How to Get and Evaluate Startup Ideas | Startup School

Y Combinator32:212022-11-17English

This Startup School talk explains how to find and evaluate startup ideas: start with a problem your team is well equipped to solve and users genuinely care about, then test it by launching rather than waiting for the perfect idea.

Overview

How to Get and Evaluate Startup Ideas

Key points

  • Don’t pick a technology first and then force a use for it. First make sure users really care about the problem and that it’s specific enough for a startup team to have a shot at solving it.
  • An idea that looks simple but that people have tried and failed to make work time and again may have structural obstacles. Look up past attempts and talk to the people behind them if you can. Don’t rush into the first idea, but don’t wait forever for the perfect one either.
  • When evaluating ideas, Jared cares most about founder–market fit. The question isn’t whether the idea is good in the abstract, but whether your team is the right one to make it work.
  • Choosing a field that suits your team and has several related opportunities gives you more room to maneuver than betting on a single solution. By pitching different data analytics tools to the same customers, Fivetran gradually discovered what they were actually willing to buy.
  • A difficult start, a boring topic, and existing competitors aren’t necessarily bad signs. Competition can show that there’s demand. The key is whether you’ve spotted a problem none of the existing products solves.
  • According to Jared’s analysis of YC’s 100 highest-valued companies, at least 70% of the ideas were found naturally. To improve your odds of finding one that way, build deep expertise in a valuable field, work at a startup, or build projects you’re interested in.
  • Even without industry experience, you can pick a field and interview people who work in it, including founders. But neither evaluation nor interviews can replace launching. If you’re unsure, launch the product and see what happens.

Chapter notes

Four common startup idea traps

  • No one can know for sure which idea will succeed. Execution matters as much as the initial idea, but starting with a more promising idea can improve your odds. The advice comes from analyzing how YC’s 100 highest-valued companies found their ideas, Paul Graham’s essay “How to Get Startup Ideas,” and experience helping companies in YC batches pivot and reviewing thousands of rejected applications.
  • The first trap is a “solution in search of a problem” (SISP): if you start by thinking AI is cool and then look for somewhere to use it, you’ll often end up with a problem that sounds plausible but that users don’t care about. Do the reverse: start with a problem users genuinely care about, one that’s specific enough for a startup to tackle, rather than something too broad, like global poverty.
  • The second trap is falling into a “tar pit.” Some common problems look easy to solve but have structural obstacles that have been hard to overcome for a long time. For example, friends trying to meet up on the weekend have to coordinate across several group chats. An app that lists events and sends invitations seems simple, but people have been applying to YC with similar ideas for about 20 years, and no one has really made it work.
  • A “tar pit” idea isn’t necessarily impossible; it’s just much harder than it looks. If you want to pursue one, use Google to see who’s tried it and talk to them if you can, so you can find out exactly where they got stuck.
  • The third trap is jumping into the first idea you think of without checking whether it can become a good business. The fourth is waiting for a perfect idea that doesn’t exist and never getting started. Paul Graham’s advice is to take a middle path: treat the idea as a good starting point.

Ten questions for evaluating a startup idea

  • A startup idea doesn’t need to be perfect from the start, because it’s likely to change. The first and most important question is: Are you the right team to do this? PlanGrid is an example: Tracy knew construction, and Ralph was good at software development. Their skills were a good fit for a product that let people view construction drawings on an iPad.
  • The second question is whether the market can reach a billion dollars. It can already be large, or it can be small for now but growing quickly, like the Bitcoin trading market Coinbase faced in 2012. The third question is how painful the problem is. Before Brex, YC startups wanted corporate credit cards, but no bank would issue them one. Customers didn’t even have an alternative.
  • The fourth question is whether there’s competition. Most good ideas have competitors, but taking on well-established rivals usually requires a new insight. The fifth is whether you or someone you know wants the product. If neither is true, talk to users. The sixth is whether a recent change has made the product possible or necessary for the first time. The expansion of DoorDash, Instacart, and Uber created a need to run background checks on large numbers of workers, and Checkr met that need with an API.
  • The seventh question is whether there’s a large company with a similar business that you can learn from without competing with it directly. DoorDash’s success in other markets suggested that Rappi might be able to make food delivery work in Latin America. The eighth question is whether you’re willing to spend several years on it. If you are, that’s a good sign, but you don’t have to love the field from the start. Your interest can grow as the business takes off.
  • The ninth question is whether the business can scale. A pure software business is easy to scale, while agencies and outsourced development companies often need highly skilled people to serve their clients. The tenth question is whether the idea is in a good field. Choose one that suits the team and offers a reasonable chance of success, so you have related opportunities to pivot to if the initial idea doesn’t work.
  • Fivetran first pitched a data analytics tool to businesses, but nobody wanted to buy it. The team switched to a different tool and pitched it to the same businesses, but still made no sales. Each attempt, though, helped the team understand what those businesses actually wanted.

Three seemingly bad signs that can point to opportunity

  • Fivetran’s founders explored a promising area and eventually found the data analytics tool businesses really needed. A difficult start, a boring field, and existing competitors can all look like bad signs but still point to an opportunity. Most founders will steer clear, leaving the opportunity to those willing to try.
  • Stripe made it easier for websites to accept credit card payments. Before it launched, thousands of developers already knew the existing options were hard to use, but no one had tried to build a similar product. Getting started meant striking special deals with banks and figuring out the details of credit card infrastructure. Those hurdles scared others off and left an opportunity worth about $100 billion to Stripe’s founders.
  • Gusto makes payroll software. Many people knew the existing software was frustrating to use, but they didn’t try to fix it because the problem looked boring. Fun ideas like finding a restaurant or choosing the next song have been tried over and over. Boring ideas often go untouched for a long time, which can make them easier to turn into successful businesses.
  • No matter how interesting the idea sounds, most of the day-to-day work at a startup is still writing code, fixing bugs, and talking to users. By month 6 or 12, when the novelty has worn off and making the product work comes down to execution, how fun the idea seemed at first has almost nothing to do with how happy you are doing the work.
  • Most good ideas already have competitors. Sometimes there’s no competition simply because no one wants the product. A better opportunity is to enter a market that already has competitors and see what they’ve all missed or why none of their products works well.
  • When Dropbox launched, there were already about 20 cloud file storage companies, but most people didn’t use any of them. That showed the existing products hadn’t solved the problem. Founder Drew noticed that they were awkward to use: users had to open a website and manually upload files one by one.

Finding ideas naturally and looking for them deliberately

  • The best way to find startup ideas is to come across them naturally while doing other things. At least 70% of the ideas among YC’s top 100 companies came about this way. Sitting down specifically to think of ideas makes you more likely to keep coming back to “tar pit ideas” that look promising but are easy to get stuck in.
  • If you’re not in a hurry to start a company, you can build deep expertise in a valuable field, work at a startup to develop your skills, or build something that interests you without treating it as a business from day one. Replit started as a project Amjad was interested in, not something he built to start a company.
  • If you need to find an idea now, the first place to look is at what your team does best, because the ideas you find there will naturally fit the founders’ experience better. Rezi’s founders had experience in real estate and debt financing. After joining YC, they looked only at the intersection of real estate and fintech and quickly found a direction similar to Opendoor, but for rental apartments.
  • The second method is to start with a problem you’ve experienced firsthand, ideally one you only got to see because of your particular background. VetCove’s two founders are brothers who grew up watching their veterinarian father call suppliers to place orders. They built a website where vets could buy supplies the way they shop on Amazon.com.
  • VetCove had no competitors when it started. Vets knew how much of a hassle ordering supplies was but rarely started tech companies, while Silicon Valley programmers didn’t know the problem existed. To find similar opportunities, have each founder go through their jobs, internships, and other life experiences one by one and list the problems they encountered, things outsiders don’t know, and opportunities they saw because of their particular circumstances.
  • The third method is to build a product you wish existed. When DoorDash’s founders were undergraduates at Stanford, they wanted to order from nearby restaurants and have the food delivered to their dorms. They couldn’t do that at the time, so they started DoorDash. This method is the most likely to produce “tar pit ideas,” so before you start, ask: Why doesn’t this product already exist?

From recent changes and interviews to launching

  • Look for opportunities in recent changes. The pandemic changed people’s everyday behavior. When Gather Town’s founders saw that their original idea wasn’t going well, they pivoted to a product that made getting together online easy and fun.
  • You can also take inspiration from a recently successful company and adapt its model to a different market. Nuvocargo is like Flexport for Latin America, helping US companies import goods from Mexico. After deciding while at YC that his original idea wouldn’t work, Deepak systematically looked for a new direction. He was drawn to the size of the market and to Flexport as a model. Although he didn’t have deep experience in imports and exports, he had contacts who could help him get started, and he believed he was good at running this kind of operations-heavy business.
  • Finding ideas through interviews takes skill. First pick a field that might have plenty of opportunities, then ask potential customers what problems they face and ask founders in the industry which problems are worth solving. AtoB’s founders didn’t know the trucking industry well, but they saw that it was large and hadn’t yet been significantly changed by startups or software, so they chose to focus on trucking software.
  • AtoB’s founders drove to truck stops themselves to ask drivers what problems they ran into, and they also sought advice from trucking industry founders. The interviews gradually helped them see which directions had potential. After comparing several ideas, they decided to build a fuel card.
  • Large industries that don’t work well may offer opportunities to change how things are done. If you have neither an idea nor a co-founder, you can use Startup School’s co-founder matching service to find someone who already has an idea and is looking for a partner.
  • It’s often hard to tell in advance whether a startup idea is good. Usually, the only way to find out is to actually launch. If you’re still unsure after doing the analysis, launch and see what happens.

What it answers

How can you tell whether a seemingly simple idea is a trap?
Look up who has tried it and talk to them if you can, so you can find out exactly where they got stuck. The fact that no one has made it work for a long time doesn’t mean it’s impossible, but it may mean the hard part isn’t building an app; it’s overcoming a deeper structural obstacle.
What should you ask first when evaluating an idea?
Jared would first ask whether his own team was the right one to do it. In the PlanGrid example, Tracy knew the construction industry and Ralph was good at software development. Their skills matched what the product needed.
Is a market worth pursuing if it’s small right now?
It’s worth considering. The talk treats both markets that are already large and markets that are small but growing quickly as good options. When Coinbase started in 2012, the Bitcoin trading market was still small, but if Bitcoin took off, the market could reach a billion dollars.
How can you actively look for a problem worth solving if you lack industry experience?
Pick a field that might have opportunities, then interview both potential users and founders in the industry. The AtoB team went to truck stops to talk to drivers and sought advice from industry founders. They compared several directions before deciding to build a fuel card.
If you’re still unsure after doing the analysis, should you keep waiting?
Treat the idea as a starting point that will evolve, not as a destination you have to pick correctly on the first try. If you’re still unsure, Jared’s advice is to launch and let the results answer the question.

Quotes

The best way to find a startup idea is to start with a high-quality problem.
It doesn't matter if something is a good startup idea for someone else if it's not a good idea for your team.
Like fun ideas get picked over. Boring ideas get left on the table for a long time.
That is my advice for you: just launch it and find out.

Mentioned

  • Jared — The speaker in this talk, who draws on YC company examples and his experience advising teams to explain how to find and evaluate ideas.
  • Y Combinator — Called YC in the talk. Its 100 highest-valued companies, examples of companies pivoting, and rejected applications are the speaker’s main sources.
  • Paul Graham — The speaker recommends his essays “How to Get Startup Ideas” and “Schlep Blindness.”
  • 创始人与市场的匹配度 — A measure of how well suited a team is to pursuing an idea, and the speaker’s most important evaluation criterion.
  • 创业方向领域 — A group of closely related startup ideas. Choosing the right group gives you nearby opportunities to pivot to if the initial idea doesn’t work.
  • PlanGrid — Built an app for viewing construction drawings on an iPad, illustrating how industry experience and software development skills can complement each other.
  • Fivetran — Gradually found the product customers really wanted by repeatedly pitching different data analytics tools to the same customers.
  • Stripe — Helps websites accept credit card payments and shows how the hurdles of working with banks and dealing with payment infrastructure can scare off other founders.
  • Gusto — Provides payroll management software, showing that a boring field can produce a good business.
  • Dropbox — Entered a cloud storage market with about 20 existing companies and replaced manual uploads with automatic syncing.
  • Replit — Started as a project Amjad was interested in, not as something he built to start a company.
  • AtoB — Provides fuel cards to truck drivers. The idea came from the team’s systematic interviews in the trucking industry.