First Lesson for Researcher-Founders: You Already Speak Lean Startup

Researchers who start companies walk in holding a card most founders don’t have.

Author: Koutian Wu; GitHub: ktwu01

A few days ago I read a post arguing that, at work, people with real research training see problems differently from people without it. The author’s example was PhDs versus master’s graduates. PhDs think more about feasibility and long-term value. Master’s graduates lean toward quick wins and what pays off right now. The author added one caveat: the label is about how much research training someone has had. A master’s student with enough of it counts as the first kind.

My first thought was, that’s just Lean Startup.

Eric Ries published The Lean Startup in 2011, and the core fits in a paragraph. Break your idea into leap-of-faith assumptions, the premises that kill the business if they turn out false. The two that matter most are the value hypothesis (do users actually get value once they use it?) and the growth hypothesis (how will new users find it?). Then test them as cheaply as you can. The cheapest thing that runs the test is the MVP, the minimum viable product. Look at the data, learn something, and decide whether to persevere or pivot. Ries calls the loop Build-Measure-Learn.

Now rename the steps.

Form a hypothesis. Design an experiment. Analyze the results. Revise.

That is what every researcher does every day. Ries says so plainly in the book: a true experiment follows the scientific method. It starts with a clear hypothesis, predicts what should happen, and tests that prediction against real data. He calls what comes out of it validated learning.

So I have long thought researchers enter founding with a privilege. For most founders, Lean Startup is a method they have to study and drill. For researchers, it’s a native language.

Where does the privilege actually show up? I count at least three places.

First, the habit of asking whether something is worth doing. Anyone who has done research has had an advisor or a reviewer ask “so what?” After enough of that, you ask yourself before you start.

Second, failed experiments don’t scare you. In a lab, nine ideas out of ten don’t work, and that’s a normal week. A negative result is still a result. It tells you where not to go next. A lot of first-time founders ship a feature nobody uses and start doubting everything. A researcher’s reflex is to ask what that taught them.

Third, you know how to state a question so it can be proven wrong. “I think users will like it” can’t be tested. A threshold set in advance can: if fewer than some share of target users will pay, this direction is wrong. You set the bar before the experiment, and you don’t move it after you see the result.

But this is the part of the first lesson that matters most.

Having the privilege is not the same as using it.

The classic researcher mistake is spending all the validation effort on technical hypotheses. Does the model work? Can we get two more points of accuracy? Researchers are good at those questions, and comfortable with them. Lean Startup says the first thing to test is the value hypothesis: does anyone actually need this? You can’t answer that in a lab. You have to go talk to users. Lean Startup grew out of Steve Blank’s customer development method, and Blank’s most famous line is get out of the building.

Honestly, researchers know benchmarks far better than they know users.

I also disagree with one part of the original post. It implies that thinking longer beats moving fast. Lean Startup actually argues for moving fast. What it opposes is disappearing for a year to build something big in one shot. The difference is that each fast move aims at a specific hypothesis, so you learn something when it’s done. Moving fast without a question gets you only the short-term payoff. Moving fast with a question teaches you where to go next, even when it fails.

So the degree doesn’t create the gap, and neither does thinking longer or moving faster. The gap comes from having a falsifiable question in mind before you start. On that, I agree with the post’s last line: what counts is the training.

The first lesson for researcher-founders fits in one sentence.

You already know how to run experiments. Now run them on people instead of data.

The card is real. It only counts when you play it on users.