetera. And then something which is very, very ambitious, potentially radical, the 10x idea, do these things just loop together in your mind or do people work on different things? - I think in any given time in your portfolio, you wanna have a mix of innovative and iterative things because you're not gonna reach the full potential of something just by that first beginning of the innovate, right, the innovative unlock something, but to make it excellent, you have to really iterate a whole bunch on it, but then you're starting like the next frontier of innovating for the next generation, right? So if you take something like lens, which is our ability to search by an image, okay? First, it was like basically not possible enough to be useful. And then we launched the first version and people were like, whoa, that's so innovative you could do it. But also frankly, the quality was not great yet, okay? But it was enough of an insight to get started, right? And to suddenly feel like it's real. And then there was a lot of iteration on top of it to get to the ones, and that iteration might consist of smaller innovations, right? But then as that got bigger, you would then go and say, okay, well now we can do not just image, can we start working on image plus text together? And that creates the next wave. And so you need that mix, but both that iterative and innovative, another good example is various efforts around AR and glasses, right? Like the first versions didn't quite work. We could do this one day, they'll be really big, okay? But it's taken lots and lots of different innovations along the way. And so if you know that there's real user value, then you want to kind of keep coming back to the question about when can I unlock that user value? - I know we're down to the final two minutes, but that famous sort of 20% rule around people being able to sort of time to work on something which is unconventional away from their day job. Is that something you still maintain within Google? - I think it's always being in varying levels about how many people did it exactly as the 20%, but this general concept of ensuring that you can't just do everything that's directly known on a roadmap and you have to make space for unexpected ideas. I think that's something that we try to continue to nourish. And sometimes that can be unexpected ideas in a particular space. Sometimes that can be actually just spending some of your time learning about what's possible in the new technology so that you can come up with better ideas. But I think we want to still always make space for ideas that none of us thought of. And it's not just what are you trying to do in your environment, but you also have to continue to look at the incentives, okay? So do you celebrate the new ideas? But also, I think one of the things we're trying to get better at is celebrate learnings where the initial goal maybe didn't succeed, but we learned something that created the next great idea, and opened the door to the thing that will work. And if you only celebrate the things that work, then people only want to work on things that they know are going to work, which is not necessarily in the direction of allowing innovation. And so you have to think about what are the ways that you are rewarding innovation, what are the ways that you're allowing people to learn and not just succeed. In the all hands a few months ago, we deliberately went through a few examples in which we had team share. Here was a project that was trying to achieve X and didn't work and how it either pivoted or how it created the room for the next one. And people kind of assumed it was on the straight line to success, and actually it was not at all on the path to success, and it had to do a massive pivot to get there. And so seeing that even the things that you think were work went through this process, and that we were also highlighting ones that weren't really well with the team and inspired a set of people to think about it differently. - Fantastic, thank you so much. - Okay, thank you so much, really, really appreciate it. - On the next episode of Boss Class, we'll get some surprising advice about something that you probably dislike doing, giving a presentation. - I'm really kind of suddenly aware of how my knees are tight. - I call myself the neat police when I'm working with people 'cause I'm constantly going, "Nees, knees." - It's all in the knees. - And across the rest of this season, we'll also bring you more meat the boss episodes like this one. We'll hear how Lars Jorgensen runs Novo Nordisk, the maker of Azempek and Wugovie. - If I tell those new leaders that I have been in position where I had to make tough business decisions that had a short-term negative impact, then I also give them the mandate to do the same. - We'll hear how the economist, Mnus Sheffiek, dealt with the Euro crisis, Brexit, and her presidency of Columbia University. - I was inaugurated on the 4th of October, and then October 7th happened, and then the sort of the world blew up. - We'll learn how the diplomat, when Dishurman, handled an almost impossible task, negotiating a nuclear deal with Iran. - This was not a negotiation of equals. We were trying to stop there having a nuclear weapon. - And next week, we'll meet Michelle Garz, the CEO of Levi Strousen Company, to talk decision-making, Denim, and doing away with PowerPoint slides. - Well, I think works really well. What I respond to is when someone comes in, and they basically give me the top three headlines, with conviction, with passion, with clarity. When I see people do that, it's like they've got me. They've got me at the edge of the chair, and that's what I try to do. To listen, you'll need to be a subscriber. Right now, you can sign up for half price, just a couple of dollars a month. Search Economist Podcasts Plus for our best offer. We'd also love to get your thoughts about what you've heard, and what you'd like to hear on the show. Email us at
[email protected], with the subject line, Boss Class. Boss Class Season 2 is produced by Alicia Barrell, Sam Colbert, and Pete Norton. Our sound designer is Wade Hong Lin, original music by Darren Ung. The series editor is Claire Reed. Our executive producer is John Shields. I'm Andrew Palmer. This is The Economist.
PodcastTranscript AI