How to split equity when there is no money yet

Almost every unfunded project starts the same way. Two people agree on “50/50, we’ll figure out the details later”, go off to their own tasks, and for six months everything is fine. The problem shows up the moment there is real money or a real buyer — and it turns out one of the two has been opening the repository once a week since spring.

What follows is an argument nobody can win. All either side has is a memory of what was promised.

Why “let’s just split it evenly” breaks first

A fixed split on day one is a bet on the future made when nothing about the future is known. You don’t know how much time each person will actually put in, who burns out in month three, who has to take a full-time job, and who ends up carrying the project alone.

People split evenly not because it is fair, but because on day one it is the only split that avoids an uncomfortable conversation. The conversation isn’t avoided — it’s postponed to the moment it becomes expensive.

The second failure is subtler: equity is usually granted for intent rather than for results. Someone gets 30% because they will handle marketing. If they never do, the 30% doesn’t go anywhere. It has already been granted.

Five models and what goes wrong with each

1. Fixed split up front

Agreed on day zero, written down, never revisited.

Works when the people involved have shipped together before and know each other’s pace. Otherwise it is a lottery: any imbalance in contribution turns into resentment, and in an early project resentment kills faster than the lack of money.

2. Fixed split with vesting

Same thing, except equity is earned over time — four years with a one-year cliff being the standard: leave before year one and you get nothing.

This is the venture-world default and it is more honest than the first option. Its weak spot is that vesting measures the calendar, not the work. Someone who spent a year nominally involved and did the minimum vests exactly as much as the person who carried the project.

3. Dynamic models (Slicing Pie and its variants)

Equity isn’t granted, it is recalculated. Every contribution — hours, cash, equipment, customers brought in — converts into units at rates agreed in advance. Each person’s share is their units over the total.

The fairest model on paper and the hardest to live with. It needs someone to keep honest records continuously. Without a tool doing the bookkeeping it collapses within a month: people stop logging, and you’re back to relying on memory.

4. Splitting by confirmed contribution

A middle ground between the second and the third. Equity is neither recalculated continuously nor granted upfront: work is broken into tasks, each task carries a weight, the delivered result is confirmed by someone else, and shares are reassembled periodically from the confirmed work.

The key difference from dynamic models is that it counts accepted results rather than hours. Hours are easy to inflate. An accepted result is not.

5. An option pool instead of equity

Equity stays with the people carrying the risk; everyone else gets the right to receive equity later — at first revenue, at incorporation, at a funding round.

Good for people helping in a bounded way: a landing page, the first sales, infrastructure setup. Poor as the primary model, because a right that materialises “someday” motivates weakly.

What to write down, even without a lawyer

Formal paperwork is usually premature at the start — there is no legal entity, no revenue, and nothing to contract about. A written agreement, though, is always needed, and the minimum is short.

What counts as a contribution. Code, design, copy, a customer brought in, a paid server — list it explicitly. Half of all conflicts come from one person counting their ideas and calls as contribution while the other does not.

Who confirms a contribution. One person whose “accepted” settles the matter. Without it, every assessment is negotiable forever.

What happens when someone leaves. The most skipped clause and the most expensive one. Agree in advance whether the share stays whole, freezes at what was delivered, or lapses below some threshold.

What happens when money arrives. Not “we’ll split it”, but specifically: invested cash returned first, then distribution; or a percentage of revenue; or nothing until there is a legal entity.

When the agreement becomes legally binding. Name the event: incorporation, first contract, revenue above X. Until that event what you have is a reputational commitment, and it is healthier to call it that.

Equity versus a promise

Worth saying plainly, because most disappointment is built on this: until there is a legal entity and a signed document, a share is not a share — it is a promise. A chat message saying “your 20%” creates no rights whatsoever.

That doesn’t make early agreements pointless. It means their job is different: they record not a right but a fact — who did what. A fact can be proven later, once there is a legal framework, and it is the fact that the equity conversation will rest on. A promise with no recorded fact does not survive the first conflict.

So the sensible order is the reverse of the usual one. First, track contribution: tasks, results, confirmations, history. Then, once it is clear who is actually in the project, divide. And only then, once there is something to divide, make it legal.

The short version

Don’t assign equity on day one. Agree on how you will calculate it, start recording contribution from the very first task, and come back to the numbers in two or three months — when you have a history instead of a memory.

An equity conversation grounded in a list of confirmed results almost always goes calmly. One grounded in recollections of who promised what over the winter almost never does.