One Order of Operations for Starting a Startup
blog.ycombinator.comHello - I've been invited to some universities lately to talk to students about Y Combinator and the question "how do I come up with a good startup idea?" comes up every time. In this essay, I've attempted to write a guide for people with the same question. One thing I'll point out from the onset is that there are many paths to starting a startup and this one is not the only one (and probably not even be the best). While providing a possible path - my primary goal was to destroy some commonly held myths about "the right way" to start a startup. Even more specifically the idea that you need a "great idea" at the beginning. Happy to hear your thoughts and questions on the subject.
Hi Michael. I appreciate your candidness and openness to accept counterpoints. How would you evaluate a startup that’s only a couple weeks old, however close to shipping a prototype but probably won’t have time to get users by Sunday ( YC deadline ). It’s an idea that could generate a billion jobs and hence we feel it’s better to get YC support than try to grow it slower with a revenue model. Should we apply anyways or wait to get traction and apply late?
In my experience, the answer is almost always “just apply” - there’s no harm in it and answering the questions can be helpful to consider various aspects of your idea in more detail.
True. It’s just going to physically hurt if the hurried demo doesn’t do the idea justice. Won’t be fair to the scope and scale of the idea. I guess just apply is the answer regardless. Thanks!
If you get rejected, you can apply again for the next batch when you'll have a working demo and maybe some users.
Just apply man :)
Thanks Michael! For teams that start out this way, what is your advice for iterating on this initial seed of an idea and finding product-market fit (and making sure that you find an idea you can be passionate about for the long haul)?
Get the MVP in the hands of users early and try to learn from them how to make something that they want. Be open to the idea that while you might have figured out a good problem to work on - your solution is just a hypothesis to be tested and refined over time.
What do you do if a decent amount of people use your MVP, but then churn out and don't respond to requests for feedback? Essentially, they don't really care about the problem or solution, they were mostly just curious. Scrap it and start again? From my experience and what I hear from friends, this seems very common.
Ideally you'll work with early users who are just one or two degrees removed from yourself, so they'll feel social pressure not to ghost you.
Unrelated: In a previous talk, Brian Chesky mentioned writing down the core values of Airbnb before they had hired a single employee - do you think (other) startups should follow this example in the early days and consider the future culture of the company?
Here's a more cynical possibility: What if casting about for brilliant startup ideas is actually a signal that you are unlikely to be a founder, by DNA.
In my experience, founders simply can't help themselves from starting things, regardless of how ridiculous (or great) their ideas are. There are obvious exceptions to this (for example, I wouldn't include people who inherit businesses, who have a lot of startup capital, in this founder pool).
this is so dumb. I'm a founder. I was chosen to be the technical co-founder of a company where the space had been selected and problem had been identified. before that I wanted to start something but because I'm not asinine nor had any useful domain experience I didn't have any notions of valuable problems to solve. i have all of the makings of a founder (work ethic, risk tolerant, creativity, people skills, etc.) and we're alive because I do. are you really going to argue because I didn't just start another Shopify clone for dogs and then pivot 10 times that I don't have founder DNA?
edit: i'm probably getting downvoted because i come off as a braggart. maybe but that was not my intention - i just meant that there are plenty of counter-examples to this claim that if you don't go out and start from scratch impulsively then you're not genetically predisposed to being a founder.
You make a great point that there's a lot more to being a founder than simply pulling the trigger on a business license. Props for making it work in your bidness.
This is possible.
This post doesn't mention when to sign an operating agreement with your co-founders and legally form a company.
IMO, the answer is: you should have a signed equity agreement before you start working on the MVP, and and you must have one by the time you finish working on the MVP.
In practice, you'll probably need to incorporate a business to build your MVP. If you're accepting money from customers, the company should accept their money, not a co-founder directly. (Y Combinator recommends a Delaware C Corp.)
Even if you're just spending money, maybe at first one of the co-founders can fund it out of a personal bank account, but it's way better to have a business with its own bank account and operate the business out of that.
Your MVP isn't really viable unless your co-founders have signed an agreement. Sign one, early.
We have a number of companies who apply to YC with an MVP before incorporation. While I agree getting this done is a good idea - its not an absolute requirement for YC.
Incorporation isn't required, but delaying equity agreement is a recipe for a painful co-founder breakup later when it emerges that I thought we were going in 50/50 and you thought it was 70/30.
You should at least have talked about all the points that go into an incorporation document and some that don't (equity shares, vesting, time & work commitments, decision-making, roles, etc.) and have a verbal agreement that everyone's on-board with. You don't have to actually file the paperwork, which is expensive and creates ongoing administrative work. You should file the paperwork ASAP, but if a founder starts working on the project and then decides that his true desires lie elsewhere or he's not happy with the working arrangement, that's often easier to fix before there's a real company than afterwards.
totally agree
My problem: lots of ideas and enough technical skills to make them happen, no idea how to build any initial buzz even with an mvp. Even when I try researching upfront, asking friends and family about problems and potential solutions, most of the responses are like, “well that could be cool I guess.” I’m not sure if this is a sign that I haven’t been coming up with the right ideas, or iterating enough, or that I just suck a selling (or perhaps all three).
Any suggestions?
I just don’t see the point in asking someone’s opinion.
To me the only thing that counts is how people respond to actual software, and even then it’s got to do what you envision the core functionality being.
I know everyone else disagrees but I think you basically got to build it to find out what people really think.
And even then the only real measure is whether they jump on it without much selling explaining or persuading.
But then again I’ve never succeeded so there you go I’m provably wrong.
Let's find out - pitch me (either in here or to my email in my profile) and I'll tell you what my initial reaction and thoughts were. Pitching over text interface isn't as effective as in person but I can hopefully at least give you some idea.
> "There are a lot of counterexamples to the “failure” scenarios I described above, both for first time and repeat founders."
As a reminder the two failure scenarios are (1) idea -> pitch -> raise -> start company and (2) non-technical team.
What are these counter-examples? I can't think of any, although I have the vague feeling they must exist. In particular, counter examples to (1) must exist since the MVP/lean/validate idea wasn't as prevalent even 20 years ago, and counter examples to (2) must exist especially outside of technical startups where there are other moats.
Can you think of some?
This is good advice. It is sometimes preferable to start with the right people though, and together explore various problem spaces.
In reality this isn't much different from the reverse order, since usually you find out what kinds of problem spaces are interesting to you from talking to friends.
Great people that you work well with come around way less often than reasonable ideas in my experience. So I think it makes sense to prioritize finding the people rather than finding the problem space.
The biggest advice I'd add is to talk to users before buidling your MVP. You can potentially save yourself a lot of work.
Also, there's a ton of details in how to identify risks and validate assumptions. Four Steps to the Epiphany is a good tactical book on how to do this.
Slightly related, the Startup Podcast has two great episodes on Justin.tv: https://www.gimletmedia.com/startup/almost-famous-season-3-e...
That was so much fun :)
I really enjoyed your episodes, thank you for taking the time to do them
Thanks for this!! I’ve started pitching my idea to incubators and accelerators (cold) but now I don’t think I’ve done a well enough job of showing progression beyond the idea phase. Time to start making my software more publicly accessible. Scary. Can’t wait though. :)
I have a lovely blockchain idea but don't really know an appropriate person to deep dive with on it. sad face
Hey Michael, nice article. I saw in a comment you said students frequently ask you "how do I come up with a good startup idea?" and I know in the article you also included "This by no means the only path to an MVP...but it is a path that I’ve seen work for a number of YC companies."
I wanted to share something tangential that has worked for me as I also speak to students regularly and get the exact same question every time. My response includes the standard "Is there a problem you are passionate about?". But I've also observed how many students don't see problems as "problems" when asked to brainstorm about it and many would-be entrepreneurs end up left out of the process and discussion. I'm not sure all the reasons why but maybe a pain point isn't significant enough for them to label it a "problem" in their head when asked ("oh that's just an inconvenience not a real problem"). Or maybe their personality leans toward accepting status quo without realizing it can be changed and they can be the ones to change it. Or maybe they are too shy or introverted at first to describe and complain of a problem out loud.
So I've modified my response to students a bit to include: I have seen startup ideas come from 1. An idea for a solution to a problem and 2. An idea for something "that would be awesome" if someone created it but no one has yet or at least not done a good job of it. Here's what I mean by "that would be awesome" ideas, to take the example of Justin.tv, it leads to the same startup being created but 2 different types of entrepreneurs might get there by thinking:
1. What's a problem you experience and want to solve: "TV used to be entertaining but TV shows now feel stale, boring and with writing that feels so formulaic."
2. What is something "that would be awesome" if someone created it: "I love TV and movies, The Truman Show is one of my favorite movies, it would be awesome if someone made a TV show for based on following someone's life similar to The Truman Show but for real and with real people not actors".
Same startup. Two ways to get there. Hopefully this all makes sense. I've had good results engaging with students by adding the "that would be awesome if that existed" to the ways to think of startup ideas. This especially true with students who are shy or less likely to complain about problems out loud for some reason as well as for students who know how lucky they are to be at a fancy university in a first world country and they feel guilty describing anything as a problem worth solving if it isn't on the level of world hunger or similar (in which case I encourage those students to go solve world hunger if that's what they want to do).
Also one more thing. In my opinion the real major problem for aspiring entrepreneurs, bigger than coming up with a problem to solve, is how to brainstorm!
Most people assume brainstorming is just "sit and think". And it can be that and some people are very good at that (and other people aren't good at sit and think brainstorming at all but they just get lucky and a good idea pops in their head one day). But there's a lot of literature out there with research on effective brainstorming as well as tips and tricks to get one's brain into a better brainstorm mode (hint there's a reason why so many people say their best ideas come while showering or jogging). I am outside the valley so your mileage may vary over there but I see guided brainstorming sessions work much better than the usual casual 'shoot the shit type brainstorming' for the majority of people as brainstorming doesn't come naturally to them. Hence everyone has 99 problems and no good ideas.
The most popular thing I do when talking to students is a class discussion to come up with a problem and get the entire group of students brainstorming and iterating solution ideas together, with the goal being "find the first 5 feasible but bad solutions" (Asking them to come up with a "good solution" is too much pressure and then no one wants to raise their hand to say an idea others in the group might think is bad, so it helps to make clear it's OK we don't need good ideas at this point). Another thing that helps is I hand out index cards to start things and get students to anonymously write down one "problem to solve" idea or one "That Would Be Awesome" idea. I collect the cards and pick the idea that best facilitates brainstorming and discussion to start things off and then do guided brainstorming together. Most students have never been part of a real brainstorming session beyond some brainstorming how to do a group school assignment that gets an A and isn't too much work for everyone and that really they don't care about. Real brainstorming is hard and they have never done it for real in a group setting.
Sorry for the long comment. Just wanted to share something that has worked for me getting more students more involved in entrepreneurship, especially for the types of students who currently are under-represented among the population of founders these days for whatever reasons. Yes. Also shout out to my fellow shy introvert founders and anyone who doesn't walk around all day thinking about problems or like complaining about them. It's OK to start a company with the goal to make something awesome a reality, everyone has 99 problems and zero good ideas, try thinking up something awesome instead.
Using Airbnb as a reason why your idea doesn't matter is like using Bill Gates as an argument for dropping out of college. The idea is super important; in Twitch's case, the process of getting to that idea was putzing around with something terrible for a while. Ideas do matter and you're not doing anyone any favors with this quixotic reductionism.
Clearly the idea (or the solution) is important - my argument is that knowing the right solution day one isn't very important. Airbnb and Twitch are great examples of this.
You're extrapolating from the winners and applying that to the masses. Read up on their history; Airbnb and Twitch almost died multiple times and that's not a good model for any startup. Even startups that actually start with a good idea fail frequently.
Yes, Michael Seibel, please read up on the history of Twitch.
Lol
I was looking for more "operations" closer to the day of nicely positive earnings.
Idea? Had that. Started with a problem that essentially every Internet user in the world has and that so far is solved piecemeal, with poor means, and, net, at best poorly. For a solution, the first good solution, and what should be a really good solution, invented one based on some original applied math based on some advanced pure/applied math prerequisites. The applied math makes good progress on the meaning of Internet content.
Team? Just me.
Technical Founder? Just me. I've been in applied math and computing for a long time, have written a lot of code. Did applied math and computing at IBM's Watson lab in an AI project. Once at FedEx quickly wrote some code that scheduled the fleet, alleviated some severe concerns of the BoD, enabled some crucial funding, and saved the company. My Ph.D. dissertation research was in stochastic optimal control. Since in that field, commonly the computational demands are absurdly high, some good computational experience was important. So, I made some progress in computational algorithms and wrote and ran some code that had the computational demands quite reasonable. Have written a lot of other successful code otherwise, especially for US national security.
Code? So, with my idea, I designed and wrote the code. To users, my work will be just a Web site. So, I needed to write the code for the Web pages. And the Web site needs to make some use of relational database and also, of course, my applied math. So, for the coding, I selected Windows over Linux and there selected the .NET version of Visual Basic (apparently essentially just a different flavor of syntactic sugar compared with C#). I designed the software and server farm architecture: There are several programs that communicate over a server side local area network (LAN). The communications are based on TCP/IP and de/serialization of object instances. One of the server types is for the Web pages. The software architecture is so that there is no session affinity between a particular user and a particular instance of a Web server, and for this there is a session state server; I could have used Redis but, instead, wrote my own based on just two instances of a .NET collection class. For the code for the Web pages, used ASP.NET. For the code for the database access, used ADO.NET. My code has about 24,000 programming language statements in about 100,000 lines of typing. So, about 76,000 lines have comments. So, the code is well documented, with the comments and with various external documents. The applied math is documented using D. Knuth's mathematical word processing software TeX. The code appears to run as intended: So far I know of no bugs. The code doesn't need refactoring. Execution timings show that the code is nicely fast. The code is for a serious service and company and not merely a minimum viable product (MVP). I wrote all the code.
The next operations I have in mind are (1) tweak the code to make a few small changes, (2) for the code for the crucial applied math, do some severe testing a third time, (3) gather a lot of data for the database, (4) do an alpha test and tweak the code as advisable, (5) do a beta test, (6) arrange to run ads, (7) get publicity, users, revenue, and earnings.
There are some user interface (UI) and user experience (UX) features: (1) There is no use of HTTP cookies. (2) The Web pages are simple; the pages should load quickly since the largest page sends for just 400,000 bits. (3) I wrote no JavaScript; Microsoft's ASP.NET wrote a little JavaScript for me, likely for some aspect of cursor positioning, but for the users enabling JavaScript is optional. (4) The site is interactive, but the UI is simple. (5) Many users should find the UX fun, somewhat game like. A good user might spend a hour a week using the site. That much time per week for a significant fraction of all developed country Internet users would be a lot of usage, Web pages and ads sent, and ad revenue.
The latest is that my development computer developed some data corruption and finally quit. I ordered parts for a new development computer and first server. So, I got an Asus motherboard, an AMD FX-8350 processor (8 cores at 4.0 GHz), 16 GB of ECC (error correcting coding) main memory, etc. As of yesterday, the system boots and runs the power on self test (POST) and BIOS (basic input/output) code successfully; the BIOS finds the details on each of the motherboard, the processor, the one disk drive installed so far, the CD/DVD device, etc.
The OP seemed to concentrate on the earlier operations I've already done, that is, the idea and the ability to design and write code, but the operations from the running code to good earnings also stand to be important.
For venture funding, I found that all the well known venture firms in Silicon Valley, Boston, NYC, etc. were not at all interested -- e.g., nothing about the description here is of any interest at all. So, I funded this work from my own checkbook and am 100% owner. Maybe once I have, say, $250,000 a month in revenue, some venture firm would be interested, say, write me a Series A check for $5 million. But with $250,000 a month in revenue, cash won't be a bottleneck; I won't need their check and won't want to take on the overhead and risk of a BoD, and won't accept their check. I got no meaningful response at all from YC. That I am a sole, solo founder seems to be a big negative for venture firms and YC.
Should I be able to be a sole, solo founder of a successful company with no equity funding? Sure: In the US, border, villages, town, and cities, people do that by the millions in pizza carryout, auto body repair, .... Should I be able to be successful as a sole, solo founder of a Web site business? Sure: The founder of the romantic matchmaking site Plenty of Fish did that.
you're getting downvoted because 1 you're rambling 2 this isn't the time or place for this.
I'm not "rambling" at all. Instead very much I'm responding quite directly to the OP. I'm responding point by point on just what the OP concentrated on, the business idea, what constitutes a promising business idea, the founders, technical founders who could write code, the code, minimum viable product, the "team" of founders and the point about a sole founder, and venture funding, in particular, YC funding. Those issues were all central in the OP and also in what I wrote. So, I'm right on topic. And I went point by point and didn't "ramble".
And to be more clear, I illustrated with a project I know well, my own.
Then from my example, readers who are a natural audience for the OP can see some likely important points: (1) Getting a promising business idea can be less obscure than suggested in the OP. (2) Getting a "team" that can do the work is easier than suggested in the OP, e.g., some sole, solo founders can do all the work from the business idea through the code ready to go live. (3) Venture funding and/or YC funding can be much more difficult to obtain than suggested in the OP.
Then I made the point that the OP didn't say much about the "operations" from the running code ready for production to good earnings -- there's more to say there, too.
And for the potential of a sole, solo founder, I gave some arguments and examples why that is not unreasonable.
You noticed how what I wrote was really a direct response to the OP, right? Apparently not. What you wrote about my post makes no sense. So, your concerns must be elsewhere, and a strong suspicion has to be that you do have some bitter reactions to what I wrote but not like anything you wrote.
So, you don't know me at all but want to mount a personal attack on me. Okay, I understand.