A 13-week cash flow forecast answers one question: will the company have enough cash, week by week, for the next quarter. It does not try to predict profit or tie to the income statement. It tracks money into and out of the bank, which is why lenders, boards and operators trust it more than any other forecast a company produces.
We build one with every client in the first ten days of an engagement. Here is the structure we use, line by line, and the weekly routine that keeps it honest.
Why the direct method, and why 13 weeks
The direct method forecasts actual cash receipts and payments. The indirect method starts from net income and adjusts for noncash items and working capital changes. Indirect is useful inside a three-statement model. For weekly cash decisions it hides timing: a large customer payment that slips by two weeks disappears inside a monthly accrual view.
Thirteen weeks is one quarter. It is long enough to catch every payroll cycle, a quarterly tax payment and a slow-paying customer, and short enough that most inflows and outflows are already known or under contract.
The structure, line by line
Each column is a week ending on a fixed day, usually Friday. Each row is one of the following.
Opening cash. Last week’s ending cash. For week 1, the actual balance across all operating bank accounts.
Receipts by source. Split by how cash arrives: collections on open receivables, new billings expected to collect inside the window, card or marketplace payouts, and other inflows such as refunds. Keep financing draws on their own line.
Payroll. Net pay, payroll taxes and benefits on actual pay dates. Contractors get a separate line.
Rent and facilities. Fixed amounts on fixed dates.
Vendors and AP. From the open AP aging plus recurring bills not yet entered. Break out the five largest vendors if they drive the total.
Taxes. Sales tax remittances, estimated income tax payments and other filings, with amounts confirmed by your licensed tax advisors.
Debt service. Interest and principal on their scheduled dates.
Capex. Equipment and capitalized spend, included only once committed.
Net cash flow. Total receipts minus total disbursements.
Ending cash. Opening cash plus net cash flow.
Minimum cash buffer. A fixed floor, typically four to eight weeks of core disbursements or any lender covenant minimum, whichever is higher. Flag every week where ending cash falls below it.
Building week 1 from bank data
Start by exporting the last eight to twelve weeks of transactions from every operating account. Tag each transaction to one of the rows above. This does two things. It shows the real timing of payroll, rent and large vendors, and it surfaces recurring payments nobody remembered to mention.
Then set week 1. Opening cash is today’s bank balance. Receipts come from the AR aging, adjusted for how each customer actually pays rather than the terms on the invoice. If a customer on net-30 terms pays in 48 days, forecast 48. Disbursements come from the payroll calendar, the open AP list and the recurring items you tagged. Weeks 2 through 13 follow the same logic, with more estimates and fewer known amounts as you move out.
The Monday update and variance review
The forecast is only as useful as its last update. Our routine takes about an hour every Monday.
Replace last week’s forecast with actuals from the bank.
Compare actual to forecast by line, and write a one-line reason for any variance above a set threshold, for example $10k or 10%.
Classify each variance as timing (the cash will still arrive or leave) or permanent (it will not).
Roll the forecast forward one week so there are always 13 weeks ahead.
Update receipt timing for any customer whose payment pattern has changed.
After four to six weeks, the variance log shows which lines you forecast well and which ones need a better source.
Five mistakes that make the forecast lie
Forecasting receipts from invoice terms. Customers pay on their schedule, not yours. Use observed days to pay.
Mixing accrual and cash. Revenue recognized is not cash collected. Keep the income statement out of this model.
Plugging the balance. If the forecast does not tie to the bank, find the missing transaction. A plug hides the exact problem the forecast exists to show.
Skipping the variance review. Without it, the same errors repeat every week and the team stops trusting the numbers.
Forgetting lumpy items. Annual insurance, quarterly taxes, bonus payouts and annual software renewals are easy to miss in a 13-week window.
Start from bank data, keep the rows few and the updates weekly, and the forecast will typically show a cash problem six to eight weeks before it arrives. If you want a head start, our free 13-week cash flow template on the Resources page has these rows, the minimum buffer check and the variance log already set up.
Written by
Leah Okonjo
Principal, FP&A
Leah Okonjo leads financial planning and analysis at Stavely & Rowe. Before joining the firm she built the forecasting function at two venture-backed marketplaces, one matching home services providers with customers and one connecting wholesale buyers with independent brands. At both she replaced a founder’s spreadsheet with a driver-based model, a 13-week cash forecast and a monthly budget-vs-actual the team could run without her.
She joined Stavely & Rowe in 2021. She builds most of the firm’s operating models, leads unit economics work for e-commerce and climate tech clients, and writes the firm’s guides on cash forecasting and SaaS metrics.
