Thaisan Tonthat | Writing

Four approaches to starting a company

In my work with students at Stanford, I frequently get asked some variation of the question, "What is the right way to start a company?" (to which the answer is: "there is no right answer!").

There are clues from how others got started, however, that can point you in the right path.

The origin stories founders tell are always cleaner in retrospect than they were in real time, so it's not obvious how successful folks actually got started. Here are the patterns I've observed of how companies get going:

  1. The Lean Startup Approach
  2. The Deep Conviction Approach
  3. The "Messing Around with Technology" Approach
  4. The Organic Approach

The Lean Startup Approach

The rapid build, test, and learn ethos emerged as a constellation of ideas in the last era of software startups. Written about by folks like Steve Blank, Eric Ries, and faculty at Stanford's d.school, this set of tools is largely centered around validating demand. After all, you can't build a business selling people something they don't want.

Some tools and attitudes that arose from this are: being hypothesis driven, bias toward action and learning, iteration over planning, user centricity, and failing fast as a virtue.

In practice, this manifests as essentially: "test out ideas and see where they land." This pattern is distinguished by being hypothesis-driven in finding the problem space itself. The extreme case of this is founders who have no commitment to any single idea at all, they just want to find something the market wants.

This pattern works well with high aptitude, curious generalists and business hustlers who enjoy exploring many spaces and testing GTM.

The tension worth noting: these ideas were developed as antidotes to over-planning and premature scaling. But absorbed uncritically, they can become permission to have no conviction at all, which is a different failure mode than the one they were designed to fix.

The Deep Conviction Approach

The founder starts with a strong hypothesis and deep conviction around a space or idea. Usually, the problem is obviously real and the solution is obviously needed. If you'd invented a cure for cancer, you wouldn't run tests to see if there was demand (it's a feasibility question). Experimentation happens within the pursuit, not around whether to pursue it at all.

This is Elon and SpaceX. The conviction that humanity needed to become multi-planetary wasn't something he A/B tested his way into.

This archetype shows up with zealots (like Elon) or domain specialists. A cardiologist who spent a decade in school and years more seeing patients isn't going to brainstorm random startup ideas. They're going to go with what they know. And that's actually a structural advantage: they see problems that only ~40,000 people in the US are even positioned to see. If 10% of those cardiologists have the temperament, desire, and circumstances to want to build a company, you're competing with 4,000 people instead of the world. Deep expertise narrows the competitive field while sharpening the conviction.

"Messing Around with Technology" Approach

In periods of rapid technological change, working at the edges of what's possible often leads to stumbling onto great use cases. You're not starting with a problem. You're starting with a capability and seeing what it wants to become.

"The next big thing will start as a toy." (Chris Dixon)

Coinbase, Figma, and Oculus all started this way. The founders were fascinated by what the technology (crypto, webGL, and VR, respectively) could do before anyone had a clear picture of the market.

This approach has a structural advantage for technical founders specifically. If you can build things yourself, you can justify the exploration on its own terms — the learning is the intrinsic reward, regardless of commercial outcome. That gives you runway to stay in the exploratory phase longer.

Non-technical founders don't have that luxury. The moment you need to recruit a technical co-founder or hire contractors, you're on the clock. You need a compelling enough idea to attract them, and real money to pay them. That pressure forces early convergence on something commercially viable. This is not always a negative, but it usually results in a different shape and trajectory of the company.

The Organic Approach

Some companies don't start with a market opportunity, they emerge as a byproduct of work already in progress. Something you're working on causes you to work on something else.

This plays out in a few ways:

  1. The adjacent thing turns out to be the real thing. You're building something, and a piece of it (a tool, a side component, something tangential) turns out to have more pull than the main project. Slack was the communication layer for a failing video game studio. The game died; the chat lived.
  2. The personal need that generalizes. You solve your own problem first, then notice the solution has broader application. Tobi Lütke couldn't find good e-commerce software to sell snowboards, so he built it, then realized he could sell a hell of a lot more than snowboards with Shopify.
  3. The natural evolution. You follow the logic of your own business model and it takes you somewhere new. Netflix wasn't a pivot. Streaming was just the inevitable next form of "DVDs mailed to your home."

The common thread is that the founders weren't hunting for an idea. The idea surfaced through work they were already doing.

This approach works with founders who are already doing the work in some format. This approach is related to Deep Conviction, but the distinction is: Deep Conviction has a strong vision of the world that doesn't exist yet and is working to realize it. Organic is doing the work already and nurturing the ideas that come out of the work.

Patterns and founder archetypes

These four founding patterns aren't mutually exclusive and most companies will have bits of more than one. But knowing which pattern fits your personal archetype helps you avoid borrowing the wrong playbook.

Are you a domain expert or a generalist? A high aptitude learn-it-all or a seasoned operator? Technical or non-technical?

A deep conviction founder who runs endless lean experiments may be stalling. A technology explorer who tries to force early commercial validation may kill the thing before it has a chance to become interesting.

Take a hard look at your own skills, advantages, and intrinsic motivation. Your path is unique and yours only but looking at how others have done it in the past might give you a hint to what shape yours takes.