Ask ten owners how much cash a business should keep and you get ten versions of three months of expenses, delivered with varying confidence and no arithmetic behind it. It is a fine slogan and a poor operating target. Three months of expenses is enormous for a business with light fixed costs and dangerously thin for one carrying payroll for forty people and a floorplan line.
A better approach is to stop looking for one number and compute four floors, each answering a different failure question. What breaks first if collections stop? What breaks if the lender's draft hits on a slow week? What does a normal bad month actually cost? And what is the largest single unplanned demand this business could plausibly face? Your operating minimum is the largest of the four, and the exercise takes about an hour. It also happens to produce exactly the reserve logic a credit desk finds credible when it asks why your balance is what it is.
Floor one: the payroll floor
Payroll is the obligation with no grace period and no workout conversation. It is the first thing that has to clear and the last thing you can defer, so it sets the hard bottom of the account.
Compute it as two full cycles of gross payroll including employer taxes, workers compensation accrual and any benefit remittance that funds on the same rhythm. Two cycles rather than one, because the failure mode is not a single missed payroll: it is a receivable that slips past one cycle and lands in the next.
If you use a payroll service that drafts the full amount two banking days ahead, use the draft date rather than the pay date. The money has to be there when the draft posts, not when employees see it.
Floor two: the debt service floor
This one is straightforward: three months of total scheduled principal and interest across every obligation, including any daily or weekly remittance annualized down to a monthly equivalent.
Three months is a reasonable working figure because it covers the realistic recovery window from a single bad event: a lost anchor customer, a delayed insurance settlement, a season that opened late. It is not a regulatory standard and it is not a covenant. It is the amount that buys you time to fix a problem instead of negotiating from a position of having already missed a payment.
Floor three: the volatility floor
Every business has a bad month, and most owners know roughly what one looks like without ever having quantified it. The volatility floor makes it explicit: take the gap between your average monthly cash inflow and your worst month's inflow over the last twenty-four months, and hold that gap.
Twenty-four months rather than twelve, because a single year may not contain a genuinely bad month, and a floor built off a good year is a floor built for a business that no longer exists the moment conditions change. If you have a seasonal cycle, use the trough month from each of the last two years and take the worse one.
Use inflow, not profit
The point of this floor is cash timing, not accounting performance. A month where you booked $300,000 of revenue and collected $90,000 is a bad cash month regardless of what the P&L says. Build the series from deposits.
Floor four: the shock floor
The first three floors handle rhythm. The fourth handles the discrete event: the largest single unplanned demand the business could plausibly face inside a quarter. It is the one owners skip and the one that most often causes the emergency.
Compute it as the largest of a short list rather than a sum, since these rarely arrive together: your insurance deductible on the most likely claim, the replacement cost of the single piece of equipment the operation cannot run without, one quarter's estimated tax payment if you do not fund taxes separately, or the receivable from your largest customer if that customer paid late or not at all.
Putting the four together
Your operating floor is the largest of the four, not the total. The floors overlap in practice; holding enough to cover the worst one generally covers the others, and stacking them produces a target so high that no operator will actually hold it, which means the exercise gets abandoned.
| Floor | How it was computed | Amount |
|---|---|---|
| Payroll floor | Two semi-monthly cycles of gross payroll and employer taxes at $61,000 each | $122,000 |
| Debt service floor | Three months of $18,400 in total principal and interest | $55,200 |
| Volatility floor | Average monthly deposits of $348,000 less worst month of $214,000 | $134,000 |
| Shock floor | Largest single item: top customer receivable at $146,000 | $146,000 |
| Operating floor | The largest of the four, not the sum | $146,000 |
Note what the exercise revealed. This operator would have named payroll as the binding constraint, because payroll is the number that keeps people awake. The actual binding floor was customer concentration, which is also the risk the business had done the least to manage. That reordering is the usual outcome and the main reason the exercise is worth an hour.
Where the floor should physically sit
A reserve that lives in the operating account gets spent, not through recklessness but through the ordinary drift of a balance that never visibly falls below a line because no line was drawn. Separating it works better, with one caveat.
The caveat is that the operating account still has to look healthy on its own. A business that keeps its entire cushion in a savings account and runs the operating account to nine hundred dollars every cycle presents as fragile even though it is not. Practical compromise: keep the volatility and shock portions in a separate account, keep payroll and debt service coverage in the operating account where it visibly steadies the daily balance, and disclose both accounts in any credit file.
A separate tax-reserve account is worth adding regardless. It is the single most common source of a shock that was entirely predictable.
Getting from where you are to the floor
Most businesses computing these floors for the first time find they are below the largest one. That is a plan, not a crisis, and it is best worked as a funded target over two or three quarters rather than a single heroic month.
Building the floor without starving the business
- Compute all four floors and write the operating floor on one line where you will see it monthly.
- Open a separate reserve account if you do not have one, and a separate tax account if you do not have that.
- Set an automatic weekly transfer into reserve: small and constant beats large and occasional, because it survives bad weeks.
- Divert the cash freed by retiring any high-remittance obligation straight into reserve rather than into operations.
- Cap discretionary distributions until the floor is funded, and write the cap down so it is a policy rather than a mood.
- Convert one concentration risk into a smaller one: deposits from the largest customer, shorter terms, or a second account of scale.
- Reprice or clear inventory that has been sitting longer than one normal cycle and route the proceeds to reserve.
- Recompute the floors annually, or immediately after any material change in payroll, debt or customer mix.
Is three months of expenses actually wrong?
It is not wrong so much as unmeasured. For some businesses it lands close to the largest of the four floors; for others it is off by a factor of three in either direction. The four-floor method takes about the same time to compute and gives you a number you can defend to yourself, your partner and an analyst.
Does a line of credit substitute for a cash reserve?
Partially, and less than operators assume. An undrawn line is genuine liquidity, but it can be reduced or reviewed, drawing on it adds debt service in the coverage calculation, and high utilization reads as strain. Treat a line as a supplement to the floors, not a replacement for them.
How does holding a reserve interact with average daily balance?
Directly and favorably. The reserve is the main structural reason a daily balance stops swinging to near zero between deposits. That is why the floors and the balance conversation are really the same conversation viewed from two angles.
Should the reserve be in a money-market or investment account?
Interest-bearing is sensible provided the funds are unrestricted and same-day or next-day accessible. What defeats the purpose is anything with a lockup, a penalty or settlement delay, a reserve you cannot reach on the day payroll drafts is not a reserve.
My business is seasonal. Do the floors change through the year?
The floors themselves are stable; what changes is how far above them you should sit. Seasonal operators typically build well above the floor during the strong months specifically so the trough draws down toward it rather than through it. The trough is the test the floors were designed for.
What if funding the floor means I cannot take distributions?
Then the honest reading is that the business does not currently generate enough cash to both support the owner and carry its own risk, which is useful information. The usual resolution is a partial distribution alongside a funded weekly transfer, so the floor builds over quarters instead of the plan collapsing in month one.
Do lenders ask about reserves directly?
Sometimes explicitly, more often implicitly through the statements, the balance either shows a cushion or it does not. Being able to state your floor and the logic behind it is a strong signal in a conversation, because very few operators can, and it varies by lender how much weight that carries.
How often should I recompute?
Annually is enough in a stable business, plus immediately after any material change: a hire that moves payroll meaningfully, new debt, the loss or addition of a large customer. The inputs are all numbers you already track, so the recompute is a short exercise once the template exists.
Where to start
Compute floor four first. It takes ten minutes, it is the one nobody has done, and it is the floor most likely to turn out binding. Then do the other three and write the largest number where you will see it at every month-end.
If you want the floors computed against your actual deposits, payroll and obligation schedule, and tracked month over month alongside coverage and average daily balance, Capital OS builds them from the same documents a credit desk would read.
Qualified Commercial Underwriting Desk
Credit and capital markets
The Qualified Commercial underwriting desk reviews commercial real estate, dealer and Main Street files daily. The Academy is written from that work (how files are actually read, priced and declined) rather than from a rate sheet.
Educational content only. Nothing here is a commitment to lend, an offer of credit, or tax, legal or accounting advice. Program terms, timelines and thresholds vary by lender, file and market conditions, and any figures shown are illustrative.