Safe to pay yourself: the number nobody gives sole proprietors

An employee gets a paycheck that is already net of tax. A sole proprietor gets a balance, and has to work out for themselves how much of it is really theirs. Here is how we compute that figure, what goes into it, and why the obvious shortcuts are wrong.

ExactSplit · Updated August 5, 2026

The question between the deadlines

Most of the year, a sole proprietor is not thinking about a tax deadline. They are looking at a checking account with a number in it and trying to decide whether they can move some of it to personal. Pay the rent from it. Book the trip. Buy the thing. The tax software does not help with this, because the tax software is about April and June and September and January. It is silent for the eleven months in between, and the eleven months are where the money actually gets spent.

The answer people arrive at on their own is usually one of two bad ones. Either they treat the whole balance as spendable and get surprised at the next deadline, or they treat it as untouchable and live off far less than they could, while cash that was always theirs sits in a business account earning nothing. Both are the same mistake in different directions: they are guessing at what part of the balance is owed to somebody else.

We built a figure for this. It is called safe to pay yourself, it sits between the headline quarterly number and the income and expense totals on the dashboard, and it answers one question: how much can leave the operating account today without creating a problem later.

This post is about how it is computed, because a number like this is only useful if you know what it assumes.

What goes into the figure

The figure starts with the operating balance: the available balance across every checking account you have marked as business or mixed, excluding the reserve account. Personal accounts are not included, because the money in them is not the business's to draw on and the question is about the business. The reserve account is excluded because the money in it is already spoken for.

From that, four things are taken away.

First, pending charges. A card payment that has been authorised but not yet posted is money that has already left in every sense that matters. Plaid reports these separately, and we subtract them rather than wait for them to post and surprise you.

Second, tax owed so far that is not sitting in the reserve. This is the year's tax liability accrued to date, less withholding to date if you also have a W-2, less any estimated payments you have recorded. If you run a reserve account and it is on track, this line is small or zero, because the tax is already set aside. If you do not run one, or it is behind, the unfunded part is subtracted here. Either way the operating balance stops looking like it includes money that is owed to the IRS.

Third, recurring bills due inside the horizon. More on this below, because it is the part that takes real work.

Fourth, a cash buffer. Also more below.

Why a percentage of the balance does not work

The tempting shortcut is to say something like: keep 30 percent for tax, keep another 10 percent for bills, spend the rest. It is the same instinct as the 30 percent rule for set-aside, and it fails for the same reason. Your tax is not a percentage of your balance. It is a function of your income, your expenses, your filing status, your state, and what you have already paid. Two people with the same balance can owe wildly different amounts, and the same person can owe very different amounts in March and in November.

Bills are worse. A percentage of the balance tells you nothing about whether your insurance premium is due Tuesday. What you need is a list of what is about to leave the account, with amounts and dates, and that list has to come from somewhere.

It comes from your own bank history.

Bills you have not paid yet

A recurring bill, for our purposes, is a merchant that has charged you a stable amount at a steady cadence at least three times. Three, not two, because two charges a month apart are a coincidence often enough that counting them would invent bills that do not exist. The cadence has to be recognisable: weekly, every two weeks, monthly, quarterly, or yearly, and every gap in the run has to look like that cadence, not just the average. A merchant you paid in January, February and then August is not a monthly bill with a long gap. It is a merchant you shop at.

The amount has to be stable too. Every charge within a quarter of the median, or the run is rejected. Your gym is a bill. Your grocery store is not, even though you go every week, because the amount is different every time and treating it as a fixed bill would be wrong in both directions.

Once a bill is recognised, the next date is projected. For monthly, quarterly and yearly bills that is done by calendar month, not by adding the average gap in days. A bill that hits on the 3rd should be projected to the 3rd, and adding 31 days to September 3rd gives October 4th, which drifts a day every month until the projection is useless. A bill that is a few days late is still counted as due now. A bill that is more than a whole cycle overdue is presumed cancelled and dropped, because the alternative is holding money back forever for a subscription you ended in March.

Only the bills whose next date falls inside the horizon are subtracted. The horizon defaults to 30 days and is yours to change. The list is visible on the dashboard, so if the detector has picked up something you have already cancelled, you can see it and it is one click to adjust.

The buffer, and why you choose it

The buffer is the one input that is genuinely a preference rather than a fact, so it is a setting rather than a computation. By default it is fourteen days of your typical spend, where typical is the average daily outflow over the last ninety days across the operating accounts. Fourteen days is enough to absorb a client paying late without the figure swinging negative. If you would rather hold a fixed amount, you can set one in dollars and it overrides the days.

There is no right answer here. Someone with one large client who pays on the first of the month wants a bigger buffer than someone with forty small clients paying continuously. The point of making it a setting is that the figure reflects your risk tolerance rather than ours.

The buffer is the last thing subtracted, and the result is floored at zero. A negative safe-to-pay figure does not mean you owe money to yourself. It means you should not be moving anything out right now.

How to read the number

The figure comes with two pieces of context. The first is its age. Every input is a balance Plaid reported at the last sync, so the panel says when that was, and after a day and a half it suggests a sync before you act on it. A figure computed from yesterday's balance is probably fine. One computed from last week's is a guess dressed up as a number.

The second is the breakdown. Every line is shown: operating balance, pending, tax unfunded, bills due with the list behind it, the buffer, and the result. Nothing is folded into a single opaque figure, because the whole point is that you can see why the number is what it is and disagree with a line if you know something the bank feed does not.

A typical reading looks like this. Operating balance twelve thousand. Pending charges one hundred and fifty. Tax owed so far and not in the reserve, three thousand, because the last two deposits have not been split yet. Bills due in the next thirty days, nine hundred: the insurance, the software, the phone. Buffer, seven hundred. Safe to pay yourself, seven thousand two hundred and fifty. If that seems low against a twelve thousand balance, the breakdown tells you exactly where the other four thousand seven hundred and fifty went, and three thousand of it goes away the moment you fund the reserve.

What it is not

It is not an owner's draw calculation for tax purposes. A sole proprietor's draw has no tax consequence on its own; you are taxed on profit whether you take it out or leave it in. This figure is about cash, not about what to report.

It is not a forecast of income. It does not assume any deposit that has not landed. If you know a large payment is coming Friday, the figure will not reflect it until Friday, on purpose. Counting money you have not received is how people end up short.

And it is not advice about what you should do with the money once it is yours. It tells you the amount that can leave without breaking anything. Whether you move it, leave it, or put it toward a retirement contribution that the optimiser says would save you tax is a separate question, and the dashboard has a separate section for it.