How to Start a Startup: From Idea to Innovation

1. Introduction: Startups, are how innovation comes into the world. If you look around yourself, and look at all the objects that are near you, all of these things are made by companies, and new things tend to get made more often by new companies. So, we all as a societies benefit from entrepreneurial activity. I'm Albert Wenger, I'm a managing partner at Union Square Ventures in New York. I'm going to teach a class about how to start a startup, how to go from nothing to something, how you take an idea, how you validate that idea, how you turn that idea into a minimum viable product, and how you start building an organization, and all that. At Union Square Ventures, we have an active portfolio of over 70 companies. We have invested in over 100 companies, and we've seen over 1,000 companies that we've looked at before we made investment decisions. Much of what I will talk about today is drawn on that experience, the experience of things that have worked, but also the experience of things that haven't worked. Both from things that I've invested in, and mistakes I've made myself as an entrepreneur. I love working with startups because and startups are inherently optimistic. You wouldn't want to try, and fight the odds of bringing a new product to market if you weren't an optimist. So, I get to work with optimist in what I do, all day,, every day. That's really infectious. I'm excited to get going. So, let's jump right in. 2. Generating Ideas: When I talk about startups, I'm talking about the kind of companies that I have experienced with from my role at Union Square Ventures as I mentioned investor. These are companies that build products or services that are technologically-enabled, that is as opposed to human services like a consulting service or a Physical Therapy Service, and the second characteristic is they're designed to grow fast which is also usually why they need to raise financial capital. You should keep that in mind when you think about the advice that I'm giving, because if you're starting let's say restaurant, not all of my advice may apply to it. The very first thing you need is an idea. What is your company going to do? Where do good startup ideas come from? In my experience, there are different sources of great startup ideas and then there are some ways that are not so great at coming up with a startup idea. The easiest one is what I call Scratch Your Own Itch. It's literally a product that you want to and are going to use yourself and somehow you're frustrated by that product not existing or not in the form that you want it to exist. A great example of that is a company that I actually helped built called Delicious. Joshua Schachter was working at Morgan Stanley and he was somebody who bookmarked a lot of things on the Internet. He was very frustrated by the fact that, not only was it impossible to find your own bookmarks inside the browser, but also they were inside your browser on one specific machine and if you were at home, you didn't have those bookmarks. Also lots and lots of people are bookmarking things and why can't we see what they're bookmarking so that we know what's interesting on a particular topic. So, those are all things that Joshua wanted for himself and so he built Delicious to do exactly those things. He had a service up and running before he even had a company. In order for Union Square Ventures to invest, Delicious the company actually had to be incorporated first. So, this is a classic example of Scratch Your Own Itch and it was quite successful after Union Square Ventures invested in it, Yahoo acquired it. Unfortunately, Yahoo didn't do a great job of maintaining that wonderful service. The second great way of coming up with a startup idea is to have a clearly identified need. A good example of that is Etsy. The founders of Etsy were actually building a community website for the crafting community as a service project for somebody else. Somebody else was paying them to build this community website and as they did that, they realized that one of the things that a lot of people in that community were talking about was, how they didn't like selling on eBay and how there wasn't a venue where you could just sell handmade or craft goods. So, they threw this other project that they were doing, came to realize this need and they came to understand how big a neat this was and then they built Etsy to meet that clearly identified need. Another example at similar is Twilio. Jeff Lawson was actually building a store, a physical retail store for skateboard equipment and he was a programmer though by training and one of the things that frustrated him to no end was the Telephony System trying to program a phone tree, trying to make the number roll over to different numbers depending on who was supposed to be answering calls and it all seemed very arcane and hard to do and expensive. That's how Jeff identified the need for a programmable phone number and that was the first product that Twilio offered. A third great source of startup ideas is actually failure. What I mean by that is, you are actually working at a company that's trying to solve a problem and it's not quite solving it and you have a better idea for how to solve that problem or you're working in a company that's solving a problem, but because all companies can do only so much, there's an immediately adjacent problem that the company doesn't have the resources to solve, but you understand it incredibly well and that you can go solve it. A good example of that is, we were investors in a company called Tacoda, that was started by Dave Morgan. It was a company, the early days of Internet advertising. Dave had built a prior company in advertising that had been acquired and after the acquisition, they weren't able to invest in new product development and he saw some clear needs for different ways of targeting advertising that wasn't being met. So, he started Tacoda to address that targeting need. Another example where something was born out of the limitations and the failure of the existing vendors, is Stripe which is a very large company today. People were able to accept credit cards on the Internet before Stripe, but all those prior payment gateways were very difficult to program. They required a separate merchant agreement which was a long process by which you had to apply and it took days, sometimes weeks to be approved and what Stripe did is basically created a very clean API and did away with the merchant agreement, that way people could literally get started overnight instead. So again, they looked at what was out there, they saw how limited it was, and they developed something that was much better along several different dimensions. One of the potential source of startup ideas are actual technological breakthroughs. An example of this is a company that we've invested in called Clarifai. Matthew was a PhD student at NYU. He built a breakthrough image recognition neural network that outperformed those of many other research teams, many of them having 20 plus researchers in them. So, Matthew had come up with a breakthrough way of recognizing objects. There are many other examples of breakthrough technologies that make the foundation for a startup. The challenge with going that route is you want to be very careful to not confuse a technological breakthrough which is a solution with the problem. So, when you have a breakthrough, you also still need to identify what the precise problem is that you want to point that breakthrough at. So, while breakthroughs can be a great source of ideas, you need to take that extra step of saying, "I have this amazing new thing, what can it be used for?" I'm a personal investor [inaudible] and Square Ventures in a startup that makes a carbon nanotube membrane. This is a breakthrough in making membranes because it can be extremely finally adjusted to what it lets through and what it filters out. But now, the important next step that the companies working on is, what are the applications? What are the different problems and what is the most immediate problem to solve with this membrane? That's an important second step. What then are some not-so-good ways to come up with a startup idea? I've yet to come across a startup where somebody said, ''Well, we did a brainstorming session and we listed 20 ideas and we talked about them and we threw out various ideas and then we came up with the idea.'' Even people who say, I've systematically studied 20 different applications and I've come to conclude that this is the best application, this is the thing that we should build. Even that approach which you sometimes get from people who have been in management consulting or who come out of MBA programs, often doesn't work. Why does this sort of brainstorming or mapping approach not seem to work? Well, it doesn't seem to work for the primary reason that it lacks the understanding, the deep understanding of a need. It usually operates at the surface where people go well. This clearly seems to be broken but it lacks the in-depth understanding of how it is broken and how it could be better or what the missing product actually is. So, I've seen that approach often result in things that sound like they're a good idea, they sound like they should work, but once people actually start to implement them, they find that there's no real need that they actually meeting. There's another reason why this approach I think is problematic. As we go along, we will see there are many ways things can go wrong in a startup. There can be many challenges. If the thing you're working on, you're only working on it because it came off a list where you kind of concluded that this was going to be a successful thing, but you didn't feel this need, you didn't experience it, you don't have a burning desire to solve that need. It's going to be a lot harder to be resilient and the face of all the challenges that you will have along the way. Sometimes people chase trends. They read about startups that are doing something and they go, ''Oh, they all seem to get funded, so I'm going to do a version of that.'' That tends to be not always, but it tends to be a bad way to proceed. Number 1, that people are getting funded will have been added for quite some time already. So, you have a whole bunch of catching up to do. So, unless you have a very precise way in which you think the way you're approaching it is different and better than what they're doing, you probably shouldn't chase something because if you're chasing it, you're likely to be late. Nothing wrong with saying, ''Hey, I'm going to build a new version of something.'' Like Facebook was not the first social network, there was Friendster and their was Myspace, but it was different and it started out as a different project. Chasing trends will most likely have you working on something that you're not actually passionate about yourself and where you will likely be late relative to the people who are actually innovating. Having said all that, I am sure that there are super successful startups out there that were created exactly the way that I just said to not do it. This is a common theme. You will get a lot of advice for almost everything. There will be exceptions to the rule and somebody will surely have been successful with the brainstorming or mapping approach. So, don't let my saying this is not the way to do it, if that's the way you want to do it, go for it, just with the caveat as to why I thought this is a hard way of doing it. At what point in your life might you have such an idea? I don't think that there's a specific H or prerequisite. People have great ideas at all different times in their life. The question is, do you feel prepared? Do you have the skills to go execute on the idea? What are some of the skills? Well, we'll see them in a second, they are the ability to create a minimum viable product, they are the ability to validate the idea, they are the ability to hire people that can come work with you. All those are things that you can learn on the fly, but it also helps if you've already learnt them by working at another company or another startup. Overall though, I would say, when the right idea hits you, you will know it, because when the right idea hits you, you will basically not be able to think of anything else other than that idea and why you are not already working on it. 3. Choosing Co-Founders: One question I get a lot is, I have an idea, how do I find co-founders to start this company with? I usually say two things. The first is, I say, "Are you sure you need a co-founder?" Because not everybody needs a co-founder. Jeff Bezos, started Amazon by himself. Now, having said that, it is still often good to have a co-founder. Why? First, the start of it is a tough challenging journey, and many times you want somebody who has an id with you, somebody you can rely on, somebody who is experiencing the setbacks with you, not as an employee, but as somebody who's been in it from the start as a co-founder. Second of all, because you may not have all the skills it takes to get going. Okay. So, how do you find those co-founders? Sometimes people go well, I'm going to systematically go look for them. That is unlikely to work. Why? Because you don't have enough time to go meet people that you can really rely on in depth, in all the challenges that you will encounter. It is much better to go with people you've already spent considerable time with, and ideally been through difficult situations. Those could be people you went to school with, those could be people you know from childhood, those could be people you worked with at another company. They don't even have to have been at the same company, it could be somebody you did a partnership with. Whatever it is though, ideally, it's somebody you have experience with not just when everything is going great, but also when some problem came up. The co-founder relationship will go on often for many years, and you will encounter many difficult situations. You do not want to find out that somebody's really terrible, and runs away at the first problem or turns angry, starts yelling at you or yelling at employees when things go wrong. Even if you think you already know somebody, you should still spend considerable time with them in the run up before formalizing the co-founder relationship. That is you should make sure your view of why you're starting this, what do you want to accomplish? Where you want to get to in the next year? That those views about very fundamental things are aligned. What should you look for in co-founders? Ideally, co-founders complement each other. So, if you are an engineer, you might want to have a co-founder. who has some business knowledge and vice versa. If you're an introvert, you might want to look for somebody who is an extrovert. If you have a very strong product vision, you might want to pair up with somebody who can be great at recruiting team building selling. Ideally, co-founders really complement each other, so that the sum is greater than the pieces. How many co-founder should you have? I generally find that two or three co-founders works best. Sometimes teams come, and they have five or more co-founders without anybody clearly being in charge, and that can be very, very difficult because one of the things that startups need to do, is they need to make a lot of decisions very quickly. If you have too many co-founders. coming to consensus, can be very difficult. A great example of co-founders who complement each other, and who knew each other very well, was the MongoDB founding team. It consisted of Kevin Ryan, Dwight Merriman and Elliot Horowitz. Kevin and Dwight had worked closely together at double-click. Dwight was a co-founder of DoubleClick, Kevin had communist president, later became CEO. Elliot was then an employee and an engineer at DoubleClick. So, the three of them had worked together. In fact, Elliot had previously helped co-found a company called Shop wiki, again with Kevin and Dwight. So, they had worked together in two contexts before they came together to start MongoDB. There wasn't any uncertainty about how they would behave with each other difficult situations, and they also complemented each other incredibly well. You now have your co-founders, you're ready to formalize things. One very important thing to do, right from the get-go is to vest founder's equity. What does this mean? It means, when you form the company entity, and you give shares to yourself and the other co-founders, those shares aren't immediately yours. You have to earn them over time. The reason this is important is because stuff goes wrong, and co-founder leave. This happens all the time for variety of reasons, and you don't want somebody walking away from the effort, and retaining all the equity in it. You're going to need that equity to hire people who will take over whatever that co-founder was doing. Many times founder vesting is seen as something that investors require, and it's seen as something that investors want to get rid of founders, and have that equity for investors, but that's not at all what's going on. I've been through many situations, almost every company that has multiple co-founders, eventually, somebody wants to leave or is basically asked to leave because they are not getting the job done. In those situations, it is essential form a fairness perspective that they don't wind up with as big a chunk of the company as those people who are going to continue building the company going forward. I've seen all crazy things including a founder who went on an India journey, and wound up staying in an Ashram, never coming back. Another question that comes up a lot is, how should equity be divided up among co-founders? While there is no rule that you absolutely have to follow, there are two considerations that matter a lot. One is, what is the relative experience and likely contribution of the co-founders? So, if you have one co-founder for instance who has had a successful startup before, and the other is a first time founder, that might make a very significant difference. The second important consideration is, how many co-founders they are? The more co-founders you have, the less likely an equal distribution is going to work. Why? Because if you have an equal distribution and a lot of co-founders, then many decisions are going to be very hard to make because everybody wants to have a word, and how to make the decision. Even given those considerations, I've seen both equal and not equal distributions work incredibly well, and I've also seen equal and not equal distributions not workout. So, there's no one size fits every startup when it comes to equity allocation. What is however key is that you are thoughtful about it, that you really think through, why are we choosing this allocation, and is it likely to work for our specific constellation of co-founders? 4. Validating Your Idea: Having an idea, potentially having co-founders is great, but how do you go about validating that idea? First of all, have you even thought about this as a step? This is such a crucial step and many people don't take it. They go from idea to building the product, skipping the validation step. That's a problem because it's very easy for us to convince ourselves that we have good ideas. I do it all the time myself. What does validation mean? Validation means the very simple step of eliciting a response from the people who want to use the product or service, and eliciting a valid response. Now, you will say, "Well, how do I do that when I don't have the product yet?" The answer is, you produce some form of mock-up. A mock-up could be as simple as a single webpage that has three bullets about what the product is going to do, or if you want to be a little more advanced, maybe it's a couple of different screens that hang together but don't actually do anything. They're not wired up to any actual software, they're just the screens of what it will look like or if it's more physical product, the mock-up could be carved out of wood instead of being made out of plastic. It might not have any actual electronics in it, it might just be the form factor. Another way you might create a mock-up is a short video that actually show somebody using the product even though it's entirely fictional, the product doesn't exist. Again, it's some thing somebody can see and react to, Why does a mock-up matter? Why not just bring a bunch of people in the room and say, "Hey, we have this great idea what do you think of it?" The reason is, people are great at reacting to things, even things that are just barely there. But people are terrible at giving you an opinion about something that's abstract, that's just a verbal description. So, sometimes people say, "Well, I'm going to run a survey, or I'm going to do a focus group," and then both of those are generally very bad ways of trying to validate a startup idea. Because they are asking people to give a reply about something in the abstract, and the kind of replies you will wind up with are like, "Yeah, sounds good," or, "Not sure that's interesting," or, "Oh yes, I have 17 ideas for what this should do." But they all exist in a vacuum, and it's quite shocking how when you give people a mock-up instead. However simple it may be an object that people can react to and it turns out people are very very good at reacting to objects. It's also incredibly important that if you do this validation, you don't bring a whole group of people together at once and let them all react in a group to the product. So, don't do a focus group, not even when you have a mock-up. Focus groups tend to be hijacked by local people, people who who like to hear themselves talk, who want to move the group in a certain direction, and it's also completely artificial. People, when they encounter your product are not going to be surrounded by another group of people debating your product it's, going to be them and the product. So, build the mock-up, keep it incredibly simple, show it to one person at a time and get their reactions. This is such a crucial step and yet it gets skipped almost all the time. I think in part it gets skipped because people are somewhat afraid to hear the reactions, and they are so enamored with the idea that they just want to build their idea and not get the reactions. Now, there's one exception to this rule. If you're scratching your own itch, then you are building the product for yourself. So, you don't really need to validate it, you are validating it, it's your own need. But if that's not the case, if you fall into any of these other categories where you're really trying to solve somebody else's problem, give them a chance to react to your proposed solution. Now, you might say, "How do I find these people to go react to my product mock-up?" There are a couple of different strategies that work. One is to literally go to a Starbucks and offer to pay people for their latte in return for them looking at your mock-up. Another great trick that lots of companies have used is you put ads on Craigslist. Turns out in most categories ads on Craigslist are free. So, you can put an ad on Craigslist, and let's say you want to create, do you think there's a need for a dog food subscription service? You might just put ads on Craigslist in the pet category and send people to a mock-up page, that doesn't really do anything other than let them at the end of it put their name down and register for an email list for when the product launches. That will give you the kind of validation that I think you should always take as a step before you go build the product. One thing that's important about validation is that you don't just go talk to your friends or your family. Your friends and your family are very likely to say, "Yes, this is great." Nobody wants to pour cold water on your startup enthusiasm. So, a big aspect of validation is to go to people you don't know and see how they react to your product mock-up how. Many people do you need to validate a product? We're not talking about hundreds of thousands of people, we're literally often talking about a dozen or more people. You're not looking for some in-depth product design, the product design is your job. You're simply looking for that one step of do people have an immediate reaction that's either very positive or very negative? Do they say, yes that meets something, I would use this if it existed? Or do they go scratched their head and go, "I don't really understand what you're trying to do here"? Or they tell you, "I have seven other things that I'm already using to solve this thing, this need that you seem to think exists"? So, you're not looking for detail, you're just looking that the idea isn't completely off the mark somehow. The biggest mistake in product validation is that people skip the step entirely, and so they wind up building things that fill an imaginary need as opposed to a real need. 5. Creating Your MVP: Now that you've validated the need, the next step is to come up with a minimum viable product or MVP. What is a minimum viable product? Well, it is the smallest instantiation of the product that actually meets the need. If the need you've identified is the need for a bicycle, then a unicycle is not a minimum viable product. A bicycle that doesn't have any lights, doesn't have a bell, maybe has only one gear, might be a minimum viable product. If your idea is a marketplace in which you can buy and sell, your minimum viable product needs to include buying and selling, but it may not need to include forums that may not need to include many different product categories, it may not include a lot of detail that you will add later. So think about what the core of your product is. What is the minimum viable product that people can actually use and that starts to meet the need that you have just validated. How should you go about building the minimum viable product? You should build it as quickly as you can and as cheaply as you can. As much as possible when you build your minimum viable product, you want to focus on where you are solving the need that you have validated, the unique value that you're adding and every other component you want to take as much as possible from an existing vendor that is well established or an existing open source project. Let's take the example of Etsy, a marketplace for handmade goods. When Rob Hyme and Chris were starting to build Etsy, they didn't go back and first write a database system, they just use an existing database. For your MVP, don't look for the perfect components, don't think everything needs to be 100 percent scalable, you don't need to be able to service millions of customers because you won't have millions of customers to begin with. Your entire goal with the MVP is to make sure that you're actually building something that really meets the need. So we had the validation as a first cut, the MVP is now we want to really solve the need and we want to show that we can solve the need. It is unlikely that your founding team will include all the skills you need to built the MVP. For instance, you might not have a designer as part of your team. Well, the great news is there are lots of wonderfully talented designers out in the world, and so you can go find a designer online, and that designer can create the initial design for your MVP. Later on when your company is further along, you can bring somebody on board full time. In creating your MVP, you want to keep things as simple as possible. Whenever you're tempted to add features that aren't absolutely necessary, just put them on a list for things to do later. It's always better to get your MVP out the door quickly, and essentially, a good guiding motive is you should be quite embarrassed by your MVP. You should think, "Oh my God, I don't really want to show this to people", but that's when you should start showing to people. Just keep in mind that the thing that's crucial about the MVP is that it can actually be used. There's some great examples of companies that today still look very much like the MVP that they launched with. Two that come to mind are Craigslist and Twitter. Craigslist today is still very much the way Craigslist was when it first launched. The MVP was literally the ability to list an item and for somebody else to go find that item. Twitter much the same. The MVP of Twitter was I can make a tweet and I can follow other people's tweets. Again, Twitter's added features since then, but the core experience of Twitter is still very much what the MVP experience was like. So when you think about the MVP, think about the core elements of the experience that you want to have, and don't think about, are they super pretty? Is it going to scale?"Those are all problems you can address down that line. But the core aspects of the experience, those need to be included in your MVP. One of the things that may be tempting, but you should really avoid is to build your MVP with a large corporation as your partner. So this could be a consulting firm or a design firm or even an engineering firm. That hardly ever works because the rhythm of large corporations just do not lend themselves to the rapid iteration that's required on the MVP. Etsy built the MVP of the marketplace in two and a half months. The Etsy team had to learn a bunch of new things as they were doing it, but they picked him up quickly themselves. The reason you need to do things relatively quickly is because if they take a long time you are spending a lot of money, and you're also not getting the crucial feedback that your MVP produces. When you're looking for external skills that are not on your team, as much as possible try to work with individual contributors. Use services such as Upwork or if you're looking for designers, go look on services like Forest or Dribbble or Behance. The reason is you will find people who know how to get things done quickly, who don't have a lot of overhead, and where you can get some sense of the quality of their work through people that they've worked with before or through projects that they've put out there. Same goes for if you need additional engineers. Always prioritize for people who can contribute quickly, and have a proven track record of producing quality output, and who don't have high overhead. Another failure mode in building MVP is to think too far ahead. You think,, What will my service be like when I have millions of listings? How will people find a listing? Let me build a very complex search for listings. Well, when your service launches, it likely won't have millions of listings, so that's not really an issue. Don't try to anticipate too many problems down the line that might arise in usage, focus on the things that people absolutely need today, and be willing to do things that are certainly not going to scale. There's a great story about Airbnb which today is a massive business, but Airbnb in the early days was hobbling along, and people weren't really using the service very much. One key insight that the founders had suddenly was that the reason was the people weren't trusting the listings and part of why they were not trusting the listings was because the photography that people submitted was really poor. So instead of saying, "How can we somehow in an automated fashion improve people's images using computer vision or artificial intelligence", they just said, "Hey, why don't we send some photographers to people and create higher-quality images?" That's definitely not going to scale, but it's going to see, it's going to test whether or not that really is the thing that's holding people back. Also once we've done that, other people can see the quality of images that should exist and that'll help raise the bar for everybody." Sure enough, that worked. So, whenever you encounter something in building your MVP where your reaction is, "Oh, I need a very complex solution that will work five years from now when I have millions of users or hundreds of customers", reign yourself into think, what do I need to launch my MVP? Sometimes, many times the answer is not more code, it's not a more complex piece of hardware, it's actually using humans to solve the problem. The much more common issue is that this isn't actually a problem, and so if you spend lots of time and money coming up with a technological solution, you may only find yourself disappointed and having invested all that time and money for something that people aren't actually encountering as a real problem on your service, or you haven't actually solved people's real need, and so as a result, your service isn't in fact or your product isn't in fact growing. That is the high-order bit problem. The high-order bit problem is, have you solved the need. Not will it work for hundreds, thousands or millions of customers. Once you have something that people love, then you can go back and fix all those problems. So, what if you put your minimum viable product out there and people aren't using? One of the biggest temptations is to add more features to your product. People are not using it because it doesn't do enough, so you think you need to add things. Coming back to the example of the bicycle, the temptation is, oh well, people do need a bell, and people do need lights, and people do need better handlebars and a better seat, and you can become obsessed with adding lots of features to your product, but in all likelihood, people are not using it because the core mechanic of the product, the core thing that it does isn't actually solving their problem. Adding more features tends to obscure that more than it attends to solve that. So when you are at the MVP stage and people are not actually using your MVP and they're not coming back and they're not giving you positive feedback on your MVP, don't go to adding features, go to understanding what exactly it is that's really holding them back from using your product or service. Even components that might eventually be core to your product, could initially be outsourced to a third party. For instance, when Etsy first launched, all payment on Etsy was handled via PayPal. It was only later that Etsy added their own payment system. It didn't matter whether that payment was deeply and elegantly integrated into the core experience, it got the job done. That's what you're looking for in MVP, get the job done. The MVP is really the ultimate validation step. It's when you know that you have actually solved somebody's need. So, the faster you can get to that, the better because once you've launched your MVP, you will see whether you're actually solving people's needs or whether despite the validation step, you've actually missed the mark. 6. Is It Working?: How do you know if you start up is working, how do you know if you found what investors call product market fit? Product market fit is a term that investors use a lot. It's what we look for when we invest in companies. Really what it means is, that the product or service you've built, it's actually solving the need that you had previously identified. It means you have, out of nothing, created something, something that people want. When you've accomplished that, you know you have a startup. Today, I want to emphasize that there are two parts that you always need to be looking at. You need to always look at quantitative and qualitative signals. What do I mean by that? Quantitative is, look at numbers, how many people are using my product. These numbers come from server logs, they come from instrumentation that you've put into your product, they come from sales numbers. Actually, look at numbers, don't be afraid of numerical analysis, but numbers never tell the entire story. Also, make sure you continue to observe users. In much the same fashion that I talked about when I talked about validation and when I talked about the MVP. Spend time with individual users, see how they use your product or service, and that is a very qualitative approach. Both of these have to be pointing in the right direction. Why is that? Well, it's very easy sometimes to think you have good looking numbers, but in the early days of a startup, you won't know yet, for instance, how long people will use your product for, how long you will retain them, or whether they will try it out because of novelty and then go away. Eventually over time, some of that will show up in the numbers and you can do what's called a cohort analysis, where you look at people who started using your product service on a certain date, how many of them continue using it? But when it's still early, like only a month into it, you can't know what your one year retention will be, you don't have data on your one year retention. But one month into it, you can spend time with users and you can see how they react to the product, whether they're enthusiastic, whether it's easy for them to use or whether they struggle, whether they're frustrated with it. So, it's very important that when you look at your metrics you look at both, am I bringing new customers, new users in, but also, are my existing users continuing to use the product? We think of the first thing as growth and the second as engagement. You want both growth and ongoing engagement. This is true, not just for consumer products, but also for B2B products. It's not good enough that somebody signs up for you, let's say, a new analytic service, if they never look at the analysis. Eventually, if people are not using your product, they will stop paying you for your product. It is crucial and I can't over emphasize this enough, to not just be stuck with your head in the numbers or to just think, "I'm going to talk to a bunch of users." You really can only understand whether things are working if you can do both. Almost all teams that I've experienced have a natural gravitation towards one or the other. Some teams are very numbers oriented, and so they really seem afraid of meeting with customers and getting that hands on experience of how is the customer reacting to the product. Some teams are very intuitive and they love spending time with customers, but they don't want to look at the numbers. Both of these tend to overlook often glaring things. So, if you want to know whether you found product market fit, make sure you study the numbers and if that's not you, find somebody else on your team or hire somebody or go show you numbers to another founder or to an investor and say, "Hey these are the numbers I'm getting, are those good numbers, what do you think of them, what should I compare them to? " Make sure you take the time to actually watch people one-on-one using your product and see how they react and what they get stuck on. Each time that I've invested in a company and that I've spent time with a company undoing both things, I and the company have learned new things that they didn't previously know. One the wonderful examples of that is, in the early days of Etsy, we did a bunch of one-on-one user testing. It turned out, nobody who came to the site for the first time first realized that these were all handmade craft goods. We all knew it, the founders knew it, that's after all what they've built the service for, we the investors knew it because that's where we had invested in, but we knew it because we had told ourselves the story of what we wanted the service to do. They did not have the context that was in our heads, in the heads of the founders, and the heads of the investors. So, we very quickly learned that we had to communicate that much more clearly and much more upfront. Another great way and maybe the most decisive way that you know you have some degree of product market fit, is when you get people to pay for your product. Many consumer services used to be free, but we're now seeing a wave of paid consumer services, consumer subscription products. If people are willing to pay a subscription for your product, clearly they're getting value from it. Often, companies are afraid in a B2C or B2B setting to ask for the order, to ask for the bill, to put a price on it. They want to launch the product for free and they want to say, let's see whether it's growing free. If you've really built something that's valuable, almost always you can charge for it, and even just attempting to charge for it, even if you decide that you want to optimize for growth, just the act of going to ask people for money will tell you whether they are truly seeing that this product that you've built is solving a need for them, or whether they're just exploring and playing around with things. People like novelty, people like to try out new services even in a business setting, but a very good measure of the value you're creating, is your ability to charge for the product. Why is it so important to really dig into the numbers and to really spend time with users? Well, it's important because, you don't want to hire a lot of people, you don't want to go raise money, you don't want to keep plugging away if you haven't actually solved the problem. I don't put the foot on the gas on your company until you know you've really got something, and if you haven't, go back to iterating, make changes. Keeping in mind, don't just add features, if you're going to add something take something away, until you have some that's working. This is important because the most valuable thing that you have in life is your time, and you don't want to be pouring your time into something that's not working. In the extreme, when things are not working, you can wind up with something like Theranos. It's a company that bound up racing hundreds of millions of dollars and dollars had a working product, and at all times they told themselves and their investors and their potential customers, that they had something that was working. A lot of time, a lot of people's important part of their lives were ultimately wasted on an effort that never really worked. There is a question as to how long you should stick with something if it's not working, and I can't answer that question for you. This depends on your conviction, this depends on your resources, that depends on your passion for it. As long as you don't try to grow things that aren't working, if you have the means of continuing to iterate on something, you probably should. 7. Hiring: Another crucial question that comes up in almost every startup is the hiring of first employees. When should you hire first anomalies? Well, a lot of it depends on how much you can cover among the founding team. The second you start adding real employees as opposed to contractors for well-specified projects, you're building up your burn rate. The biggest burn rate for a lot of start-ups are their employees. So, you want to make quite sure that you either have the funds available, or that you can procure the funds shortly before you hire employees. Otherwise, you create a huge amount of pressure. The biggest pressure that founders work under is the fear that they can't make payroll. So, before hiring employees, make sure you can pay them for some time, or that you know your found product market fit so that you can easily raise the money to go pay employees. Also people make lots of mistakes, especially first time founders, in hiring employees, and so if you hire employees too soon, you will make those mistakes very early, and they will actually slow you down substantially. The biggest issue is a lot of first-time founders don't really know what to expect in an employee and how to find a really high-caliber employee. That's especially true if you haven't worked at another place where there were excellent outstanding employees. So, you don't have a pattern of what to look for in that kind of employee. Now, there are many ways of overcoming that challenge. You can have people work up something for you before you hire them. For instance, you can have an engineer potentially work on a specific project before you hire them. Or you can go to people who have successfully hired high-quality engineers, and you can ask them to help you with that. Whenever you're hiring, a common mistake for first-time entrepreneurs is not to meet enough people for any one role, especially early on, when you don't have anybody to compare somebody new to. It's very important that you meet multiple candidates so that you create comparison for yourself. Overall, hiring is incredibly hard, and it will take up a lot of your time. So, be 100 percent sure that you are at the stage where you need an employee, and where you can afford an employee, so that then you can invest all the time that is required to hire well. You're better off without an employee than with an employee who is not getting the job done. Another common mistake in hiring the first employee is to hire somebody who is highly, highly specialized. There are very few instances where that makes sense for a start-up. Most of the time, you want somebody who is quite flexible and can work on many aspects. You want somebody who themselves is a self-starter, who doesn't need a perfectly specified roadmap for what to work on, who will take initiative. Those kind of attitudes are as important as ability when it comes to your first hire, because you will not have the time to perfectly specify the problem to break it down to an exact precise thing that somebody can work on. So, as you look for your first employee, you want to look as much for the attitude that this employee brings to your startup, the willingness to work with things that might change rapidly, that might be underspecified, that involve risk, and uncertainty, and a lack of clarity. If they don't have the attitude that goes with this, it doesn't matter how capable they are. Now, of course, attitude isn't everything, you need ability as well. But it's crucial in your early hires that you have people who have a start-up attitude that they bring to you. Ideally, when it's time to make your first hire, you already have a list of people you want to be on your team. Where does that list come from? I call it the "ABC of Hiring". Always be collecting candidates. Successful entrepreneurs, when they encounter people who are very good at what they do or highly-motivated, who show initiative, they make a note saying, "I want this person on my team." So, as you go through life, even before you have your idea, possibly even before you've incorporated your company, when you encounter such people, and when you know that you want to start a company, make sure you know who they are, they know who you are, you stay in touch with them, and you put them on the list of people you would like to be on your team. 8. Risk & Failure: Startups are really hard, and they fail at an incredible rate. Why is that? Well, it's because on everything you do in a startup, you can error in both directions. This is not a new idea. Aristotle first came up with this idea. When Aristotle defined virtues, he defined them as the golden mean between two extremes. Courage is the mean between cowardice on one end and recklessness on the other. Golden mean applies to startups. Let's take hiring. You can definitely hire too fast, too aggressively, build up your spend way faster than you have money in the bank, and put yourself in a corner, but you can also hire to slowly. Our portfolio company Coinbase for example, grew quite rapidly but didn't hire anybody in customer support, and then people who had problems with the service, were really frustrated because they couldn't get an answer. Take listening to your customers. On one extreme, you don't listen to customers at all. Customers are having problems, they're having issues, you just ignore it. You say, ''Oh, they're just not smart enough to use my product.'' So, you never fix them. Can you error by listening too much to customers? Absolutely. There'll be lots of customers who have edge needs, corner cases, and they will have feature requests. And if you keep adding all those features to your product, you will create a product that meets all the crazy edge cases, but it's actually impossible to use for the normal user. So, yes it is possible to not listen enough to customers, and it is possible to listen too much to customers. Once you're out raising money, again you can error on both sides. You can raise not enough money. So, you can't actually get to the next big milestone in your company, to the next value [inaudible] point. But it is also possible to raise too much money. Many companies that have raised too much money then wind up starting to many efforts all at once, or over complicating their product, or taking their burn rate up, and thus making the problem that they have to solve that much larger. So, fundraising too, you can error on both sides. Another place where you can error on both sides is the idea stage. In errpr on the side where you take the very first idea that pops to your mind, and just run for it, skipping the validation stage, skipping any stage of saying is this a real neat. But you can also error on the side of always thinking there will be a better idea. I'll come up with something even better, even bigger, and then you will never actually launch anything. As it turns out, a lot of people get stuck on that second thing. They want to hone the idea at the idea stage, never getting to that minimum viable product. So, if you take one thing away from today is do spend some time thinking about the idea, thinking about why this is a real need, how you came up with it. But then move quite quickly to validation, and from validation quite quickly to minimum viable product. That gets you way ahead of most people who talk about starting a startup but never actually start their startup 9. Closing: As with all advice, your mileage may vary. Take everything I've said and try to figure out how it applies to your specific situation. This is one thing you will find as a founder of a startup. Lots and lots of people will have opinions, lots and lots of people will provide advice. In the end, you're the one who has to do the doing. So, your job, a big part of your job, is to take all the advice you're getting, sift through it, but then make your own decisions. It's only decisions and actions that build products and companies, not advice. Now, I want you all to interact with each other to ask questions, to share anecdotes of companies that you've either started or worked for, what has worked for them, what didn't work for them. We can all learn from each other's experience and we can all make our own new mistakes. 10. What's Next?: