10 Academic Habits PhDs Must Unlearn to Build a Startup
Y Combinator's Startup School breaks down the academic instincts that sabotage PhD founders, from vanity metrics to novelty obsession, and how to fix them.
Why do so many PhD-led startups struggle despite brilliant founders?
PhD founders often fail not because they lack intelligence but because the habits that made them successful in academia actively work against them as founders. Years spent optimizing for grades, citations, novelty, and rigor train a mindset built for peer review, not for customers. Y Combinator’s Startup School has outlined ten specific shifts PhDs need to make to think like founders instead of researchers, covering everything from how they validate ideas to how they talk about their own competence.
TL;DR
- Vanity metrics like citations and grades don’t translate to startups, where the only question that matters is whether people will actually pay for what you built.
- Academics chase what’s intellectually interesting, but founders need to obsess over what’s important to users, even if they can’t fully explain why something works.
- Novelty is overrated in startups. Plagiarizing a working business model (Lyft copying Uber, Anthropic following OpenAI’s lead) is often a smarter path than inventing something new.
- There’s no finish line in startup life the way there is with a dissertation defense or a paper submission, so founders need a long horizon vision instead of hopping between ideas.
- Founders have to physically get out of the building and talk to customers instead of assuming they understand a problem from inside a lab or office.
- Decisions get made on partial information. Waiting for complete data before shipping, the way you’d wait before publishing a paper, means a competitor beats you to market.
- Founders can’t delegate the uncomfortable parts of the job like sales, hiring, and firing, because doing that work is how you build real empathy for customers.
What is a “vanity metric” and why does it trip up academic founders?
In academia, progress gets measured in things like citation counts, grades, published papers, and invitations to speak at conferences. These are vanity metrics: they signal status within a closed system of peers, but they don’t tell you whether anyone outside that system wants what you’ve made. Paul Graham has pointed out that one of the most damaging things people learn in school is how to optimize for grades, a skill with no real analog in the business world.
The founder mindset flips this entirely. The only metric that matters is whether people will actually pay for your product, not whether they say they like the idea. A shiny new algorithm might impress a dissertation committee and still get zero real-world adoption. Founders find this out by talking to potential customers early and often, not by assuming demand exists because the underlying research is sound.
Why does chasing novelty hurt startups more than it helps?
Academic publishing rewards originality above almost everything else. A paper gets rejected if it’s not sufficiently novel. Startups work by the opposite logic. Plagiarism, in the sense of copying a business model that already works, is frequently the better strategy.
There’s a useful distinction here between what might be called MIT-style startups and Stanford-style startups. The MIT-style company centers on a novel invention, a patented black box that does something unprecedented. These rarely scale into large businesses because customers don’t inherently want a black box, they want a problem solved. The Stanford-style startup instead starts with a boring, well-understood customer problem, like getting groceries delivered or booking a hotel room, and iterates on execution for years. Instacart didn’t invent grocery delivery. It simply executed the idea better than the companies that tried before it, and has kept refining that same core problem for over a decade.
The same pattern shows up across tech history: Lyft followed Uber, Anthropic followed OpenAI, Square followed PayPal. Novelty isn’t required to build a big company. Solving a real, underserved problem is.
How should founders handle uncertainty differently than researchers?
Academic work rewards rigor and exactness. A result needs replication, peer review, and statistical confidence before it’s considered valid. Startups can’t operate that way. Jeff Bezos has described making most decisions with something like 70% of the information you’d ideally want, because waiting for complete certainty means moving too slowly.
The historical example often cited here is the “database wars” between Oracle and Ingres. Michael Stonebraker’s Ingres had a more technically sophisticated product, but Larry Ellison pushed Oracle’s SQL into the market faster and won. Ellison’s reasoning was blunt: even if the competing product was better, speed to market mattered more. For founders coming from research backgrounds, this is often the hardest shift, because publishing with incomplete confidence feels like malpractice. In a startup, waiting for 100% of the data usually means a competitor has already shipped and captured the market.
| Vendor | Tier | Risk | ApprovalNEW |
|---|---|---|---|
| Northwind Supply | Tier 1 | Low | Approved |
| Acme Logistics | Tier 2 | Medium | Approve |
| Helios Data | Tier 1 | High | Pending |
| Granite Freight | Tier 3 | Low | Approved |
| Pinecrest Labs | Tier 2 | Low | Approve |
This same comfort with imprecision applies to marketing and sales language. Academic writing demands exactness: every claim needs a citation or a caveat. Advertising doesn’t work that way. A punchy, slightly inexact slogan that gets people excited will outperform a scientifically hedged pitch every time, as long as the underlying claims aren’t dishonest.
Can PhD founders delegate the “business side” and just focus on research?
No, and this is one of the more common traps. Many academic founders assume they can hire an MBA or a business co-founder to handle sales, marketing, hiring, and fundraising while they keep doing the research they love. This rarely works. A startup CEO is responsible for everything: cash runway, hiring and firing, product decisions, customer conversations, and company culture, not just the technical roadmap.
Abstracting yourself away from sales and customer conversations means you never build empathy for the people you’re supposedly building for. Without that empathy, it’s nearly impossible to build something people actually want. The research itself, the part academics tend to love most, often has to take a back seat once the company is in founder mode.
Why does “getting on a plane” matter so much for product market fit?
Finding product market fit is a brute force process that requires constant iteration based on direct customer feedback. Academics are trained to work in relative isolation, forming hypotheses and testing them without much outside input. That instinct, useful in a lab, is a liability in a startup, because it leads founders to assume they understand a customer’s problem without ever validating it with the actual customer.
A concrete illustration: a computer vision startup spent three months building a people-tracking system for retail stores, something every retailer claimed was a major pain point. The resulting data showed customers spent a lot of time near checkout counters and deli counters at lunchtime, an unsurprising and low-value insight. When asked if they’d pay for it, the retailers said no. The same conclusion could have come from manually watching a few hours of video footage, saving three months of engineering time. The lesson: validate demand with a scrappy, human-in-the-loop version of the product before investing heavily in building the real thing, especially in AI companies where model training is expensive and slow to iterate on.
And “talking to people” means talking to the people who actually have the problem, not investors. Investors are frequently wrong about what will become a good product, which is part of why so many of them pass on companies that go on to succeed.
Frequently Asked Questions
Why do PhDs often struggle as startup founders?
Academic training rewards novelty, rigor, grades, and citation counts, none of which predict whether customers will pay for a product. The instincts built over years of research, working in isolation, chasing intellectually interesting problems, waiting for complete data, tend to work against the speed and customer focus startups require.
Is it actually okay to copy another startup’s idea?
Yes, within reason. Plagiarism of a business model is common and often successful, as long as the company is solving a real customer problem that the market hasn’t adequately addressed yet. Examples include Lyft following Uber and Anthropic following OpenAI’s general approach. What matters is execution, not originality.
What does “founder mode” mean for decision-making?
It means acting on partial information instead of waiting for certainty. Jeff Bezos has described making most decisions with around 70% of the information founders wish they had, because waiting longer usually means a competitor ships first and captures the market.
Should PhD founders still do the research themselves?
Generally no, at least not as the primary focus. Running a startup requires handling hiring, sales, fundraising, and customer relationships directly. Delegating all of that to someone else while retreating to research work tends to disconnect founders from the customer empathy they need to build a good product.
What’s the biggest mindset shift academics need to make?
Replacing “what is intellectually interesting” with “what do customers actually want and will pay for.” That shift requires leaving the office, talking directly to the people who have the problem, and validating assumptions before investing heavily in a solution.