⇠ Essays

Game Selection in Startups

Part I in a series on early stage startup learnings

What follows are some learnings about the earliest stage of starting a startup. Part one: Game Selection

Game Selection

In poker, game selection is the art of choosing the most profitable games to play in. This is determined by the quality of the opposing players as the games in which you are clearly the best player will likely be the most profitable. If you are a 95th percentile player but you are only playing in games against 99th percentile players, you are unlikely to make much money despite being one of the best players in the world. Head to your uncle’s home game, and you’ll take the shirt off their backs.

That being said, your uncle’s game may have a buy-in of $100. You may clean up but you’ll only take home $1,000. If you enter the game with the 99th percentile players where the buy-in is $100k, you may come home with $1m. There’s a lot to consider…do you even know that you’re 95th percentile?

Startups are no different. Game selection in startups is critical. But it’s more than trying to maximize your expected value (EV) for any game you enter. Startups are binary, you either live long enough to have an HR department, or you die valiantly.

The most important games for startups are the markets for customers, talent, and capital. In that order.

Market (Customers)

Who is buying your shit? Apologies for the profane language, but I want to get my message across — who is buying your shit and why?

Let’s illustrate why this is so important. Imagine that you’re a world class engineer and so is your cofounder. You can build the best designed, most performant software the world has ever seen. Now all you need is an idea.

Here’s a bad idea: A marketplace for toys. At first glance, this sounds awesome. It’s a huge market because everyone that has kids, or even knows kids, is a potential customer. You are enamored with just how big this market is, there’s potential of every toy maker in American selling on your marketplace. This could be huge!

The problem with this idea is twofold — it’s not specific enough and incumbents with rock solid distribution aren’t going to be dislodged easily.

Now of course it is possible to build a huge marketplace for toys. In fact this sounds novel and cool! But without a clear and compelling articulation of who the customer is beyond parents, and why they are going to stop buying toys from Amazon or Toys-R-Us (or used toys from FB Marketplace), then shrewd investors and entrepreneurs know the path ahead is fraught with risk. And investors don’t like unfettered risk.

Here’s a good idea: Synchronization software for weather balloons. You realize that there is effectively no software for weather balloons. The software used to guide and track weather balloons was written in COBOL in the 80s and hasn’t been updated since. After speaking with weather balloon operators in Houston, you realize that the inability to remotely synchronize weather balloons is a huge challenge that has lead to numerous miscalculations that obfuscates the data they collect.

Furthermore, you find that this problem is the most painful for oil & gas companies in Houston. Who are willing to pay insane amounts of money to solve this in order to optimize their yields. In fact, they are willing to pay you to do this before you write a single line of code, you don’t even need to raise money — if you can write it, they will buy it. Off to the races.


Obviously, the second scenario is that of a niche. The one defining characteristic of a good niche is that it’s high fidelity, which is another way to say exact. e.g. these 3 offshore oil & gas companies based in Houston deploy this software for their weather balloons which causes this specific pain point. We can run our software on their balloons with a corresponding piece of software for the operators to view them synchronous. The blind can now see.

This is game selection at its best. There’s no one else even at the table.

But isn’t this market small? You and your co-founder are world class engineers. “We don’t want to get stuck writing software for weather balloons in the humidity of Houston for the rest of our lives!”

Did you know that the first location of global hotel giant Marriott wasn’t a hotel at all? It was a 9 seat root beer stand. In fact the second location wasn’t a hotel either, it was a Hot Shoppe (restaurant). It wasn’t until 30 years later, after the company went public as a chain of Hot Shoppes, that Marriott opened its first hotel.

This is not new advice but, it’s way easier to start small with a product or service a tiny amount of people love than trying to be everything to everyone. Would Facebook have worked without gating sign ups by school?

There are a lot of people who run weather balloons beyond Houston oil & gas companies — meteorologists, universities, the defense department, etc. You can expand by selling them your software. It’s also hard to sell into these places. Once you’re in, that’s extremely valuable. You can sell them other software, or balloons, or drones, or whatever! You have your foothold, time to become a massive company.

If you don’t take my word for it, take Paul Graham’s in his essay How to Start a Startup:

"Another way to say that is, if you try to start the kind of startup that has to be a big consumer brand, the odds against succeeding are steeper. The best odds are in niche markets. Since startups make money by offering people something better than they had before, the best opportunities are where things suck most....

...Start by writing software for smaller companies, because it's easier to sell to them. It's worth so much to sell stuff to big companies that the people selling them the crap they currently use spend a lot of time and money to do it. And while you can outhack Oracle with one frontal lobe tied behind your back, you can't outsell an Oracle salesman. So if you want to win through better technology, aim at smaller customers."

Lastly, there’s one way to reverse engineer this — think in terms of use cases. You built your toy marketplace. The use case is to buy your kids their favorite toys. Hmmm…I can already do that, can’t I? The use case for software that synchronizes their weather balloons is that it allows balloon operators who are manually trying to do this with terrible results, to do this remotely with software. The use case is clear and valuable to a specific customer. You have derisked your ability to make money.

10x Better

But Jeff, what if our product is so good, so revolutionary that we can distribute to our TAM from day one? It’s possible! Slack (basically) did it. Ford did it. It’s an extremely lucrative strategy if you have the cards to pull it off. Do you have the cards?

What we’re talking about is the 10x better convention. If you’re unfamiliar, the wisdom goes…if one product seeks to replace the incumbent product by doing the same job but better, then it needs to be at least 10x better than the incumbent to force end users to switch. The 10x better convention is true. Prior to AWS, startups had to manually rack and stack servers, or at least provision them. AWS did this job remotely while turning a fixed cost into a variable cost, this was easily a 10x improvement.

If you build something that’s 10x better, it alleviates some of the need to think about the finer details like a wedge into a niche. But it’s really hard to build something 10x better for a job that’s already being done. You’ll likely need a lot of money and a lot of human capital. Let’s also not forget that when serving a niche, you still want a 10x product.

Those are the cards, do you hold them? If not, you’re playing the wrong game.


Peter Thiel talks about the need to find undiscovered talent. You and your poor startup cannot afford the people that are talented that have already been discovered. The game selection that matters here is — where to look.

If you show up to the Stanford job fair, you will be competing against the best companies in the world. Do you know how much Facebook pays entry level engineers? More than you can afford. If you’re hiring software engineers, a better place to look may be to comb through the top contributors to a lesser known open source project. Is there someone who didn’t go to college that has contributed brilliant code? It’s unlikely that person is on the radar of the recruiters at the Stanford job fair.

You probably want your CTO to have experience scaling an engineering team, but it’s common sense that the ability to spot talent in places others are unable or unwilling to look is a worthwhile arbitrage when filling out your roster.

I’m not an expert here, but the Pistons found Dennis Rodman at the Portsmouth Invitational Tournament after he played his college ball at Southeastern Oklahoma State University, an NAIA school. Had they been playing the same game as every other team, recruiting from Duke and North Carolina, they would have missed the greatest rebounder in NBA history.


This one is simple, play your game. If you build a great company, investors will flock. Listen to Parker Conrad interviews to fully internalize this.


Game selection for startups is the secret sauce. It’s a cheat code. In many cases, it’s the only thing that matters. Had DoorDash (née Palo Alto Delivery) started by competing head on with Postmates in SF or LA, opposed to starting with only Palo Alto, they would have burned through every dollar they had. It would have been the classic case of a deep stack bleeding out a short stack at the poker table.

If you’re unsure, let me assure you — you’re not the deep stack. Do you know who else is at your table?

First published on December 13, 2023