How to Build a 13-Week Cash Flow Forecast, Row by Row

Build one column per week (week ending Friday works well) and one row per cash movement. Keep the rows few enough that you'll actually maintain them:

• Opening cash: the real balance across every operating, savings and sweep account on Monday morning, not the ledger balance.
• Receipts: customer collections split by invoiced, subscription renewals and card or marketplace payouts, plus one line for non-operating items such as tax refunds.
• Payroll and related costs: net pay, employer payroll taxes and benefits as three separate rows, because they leave on different dates.
• Vendors and rent: software, hosting, contractors, rent and insurance, each with its own payment day.
• Debt and one-offs: loan payments, card statement payoffs, annual prepayments, quarterly estimated taxes.
• Net cash flow, closing cash and headroom: closing cash minus the minimum balance you've committed to holding.

Closing cash becomes next week's opening cash, so any formula error compounds. Lock those two rows down first.

Forecast the date customers actually pay, not the date the invoice says. Pull your receivables aging report and, for each customer, look at the last three or four invoices to see how many days past terms they typically pay. Then place each open invoice in the week that habit predicts.

For context, US small businesses waited 28.8 days on average to be paid in early 2026¹. If your invoices say net 30 but your best customer pays at day 45, forecast day 45.

For subscriptions, forecast from the billing calendar: monthly plans land when the card is charged plus the processor's payout delay, while annual plans arrive as a single lump in the renewal week. Leave new deals out of weeks 1 to 4 unless the contract is signed and the first invoice is already sent. In weeks 5 to 13, include pipeline only at a haircut you can defend, and label it so you can strip it out in one click.

Outflows are easier to forecast than inflows, but they cluster. Start with payroll because it's the largest fixed item and the hardest to delay. Put net pay in the week it's funded, which is usually a few days before the pay date, and put employer taxes on the deposit schedule your payroll provider or CPA gave you.

Then add the lumpy items people forget:

• Sales tax and payroll tax deposits that follow a monthly or quarterly rhythm.
• Corporate card payoffs, which land on the statement due date rather than when you spent.
• Annual software renewals, insurance premiums and audit or tax-prep retainers.
• Quarterly estimated income tax, if your entity pays it.

Say your card statement closes on the 20th with a due date of the 15th of next month. Spend from the last two weeks of this month shows up as a single payment in week 3 of the next cycle. If the forecast shows card spend as it happens, it will be wrong by weeks and, in a tight month, by a payroll.

The forecast only earns its keep if you refresh it on a fixed day. This routine works for most small teams:

1. Update opening cash to the real bank balances.
2. Replace last week's forecast column with actuals and note any line that missed by more than you can explain in one sentence.
3. Move any invoice that didn't get paid to the week you now expect it, and call the customer if it's more than a week late.
4. Add a new week 13 by copying the pattern of the same week last quarter, then correcting it for known events.
5. Check closing cash against your minimum in every week. Circle the first week that dips below it.
6. Write down the one or two decisions that follow (delay a vendor run, pull forward a collection, draw on a line of credit) and who owns each.

Keep last week's version. Comparing forecast against actuals across a few weeks shows whether you're consistently early or late on receipts, and that pattern is more useful than any single number.

Most failures are not spreadsheet errors, they're assumptions nobody re-tested. Watch for these:

• Mixing accrual and cash. Revenue recognized in June isn't cash until the customer pays, and a bill entered in June isn't cash until you pay it.
• Netting things together. If you show payroll as one line, you can't tell whether a miss came from wages, taxes or benefits.
• Forecasting to the penny. Round weekly lines to the nearest hundred or thousand so the file is fast to update.
• Setting the minimum balance as a round number instead of a decision. A common approach is to tie it to several weeks of payroll and fixed costs, then revisit it when those change.
• Never linking it to the longer view. Once the 13 weeks are stable, use them to check the assumptions in your cash runway calculation and the direct versus indirect choice covered in automated cash forecasting.

If your bank offers real-time balances and a transaction export, use them to fill actuals; typing balances by hand is where transposition errors creep in.