GuideStrategy & Corporate Finance
How to build a thirteen-week cash flow forecast and keep it accurate week after week
A thirteen-week cash flow forecast projects actual cash receipts and payments week by week for roughly one quarter ahead, using the direct method rather than adjusting profit. It tells management when cash gets tight, by how much, and which levers buy time. This guide covers the structure, how to build receipts and disbursements from source data, the weekly roll-forward and variance review, scenarios, where automation helps and the mistakes that destroy trust in the numbers.
On this page
- What a short-horizon cash forecast answers that a monthly forecast cannot
- Anatomy of the forecast, from opening cash to available liquidity
- Building receipts and disbursements line by line
- Source data to collect before the first weekly cycle
- The weekly cycle: roll forward, compare actuals and explain the gap
- Scenarios to run when the base case is tight
- Where automation and AI help, and who stays accountable
- Errors that make a cash forecast untrustworthy
- Questions and answers
- Sources
What a short-horizon cash forecast answers that a monthly forecast cannot
A monthly forecast built with the indirect method starts from planned profit and adjusts for non-cash items and movements in working capital. It is good for planning and for explaining results, but it hides timing. A month can end with comfortable cash while the third week, after payroll and a supplier run, dips below the overdraft limit.
The thirteen-week forecast works at the level where liquidity problems happen: individual weeks and identifiable payments. Its horizon is long enough to see a quarter's tax payment or a debt service date coming, and short enough that most inputs are known commitments rather than estimates. Companies under pressure, those with tight covenants and those going through a transaction tend to rely on it, but any business with seasonal or lumpy cash flows benefits.
It complements rather than replaces the longer forecast. The monthly or annual model sets direction; the weekly forecast tells treasury whether the plan can be paid for, week by week.
Anatomy of the forecast, from opening cash to available liquidity
- Opening cash
Reconciled bank balances at the start of the week, across all accounts that can actually be used.
- Operating receipts
Customer collections forecast by customer or segment, plus other operating inflows.
- Operating payments
Suppliers, payroll and benefits, rent, utilities and operating taxes.
- Non-operating items
Capital expenditure, interest and debt repayments, dividends, transaction costs and one-off items.
- Closing cash
Opening cash plus receipts less all payments; it becomes next week's opening balance.
- Available liquidity
Closing cash plus undrawn committed facilities, less minimum cash and trapped balances.
Building receipts and disbursements line by line
Set the structure and the cut-off
Fix the week definition, the bank accounts in scope and the line items, separating operating from non-operating flows. Agree when actuals are taken from the bank each week so every cycle compares like with like.
Forecast receipts from customer behavior
Start from the receivables ageing, but forecast large customers individually using how they actually pay: average days beyond terms, payment-run days and dispute patterns. Group the long tail by segment and apply observed collection curves.
Add sales not yet invoiced
For later weeks, convert forecast billings into collections using the same payment patterns. Keep this layer visible, because it is the least certain part of the forecast and the first to revise.
Map disbursements to their calendars
Load approved payables by scheduled payment run, then add payroll and benefits on their pay dates, rent and leases, debt service from the facility agreements, tax payments from the tax calendar and committed capital expenditure.
Derive closing cash and liquidity
Calculate closing cash for each week, add undrawn committed facilities and subtract minimum operating cash and balances that cannot be moved, such as cash held in restricted accounts or certain jurisdictions.
Check covenant headroom
Where facilities carry minimum-liquidity or net-debt covenants, show the forecast against each test date and the headroom remaining. Read the definitions in the facility agreement rather than assuming them.
Source data to collect before the first weekly cycle
The weekly cycle: roll forward, compare actuals and explain the gap
Each week, replace the week just ended with actuals from the bank, add a new week at the end of the horizon and update every remaining week for new information. Then compare the forecast made a week ago with what happened, line by line.
Variance analysis is the habit that makes the forecast credible. For each material difference, record whether it was timing (the cash moved to another week), permanent (it will not arrive or will not be paid) or an error in the model. Timing differences should reverse; if they keep growing, an assumption is wrong. Over a few cycles, these notes show which lines are reliable and which need a better method.
Keep a short commentary with every version: the closing liquidity low point, what changed since last week and why, and the actions under way. Management reads the commentary first and the numbers second.
Scenarios to run when the base case is tight
- If
Liquidity headroom falls below the agreed minimum in any week of the base case.
ThenModel a delayed-receipts case in which the largest customers pay later than their recent average, and identify which payments could move without breaching terms.
Receipts are the least controllable line, so the downside usually starts there.
- If
Suppliers are shortening terms or asking for deposits.
ThenRun a case with tighter payment terms for the affected suppliers and show the cumulative effect by week.
Term changes compound across payment runs and can move the low point earlier than expected.
- If
A covenant test date falls inside the horizon.
ThenShow the forecast metric against the covenant under the base and downside cases, and agree in advance what action each outcome triggers.
Lenders generally respond better to early, evidenced conversations than to surprises on a test date.
- If
A large one-off payment or receipt is uncertain.
ThenPresent it as a separate line with both outcomes rather than burying a probability-weighted figure in the base case.
Weighted amounts produce a number that will never actually occur and hide the decision.
Where automation and AI help, and who stays accountable
Most of the effort in a weekly forecast is collection: downloading bank statements, extracting ageing reports and payment-run files, and reconciling them. Bank feeds and scheduled extracts from the ledger remove that work and leave more time for analysis. Invoice processing automation, such as the approach in our AI invoice processing use case, also makes the payables side of the forecast more current.
Machine-learning models can predict when each open invoice is likely to be paid from a customer's payment history, which is often more accurate than assuming payment on the due date. They can also flag exceptions: a regular customer who has missed a usual payment run, or a supplier payment far outside its normal range.
None of this changes accountability. A named person should own the forecast, review model suggestions before they enter the base case and sign the weekly commentary. Our Finance Operations managed service covers forecasting and treasury analytics among its processes1, but even where operations are outsourced, the decisions drawn from the forecast stay with the company's finance leadership.
Errors that make a cash forecast untrustworthy
Plugging differences
Early signalA line called other or balancing item grows from week to week.
MitigationBan plugs; every variance gets a cause, even if the cause is that the model was wrong.
Forecasting receipts on due dates
Early signalReceipts are consistently overstated in the near weeks and caught up later.
MitigationUse observed payment behavior per customer and segment, and refresh it each month.
Stale assumptions
Early signalLater weeks are identical copies of the first forecast.
MitigationReview every week of the horizon each cycle, not just the next one.
Counting cash that cannot be used
Early signalHeadroom looks comfortable while operating accounts run short.
MitigationShow trapped and restricted cash separately and exclude it from available liquidity.
Questions and answers
Who should own the weekly cash forecast?
Usually the treasurer or, in smaller companies, the financial controller or head of FP&A, with the CFO reviewing it. Ownership means assembling the inputs, running the variance review and signing the commentary. Contributors such as credit control, accounts payable, payroll and tax provide their schedules on a fixed day each week, so the owner is not chasing data.
How accurate should a short-horizon cash forecast be?
There is no universal standard, but accuracy should be highest in the nearest weeks, where most flows are known, and lower further out. Track the variance by line and by week of the horizon over several cycles. If the near weeks are regularly off by amounts that matter for decisions, the problem is usually the receipts method or missing data rather than the template.
When should a company move from a weekly forecast to a longer horizon?
Keep the weekly forecast while liquidity is tight or a transaction is under way, and connect it to a monthly forecast covering at least the next year. Once headroom is comfortable and variances are small, many companies reduce the weekly forecast's detail or frequency and rely more on the monthly model, while keeping the ability to restart weekly reporting quickly.
Can the forecast be built in a spreadsheet?
Yes, many are, especially at first. A spreadsheet is fine while the structure is being agreed. It becomes risky when several people edit it, when data is pasted in manually from many sources, or when versions multiply. At that point, automated data feeds, locked formulas and version history matter more than the choice of tool.
Sources
- Managed Services: Finance Operations — ColdAI