How to Build and Run a 13-Week Cash Flow Forecast
By Mike White, 2-Squared Advisory
A 13-week cash flow forecast is the single most useful financial tool a $3M–$15M founder-led business can install. It takes a controller a day to build, a half hour a week to update, and 30 minutes a week to review — and it will surface problems six to ten weeks before they hit the bank account. Here's exactly how to build one.
The structure
The simplest possible model has six row groups:
- Starting cash (Monday morning bank balance)
- Receipts: A/R collections by major customer, deposits, other income
- Operating disbursements: payroll, payroll taxes, rent, COGS, insurance, utilities, A/P
- Non-operating disbursements: debt service, tax payments, distributions, capex
- Net change in cash
- Ending cash + line of credit availability
How to populate it the first time
Receipts
Pull A/R aging and forecast collection week by week based on each customer's actual pattern — not their stated terms. A customer with stated net-30 who actually pays in 48 days gets forecast at 48 days. Be honest.
Operating disbursements
Payroll is the easiest — it's a known number on a known day. A/P comes from your open invoice list, scheduled to your actual payment cadence. Rent, insurance, and recurring software are easy. The trick is remembering them all the first time.
Non-operating disbursements
This is where most forecasts die. Quarterly estimated taxes, annual insurance, owner distributions, equipment purchases, debt principal — all of it has to be on the schedule. The first time you build it, sit down with your tax accountant and list every payment due in the next 13 weeks. None of these surprises should ever be a surprise.
The weekly cadence
- Monday morning: The owner of the forecast pulls actuals from the prior week, replaces forecast with actuals, and rolls the model forward one week.
- Monday afternoon: Review variances. Anything more than ±10% on a line gets a one-sentence explanation in the comment column.
- Tuesday 30-minute meeting: Owner + forecast owner walk through variances, look at the 13-week trough (the lowest cash point coming up), and confirm any actions — calling a customer, releasing A/P, drawing on the line.
Common mistakes
- Too granular. 40 customer rows is data, not forecast. Roll up.
- Not using actuals. Forecast columns must be replaced with actuals each week. Otherwise variance is invisible.
- Owner builds it. The owner should never build the forecast. The owner should review it.
- No line of credit row. If you have a line, model its balance and availability. It's part of cash.
- Hiding the trough. The whole point is to see the low week coming. Don't average it away.
What to do next
Build version one this week. It will be ugly, it will be wrong in places, and that's fine. Run it for four weeks before deciding what to change. By week six, you'll wonder how you operated without it.
Related services
Frequently asked questions
Why 13 weeks and not a quarter or a year?
Thirteen weeks is far enough out to see the cliff and short enough to be honest. Beyond ~90 days, weekly cash forecasts are guesses. Inside 90 days, every line is a real receivable, a real bill, or a real payroll run.
Who should own it?
One person — usually the controller, bookkeeper, or fractional CFO. The owner reviews it; they don't build it. If three people own it, no one owns it.
How is this different from my P&L?
The P&L is accrual and doesn't care when cash moves. The 13-week is the opposite: it only cares when cash moves. Revenue booked today might not be cash for 75 days; a tax payment may not hit the P&L at all but lands in your bank account on a specific day.
Will my bank want to see this?
Yes, and increasingly they require it for working-capital lines above ~$500K. A clean weekly 13-week is one of the highest-trust signals you can send to a lender.