I'm Camila Castellati and this is Made It. And today I sit down with Eric Reese, author of The Lean Startup, to discuss how influential the lean startup method still is today and how it's evolving in an AI world. MVP, pivot, product market fit. These words have become part of the startup vocabulary. Did you know that Eric Reese is actually the person who invented them? Back in the early 2000s, his first startup failed. He realized he had wasted months building something nobody wanted. Out of that failure, he flipped the script, co-founded his second venture, IMVU, where he tested a new method of doing things. He used a process called continuous deployment to release new software an average of 50 times a day. This planted the seed for the lean startup method, a radical way of building companies by launching fast, testing with real users, and learning before you scale. The idea turned into a global movement. His book, The Lean Startup, published in 2011, has sold over 1 million copies, been translated into 30 plus languages, taught at Harvard and Berkeley, and shaped how almost every entrepreneur builds today. Today, Eric is the founder and CEO of the Long-Term Stock Exchange, an adviser to top startups and VCs, and he's working on a brand new book, so stay tuned.
Eric, it's great to have you on the podcast. It's almost surreal to have you on the podcast because you're such a legendary figure in the startup world. And my first question was really around sort of the inspiration that led to the lean startup. I'm sure you've told this story many, many times, but I was curious to hear when you realized that this method that you were testing on your own venture could become more than just 'this is for my company, it's for me,' but actually could translate into something that could work for all kinds of startups. And I'll remind the audience that the book was written in 2011, you published it in 2011. And so it has survived the test of time and it has also survived the test of a multitude of types of startups that have tested it and used it. So yeah, how did it all start? How did you decide to write the book?
I mean, I wish I could say I had a master plan and I always knew what was going to happen. But the truth is I was very passionate about this way of working and people thought I was just totally nuts. It was incredibly controversial. People now that it's famous, people don't remember. But you know, there was a time when the name of a product contained the year that it came out, like Windows 2000 or you know, Microsoft Office 95. That told you a little bit about how fast or slow software products were developed. It wasn't considered unusual to ship a new product every several years, which is still the cycle time in many kinds of businesses. You know, a new auto platform or a new refrigerator has a very slow release cycle. So people didn't think anything of it and here I am talking about releasing monthly, weekly, daily, hourly, even faster. People thought it was quite wild. Anyway, I spent a lot of time trying to understand why we do what we do because I could see that it worked myself and I taught it to many of the people that worked for me. I was an engineering manager, you know, I was a CTO. I wasn't an author. I was a programmer. And yet whenever people would ask me like, 'Why does this work?' I was like, 'I don't know. It just seems very obvious to me that it will. Why don't you try it and see if you like it?' And then they would try it and you know, I wish I could say they always liked it but quite often people were like, 'Wow, it's exciting.' And also I'm a very curious person. I wanted to know the truth. Why do these things work? So I spent a lot of time trying to figure out myself intellectually why does this work and then try to teach it to other people. It's actually a hard thing to teach because it's very counterintuitive and you know, over a long period of time I worked it out and pulling ideas from lean manufacturing and OODA loops and all this other stuff, you know, I was able to put the pieces together and I would talk to people about the lean way of building startups. I didn't call it, it wasn't like my catchy catchphrase. It was just like manufacturing applied to startups. And at a certain point I was at a point in my career where I was like between startups and trying to figure out what to do next. And a bunch of VCs were asking me if I would come be an entrepreneur in residence or maybe they would fund my next company. Would I like funding? And I was like, well, I'm very flattered but I don't even know what the next company would be. And they were like, well, it doesn't really matter. That's okay. We'll fund you and then you'll figure it out. It's a very Silicon Valley story.
Very, I was going to say.
Yeah. And it was like part of why Silicon Valley is so great. But I was still too curious. You know, I wanted to know, before I go become the CEO of a company, I want to know the answer to some of the fundamental questions of entrepreneurship.
And so what happened was I had agreed to work with some VCs on various things. And they would bring me into companies and ask me to help them. And I think from the VC's point of view, it was like, look, I had this reputation at that time that my engineering teams were supernaturally productive. I won awards on this basis because we would get so much more done because we were using this better way of working. And they would be like, can you just sprinkle some of that magic pixie dust on the slow engineers on my team? And I'd be like, no, no, no. There's no pixie dust. It's not magic. It's a system. But they would pat me on the head and be like, yeah, yeah, but will you meet with them anyway? So I would have these meetings one after the other where I would be asked to share my experience with people. Say, 'Look, I think this could work for you. Here's how it worked for me.' And then I would get ejected from the meeting. People would get angry and they'd ask me to leave. It was very unpleasant. And I thought, 'Look, you asked me. I'm here as a favor. I don't work for you. You asked me as a favor to have this meeting.' And so I thought, 'Well, maybe I'll write some of these stories down.' I did it anonymously. That's how embarrassed I was about it. But then when somebody asks me for a meeting, I can share with them the story and if they think it's crazy, then we don't have to have the meeting and I don't get yelled at. That was really as far as I thought ahead. And those stories and that blog and everything started to really take off. And then, you know, it was many, many months before I realized that this was a thing. You know, I thought it was just fun to have a blog. It was fun to be writing about these things. But all of a sudden it turned into really a movement and people started to ask me to speak at conferences and to travel around the world and it just became a thing. And I just feel like there was a deep latent desire among entrepreneurs to find a better way. And so it wasn't so much that I had a great idea as much as we were tapping into this wellspring of desire for legitimacy, for an understanding of not just what we do but why we do it. And then eventually, you know, I advocated for it as a management science, a new branch of the management sciences, the entrepreneurial management branch. So I was like, we also need to recognize this as something serious that can be studied and that can be taught and that should be part of our economic vocabulary. That's how it started.
Do you want to know one of the most frustrating things about starting a company? Finding yourself stuck with a business account built for a world that just doesn't exist anymore. Bureaucracy, clunky interfaces, no mobile control. Every small task turns into a waste of time. We figured this out right away and that's why we chose Revolut Business. And here's the big news. We've secured 6 months of a free plan for you if you sign up through the link in this episode's description. Revolut Business is a bank account built for people who want to manage their company's finances efficiently, all in one place. No paperwork, no endless phone calls. With Revolut Business, you can issue up to 200 cards for employees with limits and approvals set by you right from your phone. You can accept payments anywhere via links, Apple Pay, QR codes or Revolut Pay, quickly, easily, and with transparent fees. And if you work internationally, you can get access to currency exchanges at interbank rates. No hidden surprises. It's a tool we use every day to run Made It. And we would just recommend it to anyone building a company. So, what are you waiting for? Sign up for Revolut Business. You won't regret it. Especially because by using the link in this episode's description, you get 6 months of Revolut Business free of charge. No strings attached.
Before we get too deep into my conversation with Eric, let's do a quick recap of what the lean startup method is actually all about. First, you start with an MVP. You don't need a full app, fancy logo, or investors. You need one simple version of your idea that you can use to get in front of your customers. Make something small. Put it in the world. Watch what happens, what works, what flops. The goal is learning. Founders need to be truth seekers. Next, validate your assumptions with real users. Have real humans, as many as you can get, use your service or product. Talk to your users. Listen to what confuses them, what they don't care about, and what they actually use and love. And when the signs are screaming, 'This isn't it,' pivot. Don't waste time and money on something you think is great, but people don't want. Shift early. Shift smart, which doesn't mean giving up. It means adapting to the feedback. Finally, track what matters. You want to know, are people signing up? Are they sticking around? Are they buying? Are they telling others? That's what really matters. If it doesn't work, you don't need a fancier design. You need to learn, tweak, or pivot.
I don't think it's controversial to say that Italy is a bit behind the US when it comes to building startups, but we are catching up and we're seeing so many founders get at it and try to apply the method and do it also in different ways. So I just wanted to go back to fundamentals and ask you questions about the core elements of the lean startup method and also see, since this is the beauty of being behind is that sometimes you can leapfrog some of it, right? You can jump ahead and especially with AI, we're seeing people move so much faster than we've ever seen people move. So I'm curious to hear from you also how you would change some aspects of the method now that AI is around. And I'm sure you've been asked these questions daily in this particular moment and I know you're working on your book. I don't know what it's about but I wonder if it has something to do with this. Let me start with MVPs.
Sure. We've talked about MVPs, minimum valuable products, a lot on this podcast. So I hope everyone now knows what we're talking about. But since you're the inventor of the term and I have you on, I'd love to hear from you a bit how you would define MVP and particularly what does minimum mean? What does valuable mean and what should a successful MVP look like essentially?
So, MVP, even here, even in these events where people have just read the book and they're so excited to come meet me and talk to me about it, still people really struggle with this concept. It is not intuitive. It is not easy to do. And for people who've just heard it secondhand, a lot of people haven't actually read the book, they haven't really thought about it. They just heard, 'Oh, it's like a cheap simple version of a product.' They get it wrong quite often. So, let's break it down. First of all, MVP stands for minimum viable product. Those are the three elements. And minimum means we want to do the least amount of work to make the most amount of progress. It's just efficiency, right? Very simple. But that begs the question, well, what does it mean to make progress in a startup? And everyone thinks they know. Ah, it's to get users, it's to get money, it's to get revenue, fundraising, what? Everyone thinks they know what the elements of progress of a startup are. And those are all wrong. A startup is not worth anything because of the customers it has or the money it has or the fundraising it has. The only thing that matters in a startup is what I call validated learning. Scientifically true facts about whether the business model works or it doesn't work. And it is actually why startups are so confusing because we think it's an economic thing. It's a business. You should be able to value a business by its assets, discounted future cash flows, and all this other stuff. Startups don't have any of that. It's too early for that. When they're big, of course, you can value them that way. But at the beginning, they have nothing but their probability and magnitude of future success. They're like a lottery ticket, you know. And people find this very confusing. If I tell you that I am going to build a company, it's going to one day be worth $100 billion. And you say, 'Well, that's not very likely.' I say, 'Sure, but you can agree that I used to have a 1% chance to become worth it?' Oh, sure. Well, if I have a 1% chance to be worth $100 billion, that company, that lottery ticket is actually worth a billion dollars already right now. That's where startup valuations come from. So people get so fixated on the material elements of a startup they miss the most important thing which is what do we know, what do we learn about what customers actually want. And this is why when people quit their job they often can make a new startup or like a startup implodes and people come right back with a new one. It happens all the time. Why, how is that possible? They don't have the factory. I was just talking to an incredible founder who makes a product in the beverages business. Coca-Cola bought his company, made a lot of money. He was doing great running the business inside of Coca-Cola. And then one day, Coca-Cola just decided to shut it down for no reason. Just was like, 'Nope, not doing that anymore.' And within six months or a year, he had a new beverage company, you know, rebuilt the whole thing from scratch and is doing great. It's like, but he didn't have any assets, no customers, no brand, no contracts, no revenue, no employees, no nothing. But what he had was an understanding of how the business works. So that's when we think about viable, that's what viable means. Is it in fact revealing something critical about the probability of future success? And people find that very confusing also because it's like when I talk about learning, it's like, oh, I can say I learned anything, you know, I can read a book now, I learned something. No, it's a very particular kind of scientific learning where we validate and test a hypothesis in a rigorous way. And then we put product in there because a lot of people are taught to do market research before they build their company, which is a giant waste of time. Market research, most market research techniques are wildly inappropriate for a startup. And I was counseling a startup the other day. A founder came to see me at an event and his questions were all like what about the total available market and B2B or is it B2C or is it this or this buzzword, that buzzword. And he was just like a barrage of buzzwords and I was like, you have spent too much time with consultants, my friend. Consulting has rotted your brain and you need to stop. There's no B2B, there's no B2C, there's no TAM. Do you have a product and do you have a customer? And because he spent so much time on market research he actually hadn't learned anything about customers. So the reason we use product is: what is a product? What does it mean to make a product? People are like, I know what a product is. It's an iPhone. That's a product. It's like, well, is a website a product? Is a book a product? Is that microphone in front of you? Like, what is a product? A product is anything a customer experiences. You know, if you go for a massage, is that a product? You didn't buy anything. You had an experience. Well, that's a product. If you have a branded item, you know, why is soda water worth a lot more than water? No reason. It's not because it's more expensive to make. It's just because of the brand perception of it. So anything a customer can experience and is willing to make a voluntary exchange of value, that's a product. So the critical thing is to put customers to the test to discover how they will actually behave. So people are always like, I talked to 10 customers and they all said they love my product. Say, oh really, you asked someone like, is my baby beautiful? And they said yes, you go right to their face. Uh-huh. I'm sure it's very reliable. And if you dig in it's like, well actually five of them were my cousins. It's like, okay, I need to know what actual people think of your actual product. That means not interviewing them, not meeting with them. You need to ask them to buy it or to do some other action that costs them something so you can figure out if they actually do it. And the shocking truth of almost every startup I've ever met in my life is that most of the things they do do not work. So when you strip away all the consulting and the business plans and the fundraising and the this and the that and just be like, have you actually sold a single thing to a single person? The answer is often no. So MVP is a way to make rigorous the idea that we are doing an experiment. We are truth seekers every bit as much as a religious scholar or a scientist with a particle accelerator. We want to know what is the truth of these underlying assumptions that make our business valuable or not. And once we find the truth then we can start to apply some of the other techniques to grow it. But we start first with the bedrock of what is true.
So in order to find that truth, what your theory is, or so what the method says is try to build the leanest, so the simplest way in which you can test this product, ship it out quick, put it in the hands of people, make sure people are using it and that they're liking it or giving you some sort of feedback so that you can continuously iterate and make it better and better and better until it's a product that a large amount of people can use or love or whatever your goal is. Whether it's like, I guess the hardest part for me and I think maybe hopefully for people is the lean part of it, like the minimum part of it. Like what is the minimum? Like for me the minimum is something that's perfect for you, it's something like, it still has... I sometimes feel people are embarrassed and they're like, oh this is too bare bones, it's too lean. If I ship it out nobody's going to love it because it's bad.
Or vice versa, you can trick yourself and say like, oh people didn't respond to it because it was not well built, because it was minimal, and maybe I should like spend more time to make it better and try. So it's really tough to like understand.
And it's very important for people to understand this. People think that MVP is anti-craftsmanship or quality. No, that is backwards. In order to understand what the word quality means, you have to know who the customer is. How do we do MVP? If we want to be high quality, there's two critical things you have to do. First is you have to take your ego out of it. Okay? What you think is quality is one important input. You know, and if all you care about is that, then you're an artist, not an entrepreneur. And you should just do art. Art you do for its own sake. And you don't care if anybody ever sees it or not. You don't care if anybody ever reads it. You know, it's fine. You're happy to work in obscurity for your whole life. And maybe you'll be discovered and become a great master after you're dead. That's fine. Entrepreneurs are not like that. 'No, no, but I want to be an artist and I want people to like it.' Oh, that's two things you want now. Well, that's the tension. So, we don't just go do whatever people say. We still bring that artistic integrity to what we do, but we test. We don't use it as an excuse. We just say, 'Look, I think the higher quality it is, the more people will like it.' Excellent. We'll show it to them as you make it, and you'll see the evidence. Easy. Second thing you have to do is you have to be much more disciplined about stating what your thesis is. What is your quality thesis? So, I say, 'Okay, I'm going to make a website, and it has to be really beautiful, otherwise people won't stick.' I do my experiment and they don't stick. Ah, it must have been more beautiful. I made this mistake myself once. And I spent hours and weeks and months making the experience better and better and better and better. And the problem was I finally had the courage to say, 'This doesn't seem like it's working. I need to understand why.' And I would sit... I actually had customers come into my office and I would sit them down. This is one of the most painful things you can do in your life. You make a thing and you watch people use it without talking. Just watch. And they would sit there and they would look at this beautiful product that I had slaved over and they just stare at it and do nothing. And I'd be like, 'Hey, um, you're here for a usability test. Could you download this product and try it?' And they'd look at me and they'd be like, 'No.' Like, 'I'm sorry. Excuse me. We're paying you to be here, so you have to do it.' And they'd be like, 'No, you can keep your money. I'd rather not.' Like, it was so bad we couldn't pay them to use it. Now, was this a quality problem? Was it because it wasn't beautiful enough? Was it because the design was flat? No. The fundamental value proposition which they could infer from the very first web page already was so broken that they didn't even want to click onto the second page. So all the work I've been doing on page 10, on the 40th hour of the experience, like I've been doing all this other quality work that they never saw. It didn't matter. So you have to be willing to actually take it in stages. And you know people say, but Steve Jobs never did that. BS. Go study his actual history. He was an absolute control freak, quality obsessed perfectionist. And like he shipped the Apple I, he shipped the Apple II. Like he did a bunch of stuff that was ugly as hell that barely worked. And my favorite story is the actual development of the iPhone was like a janky ass MVP. Like you couldn't believe. Now we remember it as this instant breakthrough. But at the time that he stood on stage and did that famous demo, they didn't even have a single working iPhone that he could show off. And the whole thing had to be faked to make it seem more real than it was so that he could get the pre-orders, so that he could generate the momentum to actually finish the product. Anyway, I feel like really taking quality seriously is about really understanding who is the customer, really understanding it through their eyes. What will they see as quality? And then being willing to put that belief to the test. If nobody had pre-ordered the iPhone, they never would have finished making it. That's okay.
And let me bring AI into this just as I said in the premise. I think it's changing things mostly in my eyes, and then you can correct my assumption too, is that it is easier to do things faster and more beautiful actually because you have someone who does it for you. You don't have to have a whole team of designers. You can quickly like write, especially if you're doing something cosmetic, right? Like make this logo prettier. I don't know. You can do a lot of it fast. So, does it make MVPs more important or less important?
Oh, much more important. Okay. Well, let's be really clear. Let's be super specific about what AI can and can't do today because this is also a very common misconception. First thing is the early evidence is that AI is much better for bringing the worst performers up to average than it is about making average performers exceptional. In fact, I don't think there's any evidence yet that this makes people exceptional. So, for graphic design, for interaction design, for the things that people talk about the fit and finish of a product, if you want to get to an exceptional level, you can't delegate that to AI. You just can't. What it can do is allow your crappy wireframe test to be somewhat passable and okay. And that's actually a really big deal for engineers who don't have any design skill who are like, can you just give me something that looks vaguely plausible so that I can actually get the feedback that I want. So it actually takes one bottleneck out of the development process that's super good. And that's true for written content also. AI is a terrible writer. We're already learning AI-ese. Like people have already become allergic to the em dash because AI is full of it. So got to be careful with that. Don't let AI write for you. It's a terrible idea. But it can really increase your productivity as a writer to, you know, take an outline and turn it into proper prose. It can help you critique, especially if you're working cross...
languages. It can help you figure out if your translations are good. Like it just it can be very a very good feedback partner. And of course, it's excellent at research as long as a human being double-checks all of its findings because it still hallucinates. So those are really useful skills to accelerate the creation of an MVP. But there's a downside, a really nasty downside that I don't think people are talking about enough, which is that vibe coding and this idea that AI is something you delegate a task to which it completes and you give it feedback and it that pattern turns people's brains off. The whole reason we do MVPs, the whole reason for lean startup is it's so easy to convince yourself that you're making progress because you're cranking out code. But code is not progress. The progress is in here. You cannot outsource thinking. You cannot outsource learning to the code or to an AI. Makes no difference. You have to actually do the hard work of forming a hypothesis and testing it. And the AI can help you if there are stumbling blocks. It can help you, you know, be a sanity check. But you, it is not something where you want the that soporific effect you get when you're like, ah, I'm just in the flow. The AI is doing all the work for me. Um, that is actually really toxic and I think generally people need to to train themselves to avoid that. It's very tempting with these chat bots to get into that mode and the evidence is just not there that that works partly because people make crap with it and you ask it is what I'm making good and it's like yeah it's great and you're like oh
I was going to I was going to make that I I use it I'm I'm not a a software developer but I use it for writing because that's what I do. I write. Um, for me the trope is that I I give something to let's say ChatGPT and I'm like can you make it better and can you make it more catchy for instance this would be my prompt right so I'm already saying can you make it better which is a trick you're tricking yourself and then it gives me something that is more catchy let's say and then I read it and I don't like it and then I question myself I'm like is it because I don't like catchy writing is it because this is actually bad but because I told this machine can you do it better then now I need to believe it that it's better. I don't know if I'm explaining myself, but then I'm like, "Oh, it must be better because a machine told me it's better." And then I'm using it. I'm like, "It's not actually that good." And the truth is it's probably not good. Like, and it's better to write in my own voice because then I sound like everybody else.
That and that's the problem. You become you become a carbon copy of everybody else. I listen, I use AI every day and it's made me far more productive as a writer in a bunch of ways. Like I use it very effectively. So, I don't want to say I have nothing bad to say about it like as a tool as in terms of its capability. And we're we're we're going to do a course on this um at Answer.AI soon. So, people should should check it out, which will be announced shortly um to teach people this like anti-vibe coding technique. But, I'll give you the short version of it, which is like in order to use it effectively, you you have to as the human, you have to break the problem down into a series of subpros. The more like these things have these incredible context windows, but they have very limited attention span. So you need to concentrate. It's like, you know, you're like spreading the intelligence they have over too wide a surface area and you get crap. You have to concentrate.
And it can help you do that. If you're having trouble like do the scaffolding. It will help you. But but generally speaking, get it to a discrete task. Then what I often do is I will if if I'm it's I use it for writer's block. If I'm stuck, if I'm if I'm like looking at a passage, I'm like, I can't figure out how to improve this. I don't know what to do. I'll be like, "Give me an assessment of this paragraph. What's wrong with it?" And it'll give me its opinion. If I don't agree, I make it do it over again. You're like, we'll go back and forth. And then I'll be like, "Okay, let's come up with something better. Here's my plan for how to make it better. I think it should be more like this." And I'll give it like and I don't with typos and just a little bullet like here's bullet points. Like help me make a plan out of this. And it's great at that. If you give it the spark of something intelligent, it can kind of like format it and help you and it makes it into a to-do list. It's like okay, here's the outline. Anyway, we'll go back and forth, back and forth, back and forth, making it better, better, better. And and I'll finally I I don't really think too hard until I get its assessment that it thinks it's created something that's a 10 out of 10 good. And then I'll be like, "Okay, let me take a look at it." Like, seriously. And I'll be like, "Oh, this is horrible." It's usually really bad. So then I'll be like, "Okay, I'm going to revise it now." And I'll erase all the stuff. I'll go way back earlier into the conversation. Erase all because like the more the more time it spends on something just like humans the more invested it gets in it and the less objective it gets in evaluating so I throw all that context away take the final product we made I'll then adjust it to put it in my own voice and I'll say what do you think of this give me an honest evaluation about this and often times it'll be like that's terrible I'm like okay let's do it again now how can I make it better so like I will do as many rounds of that as necessary so that it and I both agree we got something good and then we'll turn that into pros and then we start over again. Okay, here's the first draft. That's terrible. Okay, how about this? How about this? Going back and forth, back and forth. Get it to something that it thinks is 10 out of 10. And then I'll go in and line edit it and be like, "No, this is terrible." And like put it in my And like, so it it sounds so cumbersome. But remember, it we started with something where I was stuck.
So that's why it's a productivity unlock. If you know how to write something yourself, never have an AI do it for you. Because like the the thing people crave more and more and more in a world dominated by AI slop is people's honest, authentic voice. Never give that up. But if you get stuck, as we all do, every writer in the world is always has writer's block. It's a real thing. You get stuck, let it help you get unstuck and then go back to to writing with your own voice. It's it's a skill that that we're all going to have to master going forward. It's very important.
And I guess this also applies to code and you'll correct me if I'm wrong, but instead of asking, can you write me code to create a whole website about this? Or can you write me an introduction to this podcast episode for instance, instead of asking that broad question, do it yourself and then break it line by line and be like, can you make the sentence a little shorter? Like you give it a very specific task on a very short sentence or line of code so that it's your product, but then the single elements can be improved. Then in that way, as you say, you understand how you're breaking down the issues.
No, you got it exactly right. Yeah, we we um you know, um so that you can see every element of the computation and it's always 100% human in the loop at any point. You can go back and correct or change something that the AI did. It's like fully transparent to you as the user. You can delete something if you don't like it. You can change it. You can do it over again. You can rerun it. And it mixes AI prompts and Python code interchangeably. So you can use it to work on pros. You can use it to work on code any anything can be rendered as code. So then you can use it to work on presentations or you can make video games. People made all kinds of really cool stuff with it. But I think that that human in the loop um solving problem like it's a problem solving tool not a coding tool not a writing tool.
