Where are good startup ideas born?

Once upon a bubble, founders coveted their startup ideas like winning lottery tickets.

Since then, the landscape has inverted and the common wisdom is that “ideas are worthless and execution is everything”. This is, of course, complete malarky: Elon wouldn’t be Elon if he had chosen to build an iPhone app that reminds you to feed your dog.

The obvious truth is that a good startup idea is a necessary but not sufficient ingredient for success in the startup journey. A poorly thought-out idea is like starting a trip without a route: there’s a chance that you’ll be happy about where you end up, but you’re also not doing yourself any favors.

Coming up with a good startup idea is a three step process. You need to find a problem, find a way in which technology can help solve that problem, and then analyze your solution to see if you can build a good business around it. When it invariably turns out that your idea sucks, there’s a fourth step where you soak that idea with gasoline and light it on fire.

Find a problem. Find a solution. Analyze the business. That’s all that it takes. Now, you just have to do this over, and over, and over again until you stumble across a few ideas that make you wonder, “Huh – could that actually work?"

A fire.
Your keepsake box for your roommate's app ideas.

Creating startups and shaping the future is an inherently creative process. It turns out that discarding promising raw material until you find something interesting is very common among creative endeavors. Ken Burns, the famous American documentarian, uses a beautiful phrase to describe the importance of unused ideas: “honoring the negative space of creation”. In his words, “It takes 40 gallons of sap to make one gallon of maple syrup.” He was alluding to the fact that every second of used footage in his documentaries corresponds to 40 seconds of promising but unused footage.

For startups, this might mean that you can expect to come up with 40 startup ideas to find one that’s promising. Actually, you’re not Ken Burns. Let’s make it 80.

Of the skills that you need to find a good startup idea, the best one to learn first is probably the one that you use last in the process: analysis. After all, it doesn’t make much sense to set out on an elephant hunt if you don’t even know what an elephant looks like. So without further ado:

What makes a good business?

A giraffe in crosshairs.
An elephant hunt nears its unfortunate end.

Entire volumes have been written about this: I’m not going to rewrite them poorly here.

Wait – actually I am. Here’s my short, short version of what makes a good software business:

  1. Is there a real, pressing problem?
  2. Does your business solve the problem in a significantly better way than the alternatives?
  3. Do you have a realistic plan to grow your business?
  4. Is the total addressable market large enough?
  5. Once your business starts to succeed, is there something that will prevent competitors from just copying you?

If the answer to any of those is “no”, it’s probably not a good idea.1

Bear in mind that the above list certainly isn’t comprehensive. Theranos answered “yes” to all of these, but failed because the thing couldn’t actually be built. WeWork answered “yes” to all of these, but failed because there weren’t real economies of scale. If your idea answers “yes” to all of those questions but requires a billion dollars to get off the ground with no intermediate validation that might be used to secure funding, you have – you guessed it – a bad idea.

My advice to help develop your sense of what makes a good startup:

  • Listen to the Acquired podcast. In each episode, they dig into how a company was built and then assess the company’s prospects going forward. This podcast has been immensely helpful for me in evaluating the prospects of growth and late stage startups.
  • Read The Lean Startup by Eric Ries. While it’s fiercely debated whether this book’s frugal prescription is actually realistic nowadays, I still think that it’s a great tool to get you into a mindset of scrappily evaluating startup ideas.
  • Read Paul Graham’s essays on his website. (I recommend Do Things That Don’t Scale as a first read.) Paul Graham was a founder of the YCombinator accelerator and has mentored many successful startups in their early stages.
  • Read Andrew Chen’s book The Cold Start Problem. Andrew Chen took his experience on the growth team at Uber and as a VC at Andreesen Horowitz and wrote a wonderful primer on how to kickstart marketplace businesses.
  • Read Andrew Chen’s blog, which is similarly insightful.2
  • Read Hamilton Helmer’s 7 Powers, which is an incredibly useful framework for assessing whether companies are likely to eventually get their profits competed away.

At some point while doing these things, you might realize that you have the vocabulary to describe exactly why your roommate’s app idea is so terrible. That’s probably a good sign.

Equipped with your new superpowers of analysis, you can move on to understanding:

How does software solve problems?

Your goal here should be to develop some sort of Swiss Army knife of “how does software provide value in the world”. A few examples:

  • Software can aid with discovery (Airbnb, Google, Uber): Take an activity that people are doing or would like to do and help find other people that want to participate in that activity.
  • Software can be a trusted 3rd party (Airbnb, Wise, Uber, Kickstarter): By acting as the trusted broker between two other parties, you can facilitate exchanges that might otherwise be too risky to take place.
  • Software can provide something faster or cheaper than non-tech-enabled competitors (Gusto, Vouch)
  • Software can centralize a hard process that lots of companies are already doing poorly on their own (Plaid, Stripe, Vanta): Lots of companies can pool resources through yours to do something hard well instead of each company competing in a silos to see who can do that thing least poorly.

The same resources I listed in the previous section should go a long way towards helping you develop a sense of how software solves problems. The key here is that, as you hear smart people dissect different businesses, you should start to notice patterns in how those businesses provide and capture value. The more concisely you can use words to describe those patterns, the more effective your brain will be at using those patterns to solve new problems.3

If you know how software provides value, then you just have to:

Find a problem

There are obviously no hard rules about “how to find interesting problems”, but a few things that I’ve found very helpful in the past are:

  • Reading a book about an unfamiliar field
  • Reading books that take place in different worlds (science fiction / fantasy)
  • Become deeply integrated into a technology/culture “wave”
  • Traveling to new places
  • Going to a conference for an unfamiliar field (I haven’t done this myself, but have heard it recommended by others)

Looking at these, there’s a clear pattern: innovation is almost always the result of applying techniques from one domain to solve problems in another. It’s a little more like “domain arbitrage” than it is coming up with something new: you’re just assembling familiar components in unfamiliar ways.

There are pitfalls to this approach, though. Trying to solve a problem in a field you’re not yet familiar with is also a recipe for falling prey to the Dunning-Kruger effect, where you’re overconfident in your ability to solve a problem precisely because you know so little.4 Because of this, it’s important to approach any problem you stumble upon with a great sense of humility and an earnest desire to learn. Only after you’re knowledgeable about the problem should you feel vaguely qualified to build a solution.

Tons of startup ideas revolve around using technology to solve problems that have existing but maybe suboptimal software solutions. In my opinion, this is one of the toughest ways to build a business.

To illustrate why this might be hard, imagine that you sell shovels in a frontier town. Because you’re the only seller in the market, you just have to let people know you exist in order for your business to take off. If you sell shovels in Farmville, USA between the Tractor Supply and the Home Depot, your shovels are going to have to be God’s gift to digging in order to attract customers. Even then you may find yourself in a hole that’s hard to escape.

It’s also entirely possible to build a successful new product ignoring all of this advice: as an example, I don’t think there was any particular reason why Webflow couldn’t have been built 5 years earlier. However, this approach is certainly harder. Vlad Magdalin, the founder of Webflow, had to found the company three times and work for over 10 years to crack the formula for no code website building. Heck, just look at the Hey.com and Superhuman teams trying to break into the email space with (what I believe) are two really interesting products. In both cases, it took talented teams years of building before they had something with even the minimum expected feature set to put in front of customers. And it’s not yet clear whether those teams will ultimately succeed.

A good startup idea can act as a strong tailwind to your startup, amplifying all of fundraising, hiring, and selling that will be necessary to grow. If you’re going to put years of your life into making your startup idea a reality, why not do so with the wind at your back?


  1. #5, “Is there something that will prevent competitors from just copying you?”, is probably negotiable as long as you keep your ambitions modest enough to fly under the radar. ↩︎

  2. If you can get past the full screen, unclosable “Please subscribe!” message, there’s actually good content there. Good luck. ↩︎

  3. This process of “consolidating lots of information into concise patterns” is often called “building mental models”. The efficacy of mental models is what separates experts in a field from novices – whether that field be chess or chemistry. The Barbara Oakley’s course Learning How to Learn has a ton of great information about this process and I highly recommend it if you’re interested. ↩︎

  4. In a surprising reversal of stereotypes, I actually find programmers to be much more susceptible to this bias than MBAs. ↩︎

Want to hear when I write something new?

You may also like

» Return to all posts