A P&L can show a profitable month while the bank balance heads toward zero. That is not a contradiction. It is the ordinary gap between when revenue is earned and when cash arrives, and no income statement shows it. The 13-week cash flow forecast closes that gap.
What a 13-week forecast is for, and what it is not
It answers one question: will there be enough cash in the bank in each of the next thirteen weeks, and if not, which week does it break? That is a liquidity question, not a profitability question, and confusing the two is why most attempts fail.
So it ignores nearly everything your P&L cares about. Revenue recognition, deferred revenue, accruals, depreciation: none of it appears. A signed contract is worth nothing here until the payment clears. An expense incurred in March but paid in June sits in June. The difference between accrual and cash accounting matters here.
It is also a different instrument from your runway model. Runway and burn rate tell you how many months you have at your average pace. Averages are useless for the week payroll and a quarterly tax payment land three days apart.
Why thirteen weeks
Thirteen weeks is one quarter, fifty-two divided by four. A quarter is long enough that a gap surfaces while you can still act: finding out in week nine that week twelve is short leaves room to accelerate collections, stretch a vendor, or delay a hire. Finding out in week twelve leaves you with a phone call and an apology. It is short enough that weekly precision stays honest: you can name the customers who owe you and the bills already in your inbox.
Direct method, not indirect
There are two ways to present operating cash flow. The indirect method starts with net income and adjusts for non-cash items and working capital changes. The direct method reports gross cash receipts and payments, the money actually moving.
For statutory reporting, ASC 230-10-45-25 encourages the direct method but stops short of requiring it, and most companies present operating cash flows using the indirect method (see Deloitte's Roadmap to the Statement of Cash Flows, chapter 3, on form and content).
For a 13-week forecast it is not really a choice. Build it direct. The indirect method starts from net income, and you have no weekly net income worth starting from. More importantly, it cannot tell you that a customer's wire lands Tuesday and payroll clears Wednesday. The direct method is built from the payment calendar itself: AR aging, AP aging, payroll dates, rent, debt service.
The row structure
Columns are weeks. Thirteen of them, labeled with the week-ending date, never with "Week 1." Dates survive a rolling file; labels do not. Rows follow a five-block skeleton that does not change.
| Block | What goes in it | Where it comes from |
|---|---|---|
| Opening cash | Balance across every operating account | Prior week's closing, reconciled to the bank |
| Receipts | Customer collections, then other inflows | AR aging plus observed payment behavior |
| Disbursements | Payroll, AP, rent, debt service, taxes | AP aging plus the fixed calendar |
| Net cash flow | Receipts minus disbursements | Calculated |
| Closing cash | Opening plus net cash flow | Feeds next week's opening |
One mechanic is non-negotiable: the closing cash of week N must be hard-linked as the opening cash of week N+1. Link the cells, never retype a balance. A forecast hand-keyed anywhere in that chain will eventually disagree with itself.
Inside receipts
- List collections by named customer, not by revenue line. You are forecasting the behavior of specific payers, and payers differ.
- Split committed collections (invoices issued) from uncommitted ones (deals not closed). Keep uncommitted in its own row so you can switch it off and see the floor.
- Keep non-operating inflows separate: loan draws, equity, tax refunds. Blend them in and you hide a deteriorating core business behind a one-time deposit.
Inside disbursements
- Payroll on its actual clearing dates, including the months the calendar produces a third run. Employer taxes and benefits ride alongside it.
- Accounts payable by vendor, from the aging, placed in the week you intend to pay, not the week the terms say.
- The fixed calendar: rent, debt service, insurance and software renewals, tax remittances.
- Known one-offs on their own line: a settlement, a deposit, an equipment purchase. Anything lumpy should never hide in a general AP row.
Building the first one
- Pick a week-ending day and lock it. Friday is conventional; consistency matters more than the choice.
- Set opening cash from the bank, not the books, reconciled across every account. If the first cell is wrong, all thirteen columns are wrong.
- Pull the AR aging and place every open invoice in the week you expect collection, based on how that customer has actually paid before.
- Pull the AP aging and place every bill in the week you intend to pay it. Intent, not terms. This is where it becomes a decision document.
- Layer in the fixed calendar: payroll runs, rent, debt service, tax dates, renewals.
- Add uncommitted receipts last, in their own row.
- Calculate net and closing, then find the lowest closing balance across the thirteen weeks. Not the ending balance. The minimum.
A forecast that ends week 13 comfortably positive can still be insolvent in week 6. Put the minimum closing balance, and the week it occurs, at the top of the file. If one number reaches your leadership team, make it that one.
The weekly cadence
Every Monday the completed week drops off the front and a new week thirteen is appended, so the horizon stays constant. That is what makes it rolling rather than a quarterly ritual that is irrelevant by week seven. When a week completes, do not overwrite the forecast: keep both side by side for every row.
Once the file exists and the inputs are clean, the roll is short. It stays short only if the data is current, meaning a reconciled bank feed and an AR and AP aging you believe. Teams that struggle with the roll usually have a bookkeeping problem, not a forecasting problem. A tight monthly close is the foundation this stands on, and if closing is a scramble, fix that first.
Variance analysis is where the value is
A forecast nobody checks against reality is a guess with formatting. Each Monday, walk the rows where actual diverged from forecast and ask one question first: timing or amount?
- A timing variance means the cash was right but the week was wrong. The customer paid, just later than modeled. Fix the assumption, not the number.
- An amount variance means the figure was wrong. Someone paid short, a bill was larger than expected, a renewal repriced. Fix the source data.
- Then ask whether it is temporary or permanent. A one-week slip self-corrects. A customer who has quietly moved to net 60 never reverses, and every column built on net 30 is now wrong.
Track variance by row across several weeks rather than agonizing over any single one, because individual weeks are noisy. A row that misses in the same direction three or four weeks running is not bad luck. It is a broken assumption, and it will keep lying to you until you fix the input behind it.
Common failure modes
- Building it from the P&L. Revenue is not collections, and starting from the income statement guarantees a forecast wrong in exactly the way it exists to prevent.
- Using stated payment terms instead of observed behavior. Net 30 is an aspiration; your aging report is evidence.
- Taking opening cash from the books rather than the bank, which poisons all thirteen columns.
- Blending uncommitted receipts into committed ones, converting a liquidity document into a sales forecast.
- Reporting the ending balance instead of the trough, which hides the week that actually breaks.
- Letting it stop rolling. A file last updated three weeks ago is worse than none, because people still believe it.
- Treating it as a model rather than a document. If no decision changes, it is not a forecast. It is a hobby.
Who this is for
The 13-week comes out of turnaround work, which gives it a reputation as a distress tool. That is unfortunate, because the companies that get the most from it are the ones growing fast enough that cash timing has become complicated. Headcount, inventory, enterprise payment cycles, and debt service all widen the gap between profit and liquidity. Build one before you need it.