Here’s the full transcript, unedited after we had it transcribed—which means it has some bugs:
Hello, everybody. My name is Dan, and I work with two other guys. We run a tiny little software consulting shop in Boston, called Hut 8 Labs. We work with startups in a bunch of different ways. What I’m going to talk about today is about what I consider “the first principles” underlying Lean Startup. I want to do this because I think understanding these principles, understanding what’s going on under the surface, is a huge competitive advantage. It lets you apply these ideas—which are very powerful—very broadly to different industries, different stages of growth for a company, different types of companies. But to do that, you need to understand why these ideas work and when they don’t work.
Before I get started, there are some interlocking sets of ideas I want to walk through, but I want to start with Eric Ries’s definition of a startup: it’s a human institution, putting a new product or service under conditions of extreme uncertainty—extreme uncertainty. Hold that in your head. Here we go.
The first concept I want to talk about is opportunity cost. To talk about this, I want to tell a story about Hut 8, my consulting company. I want you to imagine that we’re trying to figure out what we’re going to do with our next month, and we’ve got two people who want to hire us, and they want to hire us to do just kind of coding work, which we actually mostly don’t do at this point, but imagine. Client A: they have a bunch of Photoshop documents, and they want to turn those into HTML, CSS, cut them up and they’re happy to pay the market rate for those, which is roughly $20 an hour. But there’s good news, because they have limitless numbers of these documents, now all three of us can get eight hours of work a day at this nice $20 an hour rate. That’s client A. Client B is big data: It’s Adobe, it’s H based, it’s very sexy, and for that they pay the market rate, which is $200 an hour. But there’s a catch. They only have four hours of work a day for all three of us if we’re going to go there. So A, B. So the question is, What do we do? Which one do we choose? Which of these clients do we work for?
And I want you to imagine that one of my two partners—let’s say my partner Matt, let’s just say, which he never would–is like, “Guys, this is really easy. There’s no problem here. We have to do the Photoshop work.” And if Edmund and I were to say “Well, why?” and he said, “Well if we take the big data work, we’re going to be sitting around for half the day not working. That’s super inefficient. That’s like a waste of our time. If we do that, it’s going to look really bad if we do that, in fact. Whereas if we do the Photoshop work, we’re going to be working steadily and efficiently and we’ll make $160 a day.” That’s pretty solid. Right? Okay.
Is Matt right? No, he’s totally insane. Right? That’s a terrible idea. Look at the math. We’re going to make $800 a day per person with the big data work and only $160 the other way. In fact, Edmund and I would be perfectly justified in saying to Matt, “We’re not making $160 a day; we’re losing $640 a day.” That’s the difference. We’re giving up to opportunity to make the $800 in order to make $160. And that’s opportunity cost per person. And my first message is: this is happening at your startup, like right now. People are choosing to work on stuff that’s not that valuable–that is not the most valuable thing they could do–and it’s costing your company a lot because you’re giving up the opportunity to work on other stuff. You might be thinking, Okay, well sure, but I doubt it’s this huge difference, like this $2,200. Because the stuff we’re working on is pretty important. And so, we can answer that in an interesting way.
Imagine a relatively typical startup–typical in some sense–raises a series A, hires 10 people, and one year later goes back to raise a series B. And I want you to imagine there can be two outcomes, and how they’re valued. Not how much they raise, but what their valuation is–what somebody is willing to give them a big chunk of money thinks the overall company is worth. That’s a really good proxy for how much the value the individual is actually generating. Down one path, they worked on pretty valuable stuff, and they’re being valued at $20 million. Down the other path, things didn’t go as well, and they’re valued at zero dollars. One way you can look at this is that the people did things: if you imagine, that’s almost like the same startup that went down two different roads. The time it went down the road where they didn’t make any money, one way to look at that is the opportunity cost is $8,000 a day per person because they were working on the wrong stuff. Also known as $8,000 a day. This toy example, the $640 a day, is nothing.
And by the way, you’re paying these people maybe $500 a day, just back in end flow. There’s this tremendous cost to working on the wrong things. And you might be thinking–if I were you, I would be thinking about this example–two things. One is, there’s a fair amount of luck involved. Well that’s true, but my belief is it’s not all luck. And one thing that’s nice about this huge $8,000 thing is, even if there’s a lot of luck, even if it’s half luck, it’s still $4,000 a day because you’re working on the wrong stuff. And the other action is: there’s a little trick here, which is it’s mostly not about who works harder. And I have what I consider fairly strong evidence for that, which is who here has worked for that startup that failed? Anyone? Raise your hand. Did people work hard at that startup? Yes. [chuckles] Always, people work super hard at startups that fail.
And so at some level what I’m saying is, if hard work and luck are important, but they don’t seem to really distinguish the ones that succeed from the ones that fail, then the choices of what we’re working on must be critical. That’s actually your biggest lever, is what you choose to work on, and it has this huge differential effect. And so, one thing I’m saying is you should be actually terrified. You should be very, very scared of working on the wrong things. I would say you should be so terrified that you actually don’t work. You’re not sure that what you’re working on is the most valuable thing to your startup, you should stop working. And I tell people this and they think I’m exaggerating, but I’m not. You should only work if what you’re working on is the most valuable thing.
And you really shouldn’t worry about working hard. And especially, what I mean by that, is you shouldn’t worry about looking like you’re working hard. And everyone gets this wrong. Because human beings, when you get them into groups this way and there’s these conditions of extreme uncertainty, it’s very hard for them not to work. They actually do the thing I describe Matt as doing, that we all made fun of, which is they feel like if they’re sitting still they’re going to get in trouble. If they can’t show their investors or their boss that they were doing something at all times there’s going to be a problem. This is an opportunity for you to try to fix this.
But there’s a problem. Now, we get to section two, which is Information and Money. In my little toy example, we knew if we do this thing we’re making $20 an hour. If we do this thing we’re making $200. Now, at a startup, no one tells you that, no one says, “Hey, you do this marketing campaign, it’s worth $8,000 a day. You work back in the database, that’s worth zero dollars a day.” You don’t know what’s worth what, right? So what do you do? But you should be afraid that you’re working on the wrong things. Well let’s go back to my toy example. I want to imagine the same setup with two potential clients: one with the Photoshop, one with the big data. But they won’t tell us which is which, and this is a funny setup. We know one is one, and then one is the other, but we don’t know which is which. So what do we do? Well we could work for one kind of at random, and half the time we’d get the really good one, and half the time we’d be losing this $640 a day. So, on average, we’d be losing $320 a day, actually. That’s how it works.
But what we should really do is spend a week researching. Let’s say we can be sneaky and we can talk to friends who work at the company or find out which one is doing a design job and which one is doing a big data job. And imagine that the end of one week, we know for certain which is which. Okay? So think about that. I want you think is, that week makes us money. That information that we gathered, it’s real money, and I’m really going to emphasize this. So you can think about it. Like before that week started, we’re going to randomly lose $640 about half the time. After the week is over, we know for certain which one we’re getting: we’re always going to make more money. Therefore, that information that, yes or no, which one is the better one, is worth a lot of money to us. In fact, it’s actually useful to think about, and it’s super important not to think about, “Oh, we have to do this thing. We have to do this research, and only then could we start to make money.” Think about that week as making you money, because, information in the sort of content when you have a decision to make is worth money. The same way that a contract that you sign–that you only get paid on when some later event happens–is worth money. You don’t consider that contract to be rigmarole until you actually have the dollars in your bank account; you consider it as money as well.
So now we get to the next idea, which is risk in information. So this is interesting, I’ve had this idea that there is information as value when you’re making a decision. But startups never had a single decision, right? In fact they have what I call chained risks, that a whole sequence of things has to be true. And if they’re all true, some level of the event happens for you, your startup is worth a lot of money or you get lots of customers, whatever, but they all kind of have to be true. And so the classic, very simple one for essentially every startup is, can we build it? And if we can build it, will they buy it? Which are often called a technical risk and market risk. What’s interesting here is, well, what’s the value of knowing the answer of one of these questions? Right before, we had this idea that which question should we answer first? Right before, I started talking about this very simply: yes, no.
Well, to talk about that–about these chains of risks–I want to talk about two different kinds of startups, two imagined startups. The first one of imagine is building a teleportation device. And you can imagine they’re some people in MIT, and they have lab, quantum mechanics. And they can take an entire molecule of salt and teleport it across their labs to the other side, and they’re very excited by this, as they should be. And they come to you, and let’s say you’re an investor or early employee, and they say, “If we can just spend a couple years researching–maybe three years–we’re fully confident that we can teleport a human being very safely anywhere on earth for $1,000.” Yeah, that sounds great, that sort of thing. Startup B is very different: it’s just this enterprise CRUD app. Does anyone know what a CRUD app is? A CRUD app is like the simplest, dumbest kind of app an engineer can make, a very simple app. This startup has a doctor friend who works in hospitals–let’s say, a health IT. They have this simple app that will take me three to six months to build, and they’re convinced that hospitals will pay $10 million a year for it.
These two startups both face the same question, which is: what should they do first? They both have to prove how whether or not they can build an amount of people to buy it. The question I’m going to ask is, Well what do you do first? What should they each do with their first month? The answer to the teleportation one is really easy: they should try to build it. In fact, if what exact happens is the CEO of the teleportation thing came to you–the early investor–and said, “Look I’ve done the lean startup thing and I built a really pretty brochure, and I printed it up. It’s really nice. And I took it to a bunch of people, and I actually got them to pay me $1,000 for when we’re ready to do teleportation. Wasn’t that awesome?” You’d be like, “No, you’re totally fired. That’s incredibly useless. Why would you possibly do that? Go back to the lab. Maybe see if you can teleport like two crystals of salt, or a shaker, or a rat, or something, right? But who cares that you sold this thing? That’s totally stupid.”
And whereas with the CRUD app, it’s the opposite. If the CEO was like, “We had this crappy demo, but, man, it was crappy.” So he spent a while building a really nice database, and that’s all really good, and we’re ready to scale it up, and you’re like, “Have you talked to any hospitals?” And they say, “No, no. We haven’t done that.” You’d be like, “You’re fired. That’s a terrible, terrible idea. What are you doing? Go to the hospitals. Because $10 million, why are they going to pay it?”
Here’s my question for you. What’s different about these two stories? We have profoundly different intuitions about what we should do in these situations. Intuitions which are correct about why we should do different stuff. But why? Why are these so different? The answer I’m going to give is what I’m going to call the degree of surprise. Okay? So when somebody finds out that a human being will pay $1,000 to be teleported anywhere on earth, you actually already knew that. You’re not surprised. There’s no surprise in that. Correspondingly, when the person building the CRUD app has got it partially built in a month, you’re not particularly surprised by that. One way to say that is, with the teleportation thing: essentially, you already knew, mostly, you weren’t totally certain, but you were almost certain that people would pay for it. So finding out that Yes is actually not very much information.
There’s this idea that information is actually equal to surprise. This idea of surprise is what I’m talking about. There’s actually a very nice mathematical theory behind this, by Claude Shannon. If you like math information theory, it’s a beautiful, beautiful thing. But the key idea is that we only get information when there’s uncertainty and risk. So basically there’s one way to look at that: you can only be surprised when there can be something you don’t know. If you largely know something, there’s no surprise, there’s no information. Now looking back to what I’m saying a moment ago. I was talking about the value of information in the context of decision like you can put a dollar figure on how much information is worth. What I’m talking about now is actually the amount of information. Those are sort of orthogonal, which is kind of cool. Basically: you actually don’t get much information when you already knew something; you get a lot when you’re uncertain. And then, what information is valuable depends on what decision you’re making. So those things are sort of related.
Now, I want to switch to Steve Blank’s definition of startup for a second here. A startup is a temporary organization formed to search for a repeatable and scalable business model. So, one way to understand that is that when I described that sort of little toy story about Hut 8–where we did a week of research and then a month of actually getting paid for something. A startup is just that week, right? A startup is a startup because it is an information-gathering entity. That’s what makes the startup. And part of the reason–I want everyone to understand these principles rather than just the tactics in the specifics of what you do–is this is entirely true. But unfortunately, there’s not one day that you wake up, and you’re no longer a startup, and now you have a repeatable and scalable business model and you just execute. It doesn’t work that way at all. It’s like you gradually get more and more evidence that you’re on to something, and you see your business changes gradually overtime until at some point it’s more executional. And then of course you have to figure out how to be information-gathering again. And part of the reason these principles are so valuable to fully understand is that it lets you operate carefully as you’re moving along that spectrum, and it lets you help your teams figure out how to operate at these different points along the path.
So now I want to talk about information in time, and here I want to talk about rates of change. So one way to think about rates of change is like speed: with a car, velocity is measured (in high school physics) in meters per second. Or a car has a speed that’s in miles per hour. And one way to look at that is in terms of the units. Meters is a measure of distance; seconds are a measure of time. So, when I say that about rates: velocity is a measure of distance over time, distance per time. Revenue–something that companies really care very deeply about–could be measured in dollars per month or dollars per year or whatever. Which is, therefore, a measure of money over time. How much money are you making in a given unit of time, and revenue, is very, very important. So profits are the key, but revenue is key.
But I’ve just said is that you really want to think about information as money. At a startup, when you are in a search, information is the primary form of money. That’s how you are actually making money, getting more value for your startup is by getting the most important information. So therefore, the revenue is really best thought of, is information over time. So what I’m trying to say here is that the thing you are trying to kind of sort of make your whole company function and the way you make your company kind of make more money is more quickly gather information, and not just in some vague sense, but sort of valuable information. How fast can your team gather information? That’s the sort of key thing you want to go after. And again, real money.
So now we get to sort of putting a bunch of stuff together and this is risk information in time. This is sort of the point of the talk. To be honest these ideas sort of all fit in a certain way and I want to kind of get here. And so, this is also the part where I think it’s fun, and I’m going to tell this almost purely in a story. So let’s go back to health IT thing. You got this demo thing, or you’ve got this crud app you’re building for someone for hospitals. And I want to say instead of just being a crud app, I’m going to adjust it slightly for the story. And imagine there’s a public data source that you’re using in addition to the crud app. Something that like Todd Park who spoke last year. They’ve put out on some data set about doctors. And you’re taking that data setting, collecting it and doing something with it and then presenting it to the hospitals. Okay, so imagine that. And you’ve done the right thing. You’ve actually done a demo based on that. You sold it to hospitals, and in fact they were willing to pay. You’ve got one hospital that give you a check for $10 million, or a contract or a promise for $10 million. And that’s great, you did the right thing.
So now your sales team is out there trying to repeat that and sell the second one and you got a bunch of engineers now building that thing. And I want you to imagine something. I want you to imagine a junior developer, someone on the team, bright guy but young – guy or girl. And some morning – it’s a Thursday of morning – and they were given a job of taking the demo app and turning it into a real production system. And they’re working with this public data set, and they discover, to their surprise, that it’s not as comprehensive as everyone thought it was. It worked well for the demo, but for the actual hospital, it’s actually not going to work. The whole product that they’ve sold is actually not going to succeed the way they’ve done it. They have to do it some other way. In the moment after this person makes this discovery, the biggest risk for the startup has changed. The biggest risk is no longer, can we repeat this sale? The biggest risk is can we actually build the thing that we promised in the first sale that we thought we could build, but we just discovered we were wrong.
If the biggest risk has changed, the thing you should be doing to gather the most information has changed. Because the way you gather the most information is by going after the biggest risk. Therefore, the thing that’s going to get you the most information, and therefore, the most money has changed. Therefore, as long as the company is still doing what it was doing before that discovery was made, they’re doing the wrong thing. And one way to look at this is that in order for your company to move fast, the entire organization, the thing that will limit them in how fast they can move and how fast they can make money, is how fast they can respond to the changing nature of risk. Because it’s only by going after the biggest risk do you make the most money, and risks are changing all the time, the entire organization has to be able to change direction. And this, really, nobody gets this.
This is the competitive advantage I want to give you. If you can organize your company this way, it’s a huge competitive advantage. I want you to think about the story I just told about the junior developer. I’ve been that junior developer. I’ve worked with them. Nine times out of ten, they have no idea why they’re doing what they’re doing. They were told, collect this data, clean it up some, put it in here, we’re already behind, work late, get it finished. They don’t say it to anybody, this isn’t comprehensive data. They don’t even know what it’s for. Or if they do, somebody tells them to stop talking and get this thing done because they’re on deadline and they’re behind and they promised a lot of stuff. And then, there’s these periods of weeks or months where everyone is working on the wrong thing, because people aren’t paying attention to the sort of what reality is telling them about the changing nature of risk. This is why people work on stuff that loses them $8,000 a day, is because they’re not focusing on the things they need to.
In summary, in the presence of extreme uncertainty you make money by extracting information from reality. The most valuable information is that what reduces uncertainty behalf of largest chain of risks. And to acquire information quickly the entire team, the whole organization must constantly adjust its understanding of risk. Now, there’s actually turning these principles into practice is fascinating to me. This is something we, Hut 8, think about, work on a lot, and we work with our clients sometimes on this kind of thing. It’s a very, very interesting question and it’s sort of non-trivial. How do you get the entire organization to work this way?
But I’m actually not going to talk about that right now. I want to close with something else which is what I think is the sort of most important thing to understand, most important sort of barrier to getting these things to work if you have to say a group of human beings operating in conditions of extreme uncertainty which is fear. And fear is a profoundly important driver in how people behave in conditions of uncertainty. The story about– one thing is, the team will be very afraid of looking like they’re not busy because people interpret that and they’ll tell you this, that it looks like people don’t care. “Don’t you know how important it is? Why you’re not working long hours? I’m the wrong thing.” It’s very, very powerful. [laughter]
They will hurry, they will demonstrate to you how afraid they are by being kind of sloppy and moving fast, but the most important place that fear sort of lodges itself is in the heads of your CEO. Specifically, one of the things – I’ve seen this over and over again in startup CEOs – is what I just described. They have a lot of fear and what I described as startup, understanding the current state of a startup as sort of a chain of risks, which then have little risks within them. Can we do this, and if we can do this, what about this other thing and that requires this and there’s that sort of nature of that, whatever. Startup CEOs don’t call that a chain of risks even to themselves, what they call it is their vision.
They have this vision that all those things will be true and in fact they can’t get out of bed in the morning sometimes without kind of literally convincing themselves that the things that are most unlikely to be true or somehow going to work out and be true. What they will do, startup CEOs tell me this, and they’re do it like, “Oh my god, I should do a whole presentation about how you’re thinking wrong” is they say, “I can’t let the team know that this thing might not be true because if I let them know that they’ll be demoralized.” Oh my god. Like really, what you’re doing if your team needs to go after the biggest risk to gather information and you’re preventing them from doing that because you want to pretend those risks don’t exist, you’re forcing your team to fail, and I promise you, you will blame them for not executing at the end of the day. So basically, don’t do that. That’s my message. [laughter]
So one of the things I want to tell you is go change your organizations, don’t work this way, get the whole organization sort of oriented around information and risk in this way. It’s a huge advantage or as you may find start your own team, start your own organization if you have to. So that’s my main talk for today. We talked about some of those principles on our blog and I’m always happy to talk more about this. So, thank you. [applause]
