पाठशाला Pathshala · दल Dal, The team · Lesson 01 · Start

Choosing a co-founder: the forty questions before you commit

A co-founder is the largest unpriced decision a founder makes. Forty questions on money, roles, risk and exit, asked in four sittings before anything is signed, replace the hope with a record.

Pathshala, The Founder Library · 11 October 2026 · 7 min read

Founders spend weeks choosing a laptop and an evening choosing a co-founder. The co-founder will decide what the company is worth, whether it can raise money and whether its first bad year is survivable. The evening is not enough. This lesson gives it a structure: forty questions, four sittings, one trial project and three disqualifiers, all before anyone files a form on MCA.

It is written for the founder who has a candidate in mind, and for the one who has nobody yet and is wondering whether to go alone. The second case has its own lesson, [the solo founder](/library/solo-founder-how-to-make-it-work). This one assumes there is a person across the table.

Why the conversation has to happen before incorporation

Noam Wasserman spent a decade at Harvard Business School studying founding teams, and his conclusion, repeated across The Founder’s Dilemma in Harvard Business Review and his book of the same name, is that people problems are the leading cause of startup failure. Not the market, not the product. The people, and the arrangements between them that were never discussed.

The research on equity shows how little discussion there usually is. In a study of 1,476 founders across 511 ventures, Thomas Hellmann and Wasserman found that roughly a third of founding teams split equity equally, that many settled the split in less than a day, and that teams which split equally after a very quick negotiation went on to raise their first round at lower valuations. Their reading is that the quick equal split is a way of avoiding a difficult conversation, and that the avoided conversation shows up later in the price an investor is willing to pay.

Wasserman’s own study of 212 American startups surfaces the deeper question. Founders want two things, wealth and control, and the two usually conflict: founders who gave up more equity to attract co-founders, hires and investors built more valuable companies, and most did not remain chief executive to the end. He calls it rich or king. Two co-founders who have not asked each other which they want are building two different companies and will discover it at the first term sheet.

The equity split takes a day because the conversation it should have followed never happened. The questions are the conversation.

Where co-founders come from

Paul Graham put a single founder first in his list of the eighteen mistakes that kill startups, on the grounds that starting a company is too hard for one person and that the low points are too low to bear alone. The pressure to find someone is therefore real, and it is the reason founders settle for the first willing person. Resist that. A bad co-founder is worse than none, because none can be fixed later and a bad one owns half the company.

The best candidate is someone you have already worked with under pressure: a colleague from a previous job, a classmate on a project that went wrong, a person you built something with over a long weekend. That history is the only data that predicts how someone behaves in month fourteen of a company that is not yet working. Everything else in this lesson is a way of manufacturing a little of that data when you do not have it.

When the network does not produce anyone, the structured route exists and is free. Y Combinator’s Co-Founder Matching platform, run through Startup School, reports over a hundred thousand matches made and lists around 1,900 active profiles in Bengaluru, the fourth largest city on its board after San Francisco, New York and London. A profile states what you are working on, what you bring and what you are looking for; the platform proposes matches and either side can decline. Treat it as a source of first coffees, not of co-founders. The coffees are where the process below begins.

How to run the conversation

Gloria Lin, who was the first product manager at Stripe and later head of product at Flipboard, spent about a year talking to six potential co-founders before committing to one, and wrote the method up for First Round Review as The Founder Dating Playbook. Her steps are the ones to copy. Early conversations to find overlap in experience and interests, looking specifically for misalignment rather than agreement. A two-week prototype of an idea together, to test the idea and the working relationship at the same time. A written questionnaire, which each person completes separately, followed by several working sessions comparing answers. Then a deliberate decision to commit or part, rather than a drift into partnership.

The forty questions below are organised for that method. Each person writes answers alone, in a document the other does not see until both are done. Then four sittings of about ninety minutes, one group per sitting with the fifth folded into whichever runs short, over three or four weeks. Writing first matters because it stops the second person agreeing with the first. The gap between the two documents is the only output that matters; where the answers match, move on in a minute.

What the answers tell you

Most pairs will disagree on a dozen of the forty and that is normal. Disagreement on how to be told you are wrong, on titles, on where to work from, is a thing two adults can negotiate. Three disagreements are different in kind, and each should end the process unless it can be genuinely resolved rather than papered over.

Rich or king. If one of you wants to raise institutional money and build something large at the cost of control, and the other wants to own and run a profitable company for twenty years, there is no compromise that holds. Wasserman’s finding is that the trade-off is real and that founders who refuse to choose end up with neither. Questions twelve, thirteen and fifteen ask it three ways because people answer the abstract version with what they think they should say and the concrete version with what they mean.

Runway mismatch. If one founder can live without salary for eighteen months and the other for four, the company has four months before it must pay someone, which means it must raise or earn within four months, which means the plan is the shorter founder’s plan whether or not anyone said so. This is resolvable: the longer-runway founder can lend, the company can pay one salary and not the other with the difference adjusted in equity, or the start date can move. But it has to be resolved in writing, with numbers, and before the longer-runway founder quits their job. The [runway lesson](/library/runway-how-many-months-you-really-have) has the arithmetic.

Hesitation over vesting. Question thirty-four is the most useful of the forty. A candidate who resists a four-year schedule with a one-year cliff on their own shares is telling you one of two things: they do not understand what vesting is, which is fixable with the [vesting lesson](/library/vesting-and-the-cofounder-cliff), or they want the option of leaving early with their equity intact, which is not fixable at all. Michael Seibel’s note on splitting equity makes the same point from the other direction: a founder unwilling to treat a co-founder as an equal has probably chosen the wrong one.

The trial project

No set of questions substitutes for working together, so run Lin’s two-week prototype before the final decision. Pick something small and real: a landing page with a waitlist, ten customer interviews run and written up, a working demo of the hardest technical piece. Set a scope on day one and a date on day fourteen. The point is not the output. It is to watch how the other person behaves when the scope slips on day nine, how they respond to a customer saying no, whether they do what they said they would do by the date they said it, and whether you want to do it again on day fifteen.

Pay attention to one thing above the rest: did you argue, and how did it end? Two people who do not disagree in two weeks of building are either not building anything hard or not being honest. Two people who disagree and resolve it without anyone keeping score have found the one quality that cannot be interviewed for.

The month, as a calendar

Week one: both write answers to all forty, separately. Week one, end: sitting one, how you work and money. Week two: sitting two, roles and risk, and the trial project begins. Week three: sitting three, exit, while the project runs. Week four: the project ends; sitting four is a review of the gaps left open from the first three and of the two weeks just spent. Then a decision, on a named day, stated plainly to each other: yes, or no, or not yet with a date.

If the answer is yes, the next two lessons follow in order and should be finished before incorporation: [how to split the equity](/library/splitting-founder-equity-equal-or-unequal), then [the co-founders’ agreement](/library/cofounders-agreement-what-it-must-contain) that records the split, the vesting, the roles and the exit terms you have just spent a month discussing. The forty answers are the brief for that agreement. Keep the documents; a founder re-reading them in year three will learn something about both of you.

Nothing here is legal advice. The questions are a method for finding out what two people actually think before a lawyer writes down what they agreed.