Build a 13-Week Cash Flow Forecast (The Only Forecast That Prevents Surprises)

A business can be profitable on paper and still fail to make payroll. The two facts aren't in conflict — they measure different things on different schedules.

Profit says: over this period, revenue exceeded expenses. Cash says: on this specific Friday, is there money in the account?

The 13-week rolling forecast is where those two get reconciled.

Why profit and cash diverge

Five common gaps, all timing:

• Receivables. You invoice in March and collect in May. The P&L books March; the bank sees May.
• Inventory. Cash leaves when you buy stock, not when you sell it.
• Taxes. Estimated payments hit on a quarterly schedule unrelated to your revenue curve.
• Loan principal. Interest is an expense; principal isn't — but both leave the account.
• Capital purchases. Depreciated over years on the P&L, paid in full today.

Each of these can be perfectly healthy and still create a week where the balance goes negative.

What you need before you start

• Your current bank balance — the real one, today
• Open invoices with realistic collection dates
• Recurring fixed costs and their payment dates
• Payroll dates and amounts
• Loan payment schedule
• Tax payment dates
• Any known one-off items in the next quarter

The structure

Columns are weeks. Rows are the movements. Build it exactly in this order:

Closing balance carries forward as next week's opening. That single link is what makes the model useful — a shortfall in week 9 becomes visible in week 2.

The rules that make it accurate

Forecast collections, not invoices. If a client reliably pays at 45 days, model 45 days. Optimism here is the most common reason these forecasts fail.

Be precise on the outflows you control. Payroll, rent, loan payments, and subscriptions are known amounts on known dates. There's no excuse for imprecision.

Include the items that aren't on the P&L. Loan principal, owner draws, tax payments, equipment purchases. These are exactly the ones people forget, and they're large.

Use conservative timing on both sides. Model inflows a week later than expected and outflows a week earlier. If it survives that, it will survive reality.

Update it weekly

This is the part that determines whether the tool works.

Every week — same day, same time — do three things:

1. Enter last week's actuals next to the forecast.
2. Calculate the variance. Where were you wrong, and by how much?
3. Roll the window forward by one week, so you always see 13 weeks ahead.

The variance column is the real output. After a month or two it tells you exactly how your business behaves: which clients pay late, how much your "variable" costs actually vary, whether your collection assumptions are fiction.

When the forecast shows a gap

The point of seeing it eight weeks out is that you have options that don't exist at week one:

Accelerate inflows — invoice immediately rather than month-end, offer a small early-pay discount, ask for deposits on new work, chase aged receivables deliberately.

Delay outflows — negotiate longer supplier terms, reschedule a non-urgent purchase, time a large payment for after a known collection.

Add a buffer — draw on a line of credit before you need it. Credit is easiest to arrange when you don't yet require it, which is precisely why the eight-week warning matters.

Reduce structurally — if the gap is recurring rather than a timing blip, the problem isn't cash flow. It's the cost base or the pricing, and the forecast has just told you so.

A gap that appears in every 13-week window is not a cash flow problem. It's a business model problem wearing a cash flow costume.

The common mistakes

1. Building it from the P&L. Accrual revenue is not cash. Start from the bank.
2. Forecasting once. A stale forecast produces false confidence, which is worse than no forecast.
3. Optimistic collection dates. Model how clients actually pay, not how the contract says they should.
4. Omitting owner draws and taxes. Both are real, large, and frequently forgotten.
5. Never comparing to actuals. Without the variance check, you never find out your assumptions are wrong.

What to do next

Build the first version this week in a spreadsheet, using your real bank balance and your actual open invoices. It will take two to three hours the first time and about fifteen minutes a week after that — which is a small price for never being surprised by a Friday again.