Founder's Guide: First time founders think about product, second time founders think about theses

I.

One of the biggest mistakes we made in my first company was that we thought endlessly about product. We constantly aimed for design polish and brand identity, we agonized over the feeling of specific animations and effects, we spent way too much engineering time and budget on features that people didn't really ask for. The final outcome was beautiful, polished, and totally useless.

This is a fairly common story for first time founders. I admit I'm a bit embarrassed — I read all the Paul Graham essays as much as anyone else, but we made all the classic mistakes anyway.

Still, we managed to raise approximately 7mil on a 25m valuation. That's a solid chunk of change!

Some of that outcome is because we told a great story. We were solving a real problem, and we were solving it in a way no one else could or would, and as a matter of fact our product was actually spot on.

But a bigger part of that outcome was because, entirely by accident, our position in the market was correct. We were an AI company doing applied representation learning at the end of 2020. We were able to raise due to the craziness of ZIRP and the COVID stimulus boom. Then we were able to raise again at the very start of 2023 on the back of ChatGPT and Stable Diffusion both being released the previous August. We had, through dumb luck, stumbled into an ideal raising environment for our company twice.

Looking back on that time period, if I was smarter, I might have framed our positioning more strategically. In 2020 I might have said something like "AI is getting bigger and bigger and there's a real information arbitrage between what AI researchers know is easy and what people think is useful but hard. Money is cheap, let's raise now so we can keep our heads down later, and validate our AI arbitrage thesis". In 2023 I might have said something like "AI has gone full consumer, and everyone wants to get in on it. We can piggy-back on that to raise, but more importantly we can ride the wave of consumer interest into product buy-in and growth."

These aren't product specs or business plans or financial objectives. They are theses about the market. And I've observed that serial founders think about market theses constantly.

II.

A long time ago, I wrote that startups are a stack of hypotheses:

People have all sorts of fun metaphors that help them reason about startups. Things like ‘startups are a bet on the future paid with time and effort’ or ‘early stage startups are like hunting in prehistoric times’ or something else. My personal favorite is: ‘startups are a hypothesis’. I like this framing because it invokes the scientific method. We were all taught this in school, hopefully, but just as a recap: the core purpose of the scientific method is to eliminate biases and perform data driven iteration towards a better understanding of the world. This is done through the construction (pre-registration) of a hypothesis -- defined as a proposed explanation for an observed phenomenon. The hypothesis is then validated (or disproved) through an experiment that is specifically designed to gather expository evidence for the claim. Ideally, the experiment is designed such that the only possible explanation for the results proves or disproves the hypothesis.

…

At the early stages of a startup, the goal is to find product market fit. I think of this as ‘create hypotheses about the world, run experiments, and discard your hypotheses until one of them proves true’. A popular phrase in the startup world is ‘fail fast and fail often’. This naturally relates to the hypothesis framework of startups. There is no value in pursuing a hypothesis for its own sake. A startup should be willing -- eager, even -- to discover evidence that the hypothesis they are pursuing is flawed, and they should do so as quickly as possible to minimize time. Fail fast. And after failing, the startup should pivot, try a new experiment, and go again. Fail often.

Failure here is a strange word, because whenever a startup has disproven a hypothesis and moved on, they have actually avoided the only failure that matters: pursuing dead ends.

The very first hypothesis you need to have is about some kind of market arbitrage — something like "Person so-and-so has problem such-and-such that can be solved by this-and-that". This is just the standard 'find a real problem' advice.

The second hypothesis you need to have is about how the market will change in 5 years. Macro-economic forces like geopolitics and technological disruption can destroy even the best companies. Put bluntly, if you are about to start a company in a space that you think is going to struggle, don't start the company even if it's a decent company at the current moment! You cannot change market and political forces, any more than a horse carriage salesman could stop the sale of Fords, or a US manufacturer could stop offshoring, or local news media can stop the decrease of its readership.¹

And, on the flip side, you want to set your company up to take advantage of where the rest of the market is going to go as much as possible. Your company is a tiny sailboat on a vast ocean; you want to be positioned so that you can catch the tailwinds in your sails. Less poetically, you are almost certainly going to pivot your company in some way at least once. If the market is moving in the right direction around you, you can continue to grow and make forward progress in your company even as you pivot and change direction.

At SOOT, we were wrong about a ton of stuff. Honestly we were probably wrong about more than what we were right about. But we were positioned such that the market buoyed us along anyway.

So when starting a new company / venture, you should be asking yourself: "What is it about the market in the next few years that I believe will propel this idea along? Is there anything about the market in the next few years that will really slow me down?"

III.

Here's a really concrete example, related to the AI stuff that's been going on.

I assume that, in five years, we will have AI models that are better and cheaper than what we have today.

This is, on its face, a pretty straightforward hypothesis. But I think there are some immediately obvious market areas that I would be hesitant to build a company in. For example, I probably wouldn't start building tools for call centers. We already have pretty good voice generation and speech to text, and we have fantastic text processors. The writing is on the wall. Even if right now there are millions of call center employees, I strongly believe they are all going to be out of a job within a few years.

On the flip side, I think that I would be more excited to develop in industries that will benefit from 'cheaper intelligence'. That can include things like building tools for clients of call centers — the Comcasts of the world, for e.g. — to sell them a better, cheaper call center experience yourself.² In this world, your business will naturally become more successful as models get better. In the previous world, your business will get worse.

In my opinion, companies that are well positioned to take advantage of the AI boom include:

• Replit — they are making a bet that more people who are “non-technical” will start producing code. Probably correct!

• Intercom — the job of a first-line customer experience operator is basically automatable already.

• NVIDIA and other chip providers — hopefully obvious.³

Companies that are going to struggle:

• Fivetran, Zapier — as automations become easier, people are going to reach for no-code plug-in solutions less.

• Design agencies in general — the price-per-design-asset has dropped and will continue to drop. On the long tail, fewer people are going to reach for specialized design agencies just because they can use Photoshop. This will likely result in consolidation of taste-makers.

• Upwork and Fiverr — hopefully obvious.

I might be wrong about any of these, so don’t go and put all your savings into shorting Upwork. But I think the general idea holds: market theses are the foundations for critical thinking about company performance.

And, like, you don’t have to take this from me, I’m just some random guy. Instead take it from Bezos. Amazon is probably the most famous example of thesis driven thinking, and Bezos has said multiple times that he started from the assumption that the Internet was going to be everywhere, and worked backwards to figure out what business made sense in that world.

Rubenstein: "I'm going to start a company selling books over the internet, and I'm going to do it from Seattle."

Where did that idea come from?

Bezos: I came across the fact—so this is 1994.

Nobody has heard of the internet, or very, very few people. And I came across the fact that the web, the World Wide Web, was growing at something like 2,300% a year. This was in 1994.

And anything growing that fast, even if its baseline usage today is tiny, is going to be big. So I looked at that and thought, "I should come up with a business idea, get on the internet, and then let the internet grow around this while we keep working on it."

So I made a list of products that I might sell online. I started force-ranking them and picked books because books are unique in one key respect: there are more book items in the book category than in any other category.

At any given time, there are 3 million different books active and in print around the world. So the founding idea of Amazon was to build a universal selection of books.

And Bezos was right! In fact, the continued growth of the internet also drove Amazon’s investment in AWS, which eventually became one of the most profitable divisions of one of the most profitable companies in the world. Imagine if Bezos started and ended with “books are hard to buy”. Probably wouldn’t have gotten as far?

With hindsight, this may all seem obvious — of course you should try and align your company with the way the market is going to swing. But I feel like I need to state the obvious, because every other week I hear about some new company that is trying to swim upstream. The founders see the failures of journalism, or of education, or of manufacturing, and think "a ha, a hole in the market! There's no competitors here, it's greenfield!" And I'm like, can they just not see all the bones scattered around? Maybe there's a hole in that part of the market because it's an extremely deadly pit of toxic gas.

Here be dragons.

IV.

You know who else thinks in market theses? VCs.⁴

O, sure, not all of them.

Smaller VCs tend to only have one market thesis, if they have one at all. Many smaller VCs invest entirely on the back of their network — finding smart people to fund early in their development cycle, betting on talent instead of on a particular opinion about the way the world is going to swing.

Larger VCs, though, will always have a thesis. In fact, they will generally have many theses about the markets at any given time. And their investments are often more about their theses than they are about the company, founders, or product. The most obvious example of market-thesis-driven-investing in the startup world is the YC Call for Startups. It's…literally just a list of things that the investors think will be buoyed by the market. The message is simple: "Come build in this area", says YC, with big flashing lights. "We will fund you." a16z is doing the same thing with American Dynamism — they saw the global instability, saw the rising nationalism, made a rational bet that both of those things were going to continue to go up, and are now trying to attract smart people to win the space. "Smart people building in markets that are naturally going to grow anyway" is the most winning formula you can get from the VC side of the table.

It's worth noting that large VCs have a structural advantage in thesis driven investing. Because they invest in so many companies, they can often get on-the-ground insights about trends in the market that other folks aren't aware of yet. "We need more supply chain software, because all of our manufacturing companies are struggling", that kind of thing. This can, in theory, help them get better terms for better deals than their smaller counterparts. To use a16z again, they continued making crypto investments in 2023 and 2024. Many crypto founders at that time were terrified of the SEC and convinced that they were going to be sued out of existence and possibly personally fined/jailed. A good friend of mine got Series A funding from A16Z and couldn't understand why they would give him anything at all. He was convinced that they were making a huge mistake.

But a16z knew, through separate channels, that Trump was going to be pro-crypto, and that they could put money behind Trump to help him win. It was a bet, for sure, but a calculated one. In the end, not only did their investments pay off, they likely got better terms from their founders who didn't have that additional information.

If you're a founder, part of the benefit of thinking in market bets is that you will be speaking the VC's language. That in turn makes it easier to get them excited about what you're working on.⁵

V.

All of this naturally leads to the question: how do you know what's going to happen in the market?

This is literally a trillion dollar question, if you can accurately predict everything that will happen in a few years you can make tons of money and probably startups are not the best ROI thing you could be doing. The simple reality is that you don't really know what's going to happen. But if you have any area of expertise at all, you can make some educated guesses. And then you go out and try and validate them the best you can, like any other hypothesis.

Net net, you might be wrong. You may find evidence that your theories of the future are backwards. That's ok! First, it's much better to learn that early rather than later. And second, the simple act of thinking in terms of theses (instead of products) will make your explorations more fruitful, and will save you a lot of time when figuring out where to pivot next.

1. People talk a lot about how specific media outlets could do ABC in order to retain relevance and viewer count. "O, NYT is too liberal, it needs to be more centrist" things like that. But the reality is that the industry is dying regardless, and individual companies left in the space are picking up smaller and smaller scraps.

2. I believe they call this “disruption”.

3. Not to brag, but to brag a little: I invested in NVIDIA nearly 6 years ago because I looked around at what I was seeing at in the ML world and went “o, duh”.

4. And, like, regular long-short investors and so on. But this is an entrepreneurship themed post so whatever.

5. It would be a mistake to read this section and think that it is somehow better to take investment from larger institutional VCs than smaller ones. Thesis driven investors can be pushy, and may interfere a lot more than one that trusts your talent as a founder. And there’s risk too — if the market thesis shifts or if you shift, your once eager VC may no longer want to continue to invest. As with all things, it’s a trade off.