Early users are a strange bunch. Here is a worthwhile thought experiment: how many products do you use that you were one of the first 10 users of? I would bet that for nearly everyone that number is 0. Most people are not early adopters. There is nearly nothing any startup founder could say that could lead them to be their first customer, and especially not a paying customer.

There are other people in the world that pride themselves in being early adopters. One of my colleagues, Gustaf Alstromer, worked at Airbnb for many years and proudly enjoyed trying out products from startups and bringing them into Airbnb. Others have such a burning issue that despite them not being an early adopter in general, they are willing to give a new product a shot. My team was once building an inference API that was our first publicly hosted offering. We wanted to ship it within a few weeks and had no interest in figuring out how to securely handle payments and a publicly facing endpoint. So, within 3 days I found and paid a startup to do this for us whose product solved this one problem. I was their first customer. We had a burning problem so it didn’t matter that the startup had little credibility.

The takeaway here is that finding your first user is more of a search process than a persuasion process. In the early days, founders should put themselves in a position to find the Gustafs that are early adopters or the Ankits that have a burning problem.

This has several implications that are counterintuitive to most founders:

  • You should probably charge a lot for your first product instead of nothing. Early adopters and people with a burning problem are rarely price sensitive. You should find the people that are willing to try it despite the price. Note that the goal here isn’t necessarily lots of revenue. It’s to get feedback from the target groups you care most about. You’re more likely to get feedback from an angry customer paying relatively high amounts of money than a nobody who isn’t willing to pay.
  • The ways you find these people probably don’t look like how you find normal people. It’s unlikely that a billboard reaches them vs a targeted cold email or a knock on their door.
  • You should probably launch surprisingly early. In the early days, you don’t know much about who these early users are yet, and you want to engineer a wide surface area for them to find you.
  • You should study these people like you’re an anthropologist that has discovered a hidden civilization. What’s their psychology? How do they make decisions? Why would they make the strange choice to trust you?
  • You should run experiments on them. Change your pricing, landing, onboarding flow, or features. You should spend time talking to them and trying to make them love you, but it’s not a big deal to lose one early customer because you don’t have many. If you annoy your first customer, you can probably use your relationship to recover it. Worst case they churn, but there are plenty more out there that have never heard of you that you could go get, especially now that you’ve learned from the early ones. This is one of the protections startups get against big companies — if a big company runs a bad experiment they get an article in the NYTimes. Startups are fighting irrelevance not bad press.

It also affects whether to start with a consumer product or a prosumer/b2b product. Most individuals have a handful of software they pay for. I’m doing pretty well and my total recurring software spend is around $150/mo (YouTube Premium, Spotify, NYTimes, iCloud, Google AI, ChatGPT, Netflix, Apple TV, Amazon). Meanwhile my corporate card has several products that each cost more than my total personal spend. In the AI era, this makes creating AI consumer products particularly hard. Consumer products usually monetize with ads. These days, it’s tough to get enough ad revenue to cover AI API costs, so most opt to charge a subscription, which competes for space in a $150 spending budget. Of course, many consumer companies will still be made, but this is why many AI founders choose to start by selling to prosumers or businesses, or targeting users like doctors that have high advertising value.

Here is an analogy I use to help founders think about their first users: they should think of a startup as a phylogenetic tree. The root node is an amoeba, and the leaf nodes are complex multicellular species like humans or dogs. Almost every product you can buy on the market has run this evolutionary process and morphed from an amoeba to the maturity of a human or dog: millions of users, a refined sales pitch, and clear value. Early startups are more like amoebas. They just have the very basic functions needed to get exposed to external pressures (i.e. what users want). With those, the founders run an evolutionary search through the tree of potential future directions.

As one case study, consider Tesla, and specifically their amoeba: the Tesla Roadster. The lore about the Roadster is that Tesla needed a high margin product to fund their capex investment to make the Model S and eventually the Model 3/Y. That’s probably true, but there is a second interpretation. Tesla was searching for early adopters. They wanted to find the people crazy enough to buy an impractical $150k car that didn’t go very far, didn’t fit much in it, couldn’t publicly charge anywhere, couldn’t be serviced easily, and looked strange.

Tesla’s story reveals another reality: product evolution is path dependent on what the early adopters wanted. Why does the Tesla Model Y, a mass market vehicle, have a faster 0-60 than a Lamborghini and better tech than a BMW, but worse suspension and plushness than a Toyota? Turns out the early adopters and second generation adopters cared more about tech and acceleration than comfort. Would a mass-market vehicle designed in a vacuum have a 0 to 60 of < 3 seconds? Probably not. But it’s an outcome of the search algorithm that Tesla ran that happened to start with a sports car. If the early adopters were willing to pay $150k for a slow, plush vehicle, I bet Tesla’s cars today would look very different.

So when just starting out, think about a minimum viable product as a minimum evolvable product. That is, one that can be subject to market pressures and evolve from there into a mature one. It’s freeing to know that the product will change a lot so it doesn’t have to be perfect from the start. On the other hand, how it changes is likely a function of where it begins.

(and to those wondering what we do at YC, we help founders run this counterintuitive algorithm.)

Thanks to Jonah Kallenbach for reading drafts of this.