How Long to Keep the Job

Almost everyone building something while employed is running an overlap, and almost nobody plans it. The job pays, the thing being built does not yet, and the question of when one replaces the other gets deferred because it feels like it will answer itself.

It does not answer itself. It gets answered by whichever runs out first, which is usually patience.

The constraint is attention, not money

The standard framing is financial: save enough runway, then leave. That framing is comfortable because it turns an uncertain decision into an arithmetic one, and it is mostly wrong about what the overlap costs you.

What a demanding job takes is not primarily hours. It is the specific kind of attention that building requires, which is contiguous, unhurried, and willing to sit with an unresolved problem. A job that consumes nine hours and returns you intact is compatible with building. A job that consumes seven hours and leaves you depleted is not, and the second kind is more common at senior levels.

This is why people with apparently plenty of evenings make no progress for a year. The evenings exist. The capacity to use them does not.

What the overlap is genuinely good for

Two things, and it is worth being precise because the overlap is usually either over-extended or abandoned too early.

“When the constraint moves from money to access, the overlap has stopped being a subsidy and started being a tax.”

It is good for validation, which requires conversations rather than deep work. Talking to twelve people in your industry about a problem is entirely compatible with a full-time job, and it is the highest-value activity available to you at that stage.

It is also good for the first paying customer. Taking on one client, at a price that reflects your experience, while employed, is the single most informative thing you can do, because it tests the whole proposition without requiring you to bet anything on it.

What the overlap is bad for is the long build. Anything requiring sustained concentration over months tends not to survive contact with a full-time role, which is why so many half-finished products sit in repositories belonging to very capable people.

When the constraint moves from money to access, the overlap has stopped being a subsidy and started being a tax.

The signal to watch for

Not a revenue number. A specific feeling, which is that the job has started to cost you opportunities rather than fund them.

That shows up concretely. A prospective client wants a call at a time you cannot take. A piece of work would obviously lead somewhere and you cannot accept it. The thing you are building needs a fortnight of attention and there is no fortnight available in the next six months.

When the constraint moves from money to access, the overlap has stopped being a subsidy and started being a tax. That is the moment to take the decision seriously, and it usually arrives before the revenue would justify it on paper.

The honest arithmetic

Replacing a salary is the wrong target and it produces bad decisions, because it is a large number that makes the whole thing look impossible for a long time and then suddenly reckless.

The useful target is smaller: enough recurring revenue to cover the fixed costs you cannot avoid, plus enough demonstrated demand that you believe the next client is findable. That is a much lower bar than salary replacement and it is the actual decision point, because what you are buying with the leap is time, and time is what converts demonstrated demand into revenue.

Nobody can tell you when. But if you are waiting for the number to equal your salary, you will wait considerably longer than the situation requires, and you will probably lose the thing to attrition before you get there.

See the employee to operator to owner path for the wider sequence, the 30 day plan to replace a salary for the financial side, and the six month profitability rule for a firmer threshold. How the system works covers structuring the overlap.