What a startup actually is

Where this lesson sits. Lesson 1 of 5 in How Startups Actually Begin. Root lesson, no prerequisites. Next: The founder is the product.

Three different people describe their work as “a startup.” One runs a 20-person agency that takes web-design clients on retainer. One sells handmade ceramics on Etsy and ships 200 mugs a year. One is a PhD researcher with a working prototype of a battery chemistry that does not yet have a customer.

None of those three is building a startup. One is building a small business. One is building a side project. One is building a research lab. The label “startup” got attached because the word is everywhere, but the actual shape of each of those things is different. The decisions each of those three should make next are different. The funding they should take, if any, is different. The success metric they should chase is different.

Most early-career founders skip this step. They call what they are building a startup because that is the word they have for “company in progress.” Then they make startup-shaped decisions on a small-business-shaped opportunity, or vice versa, and they wonder why nothing fits.

A working definition

A startup is a temporary organization built to discover a repeatable, fast-growing, profitable business model under high uncertainty.

Each piece of that sentence matters.

Temporary. A startup is not a permanent organization. It is the search phase before there is a real company. Once the model is discovered (repeatable, fast-growing, profitable), the startup becomes a company and stops being a startup. If you are 8 years in and still calling yourself a startup, either you are still searching (which is a signal) or the word has lost its meaning.

To discover. A startup is doing search, not execution. The model is unknown when you start. You are running experiments to find one. A small business is executing a known model. The work feels different in your hands when you are doing search vs. execution, and confusing the two is the source of most early-stage mistakes.

Repeatable, fast-growing, profitable. All three. Repeatable means the next customer is found the same way the first one was. Fast-growing means more customers cost meaningfully less per unit than the first one did, so the company can grow without proportionally adding people or capital. Profitable means the unit you ship costs less to produce than the customer pays for it. If any one of those is missing or not yet proven, you are still searching.

Under high uncertainty. You do not know which experiment will work. You do not know your customer. You do not know your price. You do not know your channel. The uncertainty is real and is what justifies the temporary search phase in the first place.

What it is not

Small business. A small business runs a known model. The bakery, the dental practice, the consulting firm. The model is repeatable, growable to the owner’s goals, and profitable from day one. The uncertainty is low. The work is execution, not search. Small businesses are great. They are not startups, and the venture capital industry is the wrong shape of capital for them.

Agency. An agency sells services priced by the hour or the project. It can be small or large, profitable or not, well-run or chaotic. It is a known model. There is no search happening unless the agency is using its current revenue to fund the search for a productized version of itself (which is a real thing, and is then a startup-inside-an-agency, with all the conflicts that creates).

Side project. A side project is built without growth ambition. It exists because the maker wants it to exist. The success metric is “I made it.” If a side project starts to attract customers and the maker decides to chase that, it becomes a startup. Until that decision, it is a different thing and the decisions about it are different.

Research lab. A research lab discovers technology, not a business model. The customer is “whoever pays for the research” (often grants or a parent institution). If a research project finds a real customer who pays for the output of the research, and the team decides to chase that, the research becomes a startup. Until then, the success metric is publication or proof, not revenue.

The point of this taxonomy is not to be a snob about which is which. The point is that each shape has its own decisions, its own funding, its own team structure, its own success criteria, and trying to run startup-shaped decisions on a small-business-shaped opportunity (or vice versa) is the single most common reason founders stall in year two.

The five-question test

Before you call what you are building a startup, score these five honestly.

1. Is the business model still unknown to you? (Not the product, the model. How customers find you, what they pay, how often, how you grow.)
2. Could the thing you are building plausibly reach 10x your current size in 3-5 years if it works?
3. Are you willing to take outside capital and trade equity for runway to find the model?
4. Do you have a hypothesis about why this is venture-scale, and a willingness to be wrong about it?
5. Would you keep doing this even if you discovered the model was not what you thought it was?

Five out of five means you are building a startup. The decisions in the rest of this course apply to you.

Three or four out of five means it is borderline. You may be building a startup that has not committed to being one, or a small business that is shopping for the wrong identity. Worth sitting with the answer before you take the next meeting.

Two or fewer means it is probably something else. That something else might be the better thing to build. A profitable small business with 30 employees and no investors is a much better outcome for most founders than a venture-backed startup that lost the search and burned out in year three. The taxonomy is a tool for honesty, not a status hierarchy.

When the label is wrong, name it

If the five-question test says you are not building a startup, the move is to relabel. Tell investors no. Tell yourself no. Build the thing you are actually building, with the funding and team and success metrics that fit the actual shape. The honest relabel saves you 18 months you would have spent making the wrong decisions.

A version of the sentence: “I thought I was building a startup, but the more I look at the actual business, the more I think I am building [a small business, a consultancy, a research project, a side project]. Different rules apply. I am going to make the decisions that fit the actual shape.”

That sentence is rarer than it should be. It is also one of the cleanest moves an early founder can make. Calling the thing what it actually is unlocks the right next decision. Calling it a startup when it is not stays expensive for years.

---

A note from the team. This course is from TAKE INTEREST Inc. We build tools for people whose work depends on remembering context. Every decision, every conversation, every reason a strategy moved or did not. If you are early in founding something and the five-question test made you think harder about what you are building, we are open to design partners. The contact form is the door. Short message, ~48 hour response.

Take Interest Inc. (2026). What a startup actually is. TAKE INTEREST. https://takeinterest.ai/blog/what-a-startup-actually-is