How to Get and Evaluate Startup Ideas | Startup School

[Music]

All right, hello everyone! I've got a lot of content to get through, so I'm gonna move fast. Buckle in! If you are looking for a startup idea right now, I'm going to try to help. But more importantly, I'm going to try to give you the conceptual tools to think about startup ideas in a sophisticated way, the way that we think about them at YC.

Here's the thing. No one, not even YC, knows for sure which ideas will succeed. And in any case, if your idea succeeds, has much to do with how well you execute as your initial idea anyway. But certain ideas are much more likely to succeed than others, and so my goal here is to help you stack the deck in your favor by starting with a promising idea.

The advice in this talk came from several places. First, I analyzed the top 100 YC companies by evaluation and I looked at how they all got their idea. So I started with some hard quantitative data on how recent billion-dollar companies actually came up with their idea. It also draws on a classic essay by Paul Graham that I really recommend; it’s called "How to Get Startup Ideas."

It also comes from helping YC companies that pivot in the middle of the batch and learning over the years what advice helps them to find a new good startup idea. And then finally, it comes from reading thousands of YC applications that we rejected and looking at the mistakes that caused good founders to come up with bad ideas. Those are the mistakes I want to help you all avoid.

This talk's got three parts. First, I'm going to tell you the most common mistakes founders make with startup ideas. Then, I'm going to talk about how to know if your idea is good, and then about how to come up with new ones.

Okay, the four most common mistakes with startup ideas. So the most common mistake is just building something that doesn't solve a real problem for your users. Typically, you can articulate the problem that you're solving; you can put it in words, but when you actually go and talk to the users, it's just not something that they really care about.

We call it a solution in search of a problem, or a CISP. Let's go through an example. So a lot of founders come up with an idea with this kind of thought process: they go, “Hmm, AI is cool. What could I apply AI to?” And then they go look for a problem that they could solve with AI. That's a solution in search of a problem.

The reason it's dangerous is that if you do that, you'll probably find a problem. But it will be a superficially plausible problem; it'll be a made-up problem that people don't really care about rather than a real problem that people actually care about. And if people don't really care about the problem, they won't really care about your solution.

So instead, you want to fall in love with a problem. The best way to find a startup idea is to start with a high-quality problem. Now, some founders hear this and they decide to interpret that as guidance to work on some huge societal problem, like, I don't know, global poverty or something. No doubt those are real problems, but they're too abstract to make good starting points for startup ideas. You need something that's more specific; something that's tractable with a startup.

The next mistake is getting stuck on what we call tar pit ideas. What's a tar pit idea? There’s this certain set of common startup ideas that have been around forever. They have been applying in rows to YC batch after batch for years, and when founders start working on these ideas, it's like they've gotten stuck in tar. They never seem to go anywhere. So we call them tar pit ideas.

Here’s what causes tar pit ideas. They all form around some widespread problem that lots of potential founders encounter, and it's a problem that seems like it could be easily solved with a startup. But it's an illusion. There's actually a structural reason why it's very hard or impossible to solve, which is why after all these years, no one has solved it.

You can see why ideas like this would be so dangerous and why they will cause so many founders to waste months of their life stuck on a tar pit. They're very tantalizing from a distance because they're so superficially plausible as startup ideas.

Here's a concrete example. This is a very common tar pit idea that's been applied to YC for like 20 years. This is like the stereotypical college student idea, and it goes like this: you think, “Man, every Friday or Saturday night when I'm making plans to meet up with my friends, it's so inefficient. I'm in all these different text threads and chat groups and we're like trying to make plans to meet up. I'm just going to make an app to make it more efficient.”

Well, it turns out that there are some structural reasons why this idea is hard, which is why in like 20 years of people applying to YC with this idea, nobody has actually pulled it off. You can see why so many people have been attracted to it. It's like it's a problem that almost everyone encounters at some point, and it seems like it would be so easy to solve. You can just imagine the app; it's just got like a list of events and you invite friends to it. Like it seems so simple.

The thing about tar pit ideas is that they are not necessarily impossible. I'm even open-minded that somebody will eventually make the app to meet up with your friends idea work. It's more accurate to think of them as common ideas that are much harder than they seem. So if you want to work on one, here's my advice: first, Google it. It's amazing how many founders skip the step of just like Googling for their own startup idea to see who has worked on it in the past. You should find who's worked on this in the past and actually talk to them if you can.

Try to figure out what the hard part of this idea is that has caused other people to not be able to solve it yet. The next mistake is simple. It is amazing how many founders will basically just like jump into the first idea they have without even stopping to consider whether it would actually make a good business. But more dangerous is the founders on the opposite side of the spectrum who sit around waiting for the perfect startup idea. And of course, there is no such thing.

So these people just never actually start a company. So if you imagine that there's like a spectrum between picking the first idea that comes to mind and waiting for the perfect idea, you know somewhere in the middle there is this like happy place, which is the place that you want to be, right? And the way that Paul Graham put this is that you should think of your idea as a good starting point.

A new startup idea is perfect, and no matter what you start with, it's probably going to morph anyway. So you just want to have an initial idea that has enough interesting qualities that can morph in the right direction.

So now suppose you have a startup idea and you want to know if it's good. I'm going to give you a framework for this, and the format of the framework is 10 key questions to ask about any startup idea. So the first one is: do you have founder-market fit? If I depict like one most important criteria, it'd probably be this one.

What I mean by founder-market fit is just: are you the right team to be working on this idea? A great example of what good founder-market fit looks like is PlanGrid. So PlanGrid makes an iPad app to view construction blueprints, and the founders of PlanGrid were Tracy and Ralph. Tracy had worked in the construction industry and she knew a lot about construction, and Ralph was an awesome developer who was like the perfect person to build this iPad app.

If you're going to imagine a team to start PlanGrid, the team that you would imagine would look, you know, something like that. That's what good founder-market fit looks like. It's like this team is obviously the right team to work on the idea. In fact, founder-market fit is so important that I would recast your search for a startup idea. When most people go to pick a startup idea, they try to look for a good startup idea like in the abstract. Instead, I would think about this exercise as an exercise to pick a good idea for your team.

You with me? It doesn't matter if something is a good startup idea for someone else if it's not a good idea for your team. So you may as well just look for ideas that you would actually be good at executing.

Okay, number two: how big is the market? Obviously, you need a big market, which for startups typically means like a billion-dollar market. But actually, less obviously, there are two kinds of markets for startups that are good: ones that are big now and ones that are small but rapidly growing. An example of the second one is Coinbase. So when Coinbase got started in 2012, the Bitcoin trading market was minuscule. But even at that time, it was pretty obvious that if Bitcoin succeeded the way people hoped that it would, that this would eventually be a billion-dollar market.

Number three: how acute is this problem? So as I said earlier, the most common mistake is just like working on something that just isn't really a problem or it's just not a problem that people care enough about. Here's an example of the opposite: here's an example of what a good problem looks like.

Brex, so Brex from Winter 2017 makes a credit card for startups. Before Brex, if a startup in NYC wanted a corporate credit card, they literally could not get one because no bank would give a credit card to a startup. That's a good problem! Like if your alternative to your solution is literally nothing, that's what a good problem looks like.

Okay, next: do you have competition? Now, most founders think that if you have competition, that that's bad. But counter-intuitively, it is the opposite. Most good startup ideas have competition, but if you are going up against especially entrenched competition, you typically need a new insight.

Next one is: do you want this personally? Do you know people personally who want this? It's amazing how often people start companies where the answer to both these questions is no. If that's the case, you've definitely got to worry that, you know, maybe nobody wants this.

So definitely time to go talk to some users. Only recently become possible or only recently become necessary. So something has recently changed in the world, like a new technology, regulatory change, or a new problem that is often what creates a new opportunity.

A great example of this is a company called Checker, which does background checks via an API. So it's an API for doing background checks on people. And roughly the story of Checker is delivery services like DoorDash, Instacart, and Uber started to take off, and they were all hiring huge pools of delivery people and workers, and they needed to run background checks on all of these people.

There were at the time already a bunch of large existing companies that ran background checks, but they weren't well-suited for this very new use case. That is like exactly the kind of change in the world that creates a new opportunity.

Let's talk about proxies. So a proxy is a large company that does something similar to your startup, but it is not a direct competitor. A good example of this in practice is a company called Rapid, which does food delivery in Latin America. When Rapid got started, there were already a few delivery companies in other parts of the world, like DoorDash, doing very well; they just hadn't caught on in Latin America yet.

So DoorDash was a great proxy to show that this idea of doing food delivery in Latin America would probably work. Is this an idea you'd want to work on for years? But this is a tricky one. Sure, if the answer to this question is yes, that's a good sign, but often it's not.

Often an idea grows on founders over time as it starts to work. As I'm going to talk about in a moment, a lot of the best startup ideas are in boring spaces, like tax accounting software or something like that. Like no one is particularly passionate about; nobody starts off being passionate about tax accounting software.

But tax accounting software is probably a good business, and if you're actually running a successful business, you tend to become passionate about it over time.

Okay, is this a scalable business? So if you're building pure software, the answer is yes because software scales infinitely, and you can just like check this one off. The place where founders most often get into trouble here is with service businesses, like agencies or dev shops—anything that requires like high-skill human labor in order to serve your customers.

Okay, and my last question is: is this a good idea space? Which of course means I need to tell you what an idea space is. This is a concept from my colleague Dalton, who you'll hear from later in this course. An idea space is like one level of abstraction out from a particular startup idea. It is a class of closely related startup ideas like software for hospitals, or infrastructure monitoring tools, or food delivery services.

And here's the thing: different idea spaces have wildly different hit rates. Over the last 10 years, if you started a company that did like fintech infrastructure or vertical SaaS for enterprise, the probability that your company became a billion-dollar company was astonishingly high. Whereas if you started something in consumer hardware or social networks or ad tech, the success rate was like orders of magnitude lower.

I can't say that that will continue to be the case for those specific areas because spaces flip from hot to cold over time, but it’s still worth thinking about picking a good idea space. A good idea space is really just one that you expect is going to have a reasonable hit rate for new startup ideas, and one that has founder-market fit.

That way, even if your initial idea isn't quite right, there are probably good adjacent ideas that you can sort of like drift into. I'm going to give a good example of how this played out in practice.

A good example of picking a good idea space is this company called Fivetran from YC1213. Basically, Fivetran is—they started making this tool for data analysis, and they went to some companies, and they tried to sell it. The companies didn't want it.

So they pivoted, and they built a different tool for data analysis, and they went back to the same companies and tried to sell them that one, and the companies didn't want that either. But each time they went to companies and tried to sell them some tool for data analysis, they would learn more about what those companies actually wanted.

So eventually, they sort of stumbled into an actual problem, into an actual tool for data analysis that companies actually wanted. This is why picking a good idea space to start with is so important. Because the Fivetran founders were shopping for ideas in a fertile idea space, they put themselves in a good position to bump into a good startup idea. If they had picked a bad idea space, they probably wouldn't have found anything.

Okay, before I talk about how to generate startup ideas, I've got one important topic to tell you about. These are three things that make your startup idea seem bad but actually make them good. The reason that they make them good is that most founders will shy away from ideas like these, which leaves them on the table for smarter founders to go and grab them.

And here they are: ideas that are hard to get started, ideas that are in a boring space, and ideas that have existing competitors.

Ideas that are hard to get started. So Paul Graham wrote a terrific article about this called "Schlep Blindness," which I'd really recommend reading. The example that Paul discusses is Stripe. You all know Stripe—they make it easy to integrate credit card payments to your website.

The fascinating thing about Stripe is that when Stripe launched, there were thousands of developers who already knew that this was a problem. They had tried to integrate credit card payments to their site and they realized that the existing options sucked. But strangely, not one of them even tried to start Stripe.

It's kind of a fascinating question why nobody else tried when so many people were in a perfect position to see this problem. The reason is that getting started building Stripe required some things that seemed really hard. You had to get a special deal with a bank; you had to learn a lot about the nitty-gritty details of credit card infrastructure. Those things seemed so hard that they scared off all the other people who might have started Stripe, which caused them to leave this like 100 billion-dollar opportunity on the table for the Stripe founders to go and pick up.

Okay, the second one is ideas that are in a boring space. A great example of this is Gusto, which makes payroll software. Payroll software is pretty boring, right? The thing is, there are thousands of people who must have realized that payroll software sucked, but because it was kind of a boring problem, nobody tried to fix it until the Gusto founders came along.

Because most founders shy away from ideas like this, boring ideas like payroll software have a much higher hit rate than fun ideas, like apps to find new restaurants to eat at or like apps to find the next song to listen to, something like that. Fun ideas get picked over; boring ideas get left on the table for a long time.

Now, you might be thinking, “Jared, why would I want to work on a boring idea? That sounds, you know, boring.” But here's the thing: even if you work on an idea that sounds fun at the outset, the day-to-day reality of your startup is going to be mostly the same anyway.

Either way, you'll be mostly writing code, fixing bugs, talking to users—like pretty much the same stuff. I would argue that once the initial excitement of your idea has worn off and you are 6 or 12 months in and you are grinding out the execution that makes your idea actually work, how fun the initial idea sounded actually has little to no correlation with how much fun you will actually be having working on your company.

And the last one is that founders incorrectly shy away from spaces where there are existing competitors. Counter-intuitively, most startup ideas—most good startup ideas—have existing competitors. When founders go into spaces with no existing competitors, they often find out that the reason that there are no competitors is because no one wants the product.

A great situation is actually a market where there are existing competitors, but you've noticed something that they all seem to have missed, or they all just kind of suck. A classic example of this is Dropbox. So when Dropbox launched, they were already about 20 cloud-based file storage companies. Dropbox was like the 20th company in that space to launch. Naively, you might have thought that that made this a bad market.

I mean, you would have gone and looked and said, “Oh, there are already 20 competitors; it seems like a bad market to go into.” But if you were savvy about startup ideas, you would have realized that it actually made it a great market. And the reason is that even though there were like 20 companies doing this, most people didn't use any of them.

That strongly suggests that there actually was a problem here, but the existing products hadn't solved it. Drew, the founder of Dropbox, had a very specific insight about what all of them were missing. His insight was basically their UI sucked.

The reason their UI sucked was that at the time, the way that they all worked was you had to like go to their website and manually upload your files one at a time into their website, which sucked, of course. Drew had really a technical insight, which was that if he integrated directly into the host operating system, he could just sync your files automatically without you having to do anything.

That was a real step-function change in how convenient these services were to use, and that was the right insight.

Okay, let's talk about how to come up with startup ideas. So it is possible to sit down and explicitly think of startup ideas, and in a moment I'm going to talk about how to do this. But it is actually not the best way. The best way to have startup ideas is to just notice them organically.

If you look at the YC top 100 companies, at least 70 percent of them have their startup ideas organically rather than by like sitting down and explicitly trying to think of a startup idea. The problem is that when people sit down and try to think of startup ideas, they tend to think of bad ones. They're especially likely to think of the same set of tar pit ideas that I talked about earlier.

Whereas startup ideas that occur to you organically are more likely to be good ones. So if you're not planning on starting a company imminently and you just want to put yourself in a position to have organic startup ideas in the future, here are three ways to do that. This is like playing the long game to set yourself up for future success.

First, just like become an expert on something valuable. If you're working at the forefront of some field, you'll see good startup ideas in that field. A great way to do that is to go work at a startup. Harge talked about this in his talk last week. If you're working at a startup, you will become an expert in the thing that that startup does, and that is really putting yourself in a position to have great startup ideas.

Finally, if you're a programmer, one thing that can work is to just build things that you find interesting, even if they're not businesses, or they're not clearly startup ideas. Sometimes they like turn into them over time. A really striking example of this is the story of Replika.

This is exactly how Replika started; it was just something that Amjad found interesting. It wasn't supposed to be a startup originally.

Okay, but if you want to generate ideas for startups right now, I'm going to walk through seven recipes for doing that. I've tried to list these in order of how likely they are to lead to actually good ideas. So start with the first ones.

Okay, here's the first one and the best one: start with what your team is especially good at and think of ideas that take advantage of your expertise. The reason that this is so effective is that any idea you come up with this way has automatic founder-market fit. Do you see how this is almost like a hack to generate the set of ideas that has founder-market fit?

Here's a great example of how this worked in practice. A good example is this company Resi, which is like OpenDoor for rental apartments. Before starting Resi, the founders had worked in real estate and debt financing, and they were experts in those areas.

When they got into YC, they spent the first month looking for ideas, but the smart thing that they did is they only looked at ideas in that idea space—in like the rough idea space at like the intersection of like real estate and fintech. That was a smart move because that is a very fertile idea space for startups. Like, many billion-dollar companies have come out of that idea space.

The Resi founders were experts in that idea space, and so because of that, their search for a startup idea was pretty quick and painless. Pretty quickly, they came up with the idea for Resi, which is like an excellent idea and has perfect founder-market fit.

So if you have specific expertise like the Resi founders did, you should definitely start by looking at ideas in the things that you're experts in. It is weird how many startups apply to YC, and we look at their applications, and the founders like are actually legitimate experts in something, but the idea that they're applying with is like something completely different.

Now, if you're a young founder, if you're in college or something, you may not have had a chance to develop the level of domain expertise that the Resi founders had. So this may not be the right recipe for you. So let's talk about some other ones.

The next recipe is to start with a problem you've personally encountered—ideally, one that you're in an unusual position to see. So Vetco is a website for veterinarians to order supplies. You can think about it like Amazon.com for vets.

The short story of Vetco is that the founders are brothers and their dad is a veterinarian. Growing up, they would notice that the way he ordered supplies was like super old-fashioned. Like, he'd have to call up a supplier on the phone and order stuff through like a 1-800 number or something. It was like very obvious that you could build an Amazon.com kind of thing that would replace that.

I love the story of Vetco because it’s such a great example of what a great startup opportunity looks like. Thousands of veterinarians must have known that this was a real problem, that it was really annoying there wasn’t like a basic website where you could just go to order supplies.

But the thing is, veterinarians don't start tech startups very often. On the other hand, you had thousands of programmers in Silicon Valley who were banging their heads against CISPs and tar pit ideas, totally unaware that over here there was like a really great genuine problem to work on. Because of this, when the Vetco founders got started, they had no competition. This amazing idea just got left on the table for years.

So if you want to try to actually use recipes one and two, here's a specific set of instructions for how to do it. This is advice I often give to founders NYC who are pivoting and who are looking for a new idea. Here's how it works: for each founder on your team, go through every job you've ever had, plus all your internships, plus like other life experiences, and think really carefully about each of them. What problems did you come across?

What did you learn that other people don't know? What are problems or opportunities that you've been in kind of a special position to see? Those are the best places to start looking for startup ideas.

Okay, the next one is to just think of things you personally wish existed. This is like a really classic recipe; this is very common advice. A great example of this is DoorDash. The short story of DoorDash is that the founders of DoorDash were undergrads at Stanford, and the thing they really wanted was to be able to just order food from local restaurants and have it delivered to their dorm. Before DoorDash, you couldn't do that, so they started DoorDash.

This is a great recipe, but this is the recipe that is most dangerous and potentially leads to tar pit ideas. So if you're using this recipe, you just gotta stop and think for a second: is there a reason why this thing doesn't exist yet?

Okay, the next one is to look for things in the world that have changed recently that might have created a new opportunity. A great example of this is COVID. So when the pandemic started, it changed daily life for all of us, and many founders realized that this created the opportunity for new companies.

Some very successful startups came out of it, and one of them is this company called Gather Town, which builds like this fun way to hang out with other people online. The founders of Gather Town were actually working on a different idea that wasn't going so well. When the pandemic started, they pivoted to this because it was obvious that like the change in behavior of the pandemic had created a bunch of new opportunities.

You can also look for companies that have been successful recently and look for new variants on them. A good example of this is a company called Nuvo Cargo. Nuvo Cargo is a good example of explicitly sitting down to try to think about startup ideas and actually finding a good one, which is the thing that I told you is hard to do. Nuvo Cargo is basically Flexport for Latin America; they help U.S. companies to import stuff from Mexico.

The story of Nuvo Cargo is that Deepak, the founder, was working on a different idea in NYC and he realized that his idea wasn't going to work. He went on like a systematic search for better ideas, and he picked Nuvo Cargo for very analytical reasons. He picked it because it was a large market, because there were good proxies from other companies like Flexport, and he picked it even though he didn't have deep domain expertise in the import-export space because he had some connections that would enable him to get started.

He felt like he would just be very good at running this kind of operationally intensive business, and that worked really well. Nuvo Cargo is doing super well.

You can also go and talk to people and just ask them what problems they have. This can work. The downside with this recipe is it actually requires a lot of skill. If you want to do this, I would recommend first picking a fertile idea space and then going and talking to people within that idea space.

I would also recommend talking not just to potential customers but also to founders of companies in that idea space to get advice about what ideas are actually worth pursuing. A good example of doing this successfully is A to B. A to B was in the position that a lot of founders that struggle to find good startup ideas are in, which is that the founders were pretty young and they hadn't acquired a lot of specific domain expertise like the Resi founders had yet. But they really wanted to do a startup, and they wanted to do a startup that had like a genuinely good idea.

The way that they came up with the idea for A to B was very systematic, and so I'm going to walk you through how they did it and really break it down for you. So A to B makes fuel cards, and if you haven't heard of a fuel card, it's like a special kind of credit card for truck drivers.

When the A to B founders got into YC, they pivoted, and they spent the whole YC batch looking for a new idea. Here's how they did it. First, they picked an idea space, and the idea space that they picked was essentially software for the trucking industry.

They picked this idea space despite not being experts at the trucking industry because they felt that it just should be a fertile idea space to go hunting for startup ideas. The trucking industry is a big industry that hasn’t been that disrupted by startups and software yet, so they just felt like there were probably some good problems to work on in the trucking industry.

But the problem is they didn't know that much about the trucking industry, so they didn't know what those problems were. They decided that they would turn themselves into experts in the trucking industry, and what they did is they actually physically drove to truck stops, which were places where truck drivers are just kind of milling about, and they would just walk up to truck drivers and like start talking to them and ask them questions about what their problems were.

They also talked to a lot of founders who had started companies in the trucking space to get ideas for what problems were actually worth working on. They would basically talk to like anyone who knew anything about trucking that was willing to talk to them, and as they did that, they began to put together like a mental map of the space and where the good ideas were, where the bad ideas were.

They actually went through a whole bunch of different potential ideas before eventually deciding to work on fuel cards. I love the example of A to B because A to B is one of the best new ideas to come out of YC in several years. A to B is a phenomenal company, and the approach that they used to find this idea is something that really anyone could do.

Most founders don't do this because it just sounds like too much work. So if you're willing to put in the work, this is an amazing way to find a startup idea.

Okay, and my last recipe is to just like look for big industries that seem broken. Any big industry that seems broken is probably ripe for disruption. And finally, I've got kind of a bonus recipe, which is to just find a co-founder that already has an idea.

There are actually a lot of people just on Startup School co-founder matching right now that already have an idea and are looking for a co-founder. So if you don't have a co-founder and you don't have an idea, that could be a great hack to getting both at the same time.

The last point that I want to leave you with is just to remember that it's often hard to tell if a startup idea is good or not. While I hope the concepts that I talked about here will help, typically the only way to know for sure if your startup idea is good is to just launch it and find out.

So if after all this, you've got a startup idea and you're still kind of on the fence about whether it's actually a good idea or not, that is my advice for you: just launch it and find out.

[Music]