What is a 13-week cash flow forecast?
A 13-week cash flow forecast is a weekly, direct-method projection of cash receipts and disbursements over one quarter. It starts from bank activity rather than net income, so it answers a different question than your financial statements: not whether the business was profitable, but whether there is money in the account on a specific Friday.
That distinction is the whole reason the tool exists. A company can be comfortably profitable and still miss payroll, because profit is an accrual concept and payroll is a cash event. The 13-week is built to catch exactly that collision.
Why thirteen weeks, and why weekly
Thirteen weeks is a quarter. That is not an arbitrary horizon — it lines up with the test dates in most credit agreements and with most sponsor reporting cycles, so the forecast ends where a covenant calculation begins.
Weekly granularity is the more important half. Cash problems are almost always timing problems, and a month that nets comfortably positive can still contain one week where a debt service payment, a payroll run, and a quarterly insurance premium land before a large receivable clears. Monthly forecasting averages that week away. The company then discovers it on a Tuesday.
Beyond thirteen weeks, weekly precision becomes false precision. You do not know which week in month five a customer will pay. That horizon belongs to the three-statement model, which forecasts monthly and answers a different question.
The line-item structure
Below is the structure that works for most lower-middle-market companies. It is deliberately shorter than the chart of accounts: the forecast is a cash instrument, and detail that does not change a decision makes it harder to update every week.
| Section | Line item | Source |
|---|---|---|
| Opening | Beginning cash | Prior week closing, tied to bank |
| Receipts | Collections — AR ageing | AR ageing × collection curve |
| Collections — new billings | Revenue forecast × terms | |
| Other receipts | Refunds, insurance, asset sales | |
| Total receipts | Sum | |
| Disbursements | Payroll and payroll taxes | Payroll calendar |
| Benefits and PTO | Monthly, by due date | |
| Accounts payable | AP ageing × payment terms | |
| Inventory / direct materials | Purchase plan | |
| Rent and occupancy | Lease schedule | |
| Insurance | Policy calendar | |
| Capital expenditure | Approved capex plan | |
| Debt service — interest | Debt schedule | |
| Debt service — principal | Amortisation schedule | |
| Taxes | Estimated payment dates | |
| Other operating | Everything remaining | |
| Total disbursements | Sum | |
| Net | Net cash flow | Receipts less disbursements |
| Ending cash | Opening + net | |
| Facility | Revolver draw / (repay) | Plug to minimum cash |
| Revolver balance | Running | |
| Borrowing base availability | Eligible collateral × advance rate | |
| Excess availability | Availability less balance |
The last block is the one companies most often leave out and lenders most often turn to first. Ending cash of $400K means one thing with $3M of untouched revolver behind it and something very different with $150K of availability left.
Building the receipts side
Receipts are where forecasts go wrong, because it is tempting to assume customers pay on terms. They do not.
Take the AR ageing and build a collection curve from actual history — what percentage of a month's billings is collected in week one, week two, and so on. For most businesses this is stable enough to forecast against and materially different from stated terms. A company on net-30 with an average collection of 47 days is not experiencing a series of unfortunate exceptions; 47 days is its actual behaviour, and the forecast should say so.
A worked version. Suppose $2.4M of open AR:
| Ageing bucket | Balance | Expected collection | Forecast week |
|---|---|---|---|
| Current | $1,100,000 | 92% | Weeks 3–6 |
| 1–30 days | $780,000 | 88% | Weeks 1–4 |
| 31–60 days | $340,000 | 74% | Weeks 2–5 |
| 61–90 days | $120,000 | 45% | Weeks 4–8 |
| 90+ days | $60,000 | 15% | Unscheduled |
Applying those rates: $1,012,000 + $686,400 + $251,600 + $54,000 + $9,000 = $2,013,000 of the $2.4M ageing is expected to convert, with the rest either collected later or eventually reserved. Spreading that across the weeks shown gives the collections line its shape.
Two customers concentrated enough to move a week get forecast individually rather than through the curve. Averages are for populations, and a customer representing 18% of receivables is not a population.
Building the disbursements side
Disbursements are more predictable, which makes the errors here more embarrassing when they happen. Most of the line items are calendar-driven and can simply be scheduled: payroll dates are known for the year, rent is known, debt service is in the credit agreement, insurance renews on a date.
The one that requires judgement is accounts payable. Build it from the AP ageing and your actual payment behaviour, then be honest about stretch. If the company has been paying at 52 days against 30-day terms, the forecast should show 52 — and the stretch should be visible as a decision rather than buried as an assumption, because it is a form of financing and a lender will read it that way.
The weekly routine
The forecast earns its value from being rolled, not from being built. The routine is short:
- Enter last week's actuals beside what was forecast.
- Review the variance line by line, not in total. A net variance of $8K can be a $300K collection that arrived early against a $292K payment that went out late, and those are two separate things to know.
- Adjust the assumptions the variance disproved.
- Add week 14, so the horizon stays at thirteen.
- Circulate one page: opening cash, receipts, disbursements, closing cash, availability, and a line on anything unusual.
Step two is the one that gets skipped and the one that makes the model better. A forecast nobody compares to reality does not improve; it just keeps being produced.
How lenders actually read it
A lender is not reading your forecast to admire the modelling. They are checking a small number of things, roughly in this order.
Does it tie to the bank? Opening cash must equal the bank statement. A forecast that starts from an unreconciled figure is not evidence of anything, and this is the fastest way to lose credibility with a credit analyst.
Where is the trough? Not the ending balance — the lowest point across thirteen weeks, and how close it comes to the minimum cash the business needs to operate. Ending cash of $1.2M is irrelevant if week seven dips to $80K.
How much availability is left? Excess availability under the borrowing base is the real liquidity number. Cash plus availability is what the company can actually deploy.
Has the forecast been right before? This is why the variance history matters. A company presenting its ninth consecutive weekly update, with the prior weeks' variances visible, is making a much stronger claim than one presenting a first forecast — regardless of what the two documents say.
Is anything being stretched? A DPO drifting from 38 to 54 days across the forecast is visible in the AP line, and it is read as borrowing from suppliers.
Four mistakes worth avoiding
Forecasting revenue instead of collections. A sale is not cash. The gap between them is the forecast's entire subject.
Rebuilding rather than rolling. If the file is reconstructed each time it is requested, no variance history accumulates and the model never learns.
Burying the assumptions. Collection rates and payment terms belong on a labelled input tab. Hardcoded inside formulas, they cannot be reviewed, and nobody will find them a year later.
Treating the revolver as a plug and stopping there. If the model draws to hold minimum cash, someone must check the draw is actually available under the borrowing base. A forecast that quietly assumes a draw exceeding availability is describing a facility you do not have.
A starting template
A workbook with the structure above — sections, a rolling thirteen-week grid, the input tab, and the variance comparison — is available here:
Download the 13-week cash flow template (XLSX)
It is a starting structure rather than a finished model. The parts that make it yours are the collection curve built from your history, the payroll and debt calendars, and the borrowing base definition from your specific credit agreement — which is exactly where the engagement work goes.
What the 13-week cannot tell you
It is worth being precise about the limits, because the tool gets asked to answer questions it was not built for.
It says nothing about profitability. A company can show comfortable weekly cash while losing money, because collections in any given week reflect sales made sixty days ago. Cash timing and earnings are different quantities, and a forecast that looks healthy is not evidence that the business model works.
It does not forecast beyond a quarter, and should not be extended to try. Weekly precision at month five is invented — you do not know which week a customer will pay four months from now. That horizon belongs to the three-statement model, which forecasts monthly and carries the balance sheet.
It does not value working capital improvements. The 13-week will show that receivables are collecting slowly; it will not tell you what fixing that is worth over a year, because it has no annual view. That calculation happens in the model.
And it assumes the opening balance is right. Every figure downstream inherits the accuracy of one cell. If the bank reconciliation is stale, the forecast is confidently wrong for thirteen weeks.
The practical arrangement is both instruments running together: the 13-week for liquidity and covenant proximity, the monthly model for the year and the strategy. Companies that maintain only one usually maintain the wrong one for the question they are being asked.