Start from your reconciled bank balance, lay out thirteen weekly columns, then place every expected receipt and payment in the week it will actually move. Receivables are phased by each customer's payment history, not their terms. Payroll goes in on its real dates. Each week's closing balance becomes the next week's opening. Then mark every week that drops below your cash floor, because those are the weeks you still have time to do something about. Rebuild it weekly, replacing forecast with actual, and note where the last one was wrong so the next is better. Expect two to four hours to build the first and about thirty minutes a week to keep it current. The early weeks will be close and the later ones directional, which is exactly as it should be.
Profitable businesses fail on cash timing. Not on profitability — on the gap between when money is earned and when it arrives. A contractor can have a record quarter on paper and still be unable to make payroll in week seven, because two large invoices landed on net 45 and the material bills did not.
The 13-week forecast exists to make that visible while it is still fixable.
Why thirteen weeks
Thirteen weeks is one quarter, and the length is a deliberate compromise between two failures.
Too short — four weeks — and everything is already committed. You can see the problem but not act on it, because the invoices are sent, the bills are due, and the levers are gone.
Too long — twelve months, weekly — and the arithmetic outruns the evidence. Week 40 is not a forecast, it is a guess with a decimal point, and forecasts that contain guesses stop being trusted for the parts that were real.
At thirteen weeks the early columns are built from invoices and bills that already exist, and the later ones still leave time to collect faster, delay a purchase, or arrange finance before the week arrives.
Monthly cash forecasts hide exactly the problem you are looking for. A month can finish comfortably positive while containing a Thursday where payroll did not clear. Cash crises happen on days, not months, and a weekly grid is the coarsest view that still shows them.
What you need before you start
- A reconciled bank balance. Not the book balance — the reconciled one. Starting from an unreconciled figure means every one of the next thirteen weeks is wrong by the same amount. Reconcile first, per the close checklist.
- The AR aging — Reports → Who owes you → Accounts receivable aging detail, at invoice level.
- The AP aging — Reports → What you owe → Accounts payable aging detail.
- Your payroll calendar with exact run dates and typical gross plus employer taxes.
- Twelve months of payment history by customer, which is what makes the phasing honest.
- Your fixed monthly outflows — rent, loans, insurance, subscriptions, tax deposits.
Building it, row by row
Columns are week-ending dates, thirteen of them. Rows are as follows.
Opening balance
Week 1 opens with the reconciled bank balance across all operating accounts. If you keep a reserve account you genuinely will not touch, leave it out — a forecast including money you have decided not to use overstates your position.
Receipts — existing receivables
Every open invoice from the AR aging, placed in the week you expect payment. Not the due date. The expected date, which is a different thing and the subject of the next section.
Receipts — new sales
Work you will invoice during the period, phased conservatively. Keep this on its own line. When the forecast is wrong, this is usually the row responsible, and you want to be able to see that rather than have it blended into the receivables line.
Receipts — other
Loan draws, tax refunds, asset sales, owner contributions. Anything real but not routine.
Payments — payroll
Every pay run on its actual date, gross plus employer taxes. If tax deposits leave on a different date from wages, split them into two rows — for many businesses the deposit is large enough to matter on its own.
Payroll is the one row to get exactly right, because it is the one obligation that cannot slip. Whether the wage run clears is covered in detail in payroll cash planning.
Payments — payables
Bills from the AP aging, in the week you intend to pay them. Intend, not due — if you habitually pay a vendor two weeks late and they tolerate it, forecast reality.
Payments — fixed and periodic
Rent, loan payments, insurance, software, estimated tax deposits. The quarterly and annual ones are the dangerous entries, because they are easy to forget and large enough to convert a comfortable week into a crisis.
Closing balance
Opening plus receipts minus payments. Each week's closing becomes the next week's opening. That chaining is what turns thirteen independent estimates into a forecast.
Phasing receipts honestly
This is where forecasts are made useful or useless, and it takes about ten minutes.
For each significant customer, look at the last twelve months: invoice date against payment date. You will get an average lag, and it will almost never equal their stated terms.
A customer on net 30 who reliably pays in 45 days should be forecast at 45. Forecasting them at 30 builds two weeks of cash you will not have into the plan — and does it silently, in the row you are least likely to question.
Behavior is what funds payroll. If the two disagree, forecast the behavior and take up the agreement separately as a collections conversation. Do not resolve a collections problem by pretending it into a forecast.
Handling the doubtful ones
An invoice at 90 days past due from a customer who has stopped replying is not a receipt. Either place it well beyond week 13 or leave it out and note it. Carrying doubtful debt as forecast cash is the most common reason a forecast that looked fine is followed by a scramble.
Concentration
If one customer is more than about 20% of a week's receipts, note it. That week is not a forecast so much as a bet on one payment, and it is worth knowing which weeks are exposed that way before they arrive.
Setting a cash floor
A forecast without a floor is just a table. The floor is what converts it into a decision.
Your floor is the minimum balance below which you are not willing to operate — enough to absorb a late payment or a failed truck without improvising. Many small businesses use two to four weeks of operating costs. The specific number matters far less than having chosen it in advance, because a floor decided during a crunch is not a decision, it is a rationalization.
Mark every week that closes below it. Those are the only rows that require action, and having them identified in week 2 rather than week 9 is the entire return on the exercise.
Rolling it forward
Every week, three things. Replace week 1's forecast with what actually happened. Add a new week 13 at the far end. And — the step everyone skips — note where the forecast was wrong and why.
That last step is what makes forecast four better than forecast one. Over a couple of months you learn that a particular customer always pays a week later than you think, that material costs cluster before large jobs, that the tax deposit is bigger than you remember. Those corrections compound.
A forecast built once and never compared to reality is worse than none, because it carries the same authority while being unexamined.
Running the two scenarios that matter
A single-line forecast implies a confidence you do not have. Once the base version works, build two variants — it takes about twenty minutes and changes what the forecast is for.
The downside case
Push every receivable out by two weeks and remove the least certain 25% of new sales. Leave every payment exactly where it is, because costs do not politely defer when revenue does.
This is the version worth acting on. If the downside case never breaches your floor, you can stop worrying about cash and go run the business. If it breaches in week 6, you have found the thing to work on and roughly a month to work on it.
The stress case
Remove your largest customer's payments entirely for the quarter. Not because you expect it, but because concentration risk is invisible until it is measured. If losing one customer's payments for a quarter would end the business, that is a strategic fact you want to know on an ordinary Tuesday rather than during the event.
Most owners discover their real exposure is not the number of customers but the timing — one customer paying two weeks late does more damage than a smaller one not paying at all, because the large one is load-bearing for a specific week's payroll.
Not prediction. They exist to convert a vague unease about cash into a specific week and a specific amount, because a specific problem has options — collect early, delay a purchase, draw on a facility, defer an owner distribution — and a vague one only has worry.
Where forecasts go wrong
Starting from an unreconciled balance
Every week inherits the error. Reconcile first.
Forecasting terms instead of behavior
The single largest source of optimism, covered above.
Forgetting the quarterly and annual items
Insurance renewal, tax deposits, annual software. Large, infrequent, and easy to omit precisely because they are not part of the monthly rhythm. Walk last year's bank statements once and list everything that only happened two or four times.
Netting things together
A single row of "expenses" hides which lever to pull. If a week is short, you need to know whether it is payroll, materials or a loan payment, because you can move one of those and not the others.
Building it and abandoning it
The most common failure by a distance. The first build is interesting and the twentieth Monday is not. A forecast is only worth anything if it is current.
How long this takes
The first build is genuinely worth doing yourself. You learn where your cash goes in a way that no report communicates, and that understanding is not transferable.
The maintenance is the problem. Thirty minutes every Monday for a year is roughly a full working week, spent on a task with no external deadline forcing it — which is why most of these are abandoned by month three, usually in a busy period, which is exactly when the forecast was about to become useful.
When it stops being worth doing by hand
The build is a skill worth having. The weekly rebuild is data assembly — pulling the AR aging, re-phasing by history, updating payroll dates, re-chaining the balances. It is the same motions every week and none of it needs your judgment, only your time.
The signal is when you notice the forecast is two weeks stale at the moment you actually need it.
Have the forecast arrive instead of building it
The Weekly Cash Flow Snapshot lands Monday by 9 AM: 30/60/90 days out, phased on your customers' real payment behavior, with the tight weeks flagged before they arrive. It is part of the Books plan at $897 a month, or $297 on its own. Start with one closed month for $497 and see the underlying numbers first.
Get Your First Close — $497 See the plansCommon questions
Why 13 weeks specifically?
One quarter. Long enough to still act on what you see, short enough that the assumptions rest on invoices and bills that already exist.
Isn't this what the cash flow statement does?
No. That is historical and monthly. This is forward-looking and weekly — and a healthy month can contain a week where payroll does not clear.
How accurate should I expect it to be?
Weeks 1–3 close. Weeks 4–8 drift. Weeks 9–13 directional. A forecast that is precise about week 12 is lying to you.
Should I use my payment terms or actual payment history?
History, always. A net-30 customer who pays in 45 days should be forecast at 45, or you have invented two weeks of cash.
What is a cash floor?
The minimum you will operate above — often two to four weeks of costs. Choosing it in advance is what matters.
Spreadsheet or software?
Spreadsheet is fine, and building the first by hand teaches you things. Maintaining it every week for a year is where they die.