How to Find a Co-Founder: A Sourcing Playbook for 2026

I have spent fifteen years running searches for companies that had a budget, a job title and a hiring manager. Founders looking for a co-founder have none of those. There is no salary to anchor the conversation, no job description to argue over, and no manager to make the final call. What is left is a search problem, and search problems respond well to method.

This article is the method I would use if I were running your co-founder search as a mandate. It covers the brief, the channels, the funnel arithmetic, the outreach and the trial period. Where I use numbers, I state the assumption behind them so you can substitute your own.

What are you actually searching for?

A co-founder is not a senior hire who accepts equity instead of cash. The distinction matters because it changes what you screen for. A hire executes decisions inside a scope you define. A co-founder owns a scope you will not be able to supervise, at a stage when the company has no processes to fall back on.

So the first question is not who but what: which two or three decisions do you want to stop being the final word on? Typical answers are product architecture, the technical hiring bar, the go-to-market motion, or the pricing and packaging model. Write them down. That list is the real specification, and it is far more useful than a title.

Write a brief that can actually be assessed

Almost every co-founder search I see starts from a wish list: technical, ambitious, based in the same city, has shipped a product, understands our market, wants to build something big. Every item is defensible. Together they describe nobody, because none of them can be checked against a public profile and none of them is ranked.

Split your requirements into three columns instead. The exercise takes about forty minutes and it is the highest-leverage thing you will do in the whole search.

Deciding criteria are things like: has personally built and operated a production system at the scale we will hit in eighteen months; has sold into hospital procurement; has run a team through a funding round. Shaping criteria are things like: worked at a company we admire; based within two time zones; has founded before. Disqualifying criteria are things like: cannot commit full time within ninety days; needs a market-rate salary from month one; is subject to a non-compete that covers our category.

A worked example

Take a non-technical founder building a compliance product for mid-market European banks. The wish list said senior engineer, fintech experience, English speaking, entrepreneurial. Reframed as a brief:

• Deciding: has shipped software that passed a bank or insurer security review; has been the most senior engineer in a team of five or more; has worked under EU data-residency constraints.
• Shaping: has been a founder or an early employee at a company under thirty people; based in the EU for customer meetings; has a public body of work I can read.
• Disqualifying: requires more than 40% of market salary in cash in year one; cannot start within ninety days; is currently a co-founder elsewhere.

The wish-list version matched tens of thousands of people and gave no way to rank them. The brief version is narrow enough to search, and every Deciding item can be checked before a first call. That is the whole point: a good brief converts an opinion into a query.

Where to find a co-founder: nine channels compared

There is no single best place to find a co-founder. There are channels with different costs, different speeds and, critically, different levels of control over who enters your funnel. The table below is my working estimate from running searches; treat the time column as an order of magnitude rather than a promise.

Notice the pattern. The channels that feel easiest, such as posting publicly, joining a matching platform or going to events, are the ones where you have the least control over who shows up. They are worth running in parallel because they are cheap, but they should not be your primary channel. The primary channel should be the one where you decide the names.

How many people do you need to talk to?

Founders consistently underestimate the top of the funnel, then conclude after eleven conversations that there is nobody out there. Here is the arithmetic I plan with. The conversion rates are my own estimates from executive search adapted to founder-level outreach, and they assume personalised messages written by the founder rather than templated sequences.

Two consequences fall out of this table. First, fifty is the number that matters, not five, and fifty qualified names is a research task rather than a networking task. Second, a 40% reply rate is achievable at this volume only because the messages are individually written. Sending three hundred templated notes does not scale this funnel; it collapses the reply rate and burns the names.

If your reply rate comes in under 20%, the problem is almost always the message or the targeting, not the market. If fewer than one in five calls leads to a second conversation, the problem is the brief: you are talking to people who were never a fit.

A six-week co-founder search plan

1. 1

   Week 1: write the brief and the story

   Produce the three-column brief. Then write the 200-word version of why this company should exist and why now, in language a domain expert would respect. You will reuse that paragraph in every first message, so it is worth a day.

2. 2

   Week 2: build the long list

   Assemble 120-200 names from public profiles, open-source contributions, conference speaker lists, patent filings, and the second-degree connections of five people who know your market. Do not filter for likelihood yet. Filter only against the Disqualifying column.

3. 3

   Week 3: cut to a shortlist of 50 and rank it

   Score every name against the Deciding criteria with evidence attached: a link, a repository, a talk, a shipped product. If you cannot attach evidence, the person does not belong on the shortlist. Rank by strength of evidence, not by how impressive the logo is.

4. 4

   Weeks 3-5: contact in waves of fifteen

   Send fifteen individually written messages, wait five working days, then send the next wave. Waves let you fix a weak message after fifteen data points rather than after fifty. Expect around twenty replies across the shortlist.

5. 5

   Weeks 4-6: exploratory calls

   Thirty minutes, no pitch deck. Two thirds of the call should be them talking about a decision they made and later regretted. You are looking for how they reason under incomplete information, because that is the entire job.

6. 6

   Week 6: convert to paid trials

   Take the best three into a scoped, paid 60-90 day project. This is where the real evaluation happens. Everything before it was filtering.

Six weeks gets you to trials, not to a signed cap table. Budget another three months for the trial itself. A realistic end-to-end expectation is four to five months from a blank page to a co-founder you have actually worked with.

Sourcing deliberately instead of waiting to be introduced

The difference between a six-week search and a twelve-month one is whether you generate names or wait for them. Generating names means starting from an attribute and working towards people, rather than starting from people you already know and hoping one of them fits.

The useful attributes are the ones that leave public traces. Someone who has passed a bank security review has usually written or spoken about it. Someone who has scaled a team has usually appeared on an engineering blog or a conference programme. Someone who has worked under EU data-residency constraints has usually left a trail in their job history rather than in a skills list.

This is why keyword search alone under-performs for founder-level searches: the thing you care about is rarely a keyword on the profile. It is an inference from a career history. Practically, that means combining hard filters such as seniority, geography, industry and company size with a plain-English description of the experience that matters, then reading the evidence rather than trusting a score. The mechanics of building those queries are covered in Boolean search for founders.

What goes in the first message

Founder outreach fails for a predictable reason: it asks a stranger for a life-changing commitment in the first sentence. Nobody replies to that. The first message should ask for a conversation about a problem, and should demonstrate that you chose this person on purpose.

• One specific reason you contacted them. Not their title. The thing they built, wrote, shipped or decided.
• Two sentences on the problem, in the vocabulary of their domain rather than of your pitch deck.
• What you have already done. Evidence that this is real: customers spoken to, a prototype, a letter of intent, revenue.
• A small ask. Twenty-five minutes to test whether the problem is as bad as you think. Not a co-founder proposal.
• A visible exit. One line saying that if the timing is wrong, a pointer to someone else would be just as useful.

Do not mention equity in the first message. Equity is the answer to a question the person has not asked yet, and putting a number in early converts a peer conversation into a job offer. Full templates and the sequencing that follows are in Cold outreach to co-founders and advisors.

Test the partnership before you sign it

Interviews measure how someone presents. A trial measures how someone works. For a co-founder, where the cost of being wrong is a year of your life and a contested cap table, the trial is not optional.

1. 1Scope a real project, 60-90 days, paid. Paid matters: it signals that you value the time and it gives you a legitimate basis for expectations.
2. 2Choose work with a visible outcome, such as a shipped module, five customer discovery calls with written findings, or a pricing model with a defence of its assumptions.
3. 3Deliberately include one disagreement. Put them on a decision where you already hold a view, and watch how they argue for a different one.
4. 4Write down, before you start, what a good result looks like. Otherwise you will rationalise whatever happens.
5. 5Review it in writing at the end, in both directions. The way someone receives a critical review predicts the next three years accurately.

The detailed version of this, including the questions I ask and the reference calls worth making, is in Co-founder vetting.

Five mistakes that cost founders a year

• Searching for a friend, not a partner. Comfort and complementary judgement are different variables. Optimise for the second and accept the first if it comes.
• Recruiting for the stage you are in. A brilliant zero-to-one builder may be the wrong person for the twenty-person company you will be in two years. Ask what they want to be doing at that point, explicitly.
• Treating a 50/50 split as fairness. Equal splits are fine when contributions are genuinely equal. When they are not, an equal split starts a resentment clock. Decide it deliberately and write down why.
• Waiting for certainty. You will never be certain. You will only ever have evidence, and a trial produces more of it in ninety days than eighteen months of coffees.
• Running the search in secret. Some founders never source properly because they worry the market will read it as weakness. In practice a specific, confident search reads as competence.

What to do this week

1. 1Write the three-column brief. Forty minutes. Nothing else in this article works without it.
2. 2Build a list of twenty names from public evidence, with a link attached to each showing why they meet a Deciding criterion.
3. 3Write one first message to the strongest name and send it. The first message you send teaches you more than the next week of planning.