A thirteen-week forecast can be arithmetically correct and still be unusable. The opening balance may be stale, receipt dates may come from invoice terms rather than collection evidence, payment changes may be mixed into the base case, and nobody may be responsible for explaining why last week’s forecast missed.
The operating question is therefore not only how to extend a spreadsheet across thirteen columns. Treasury needs a repeatable process for deciding what cash is in scope, when inputs close, who owns each timing assumption, how scenarios are changed, and which exceptions require review. This guide covers that weekly control cycle. It does not replace an annual FP&A forecast, authorize payments, prescribe a financing response or rank treasury software.
Quick answer
The forecast becomes decision-useful only when current bank cash, owned timing assumptions, scenario changes and forecast-to-actual exceptions are reviewed on a fixed weekly cadence.
Decision: Set one weekly operating standard for scope, data cut-off, assumption ownership, scenario use, variance thresholds, sign-off and the roll into a new thirteenth week.
Key takeaways
- Start from reconciled, available cash at a defined cut-off, not an unqualified general-ledger balance.
- Forecast receipts and disbursements on expected cash dates, with a named owner and evidence for every material timing assumption.
- Keep the best current estimate, downside assumptions and proposed management actions in separate layers.
- Compare the locked prior forecast with actual cash movements and classify each material variance before changing the model.
- Roll the horizon only after source checks, exception review and accountable sign-off are complete.
What the forecast must control
In this guide, a 13-week cash flow forecast means a rolling, weekly view of expected cash receipts, cash disbursements and resulting balances over the next thirteen weeks. Its purpose is near-term liquidity visibility, not earnings prediction. AFP’s cash-forecasting overview describes treasury forecasts as operational views of receipts and disbursements and notes that the model design depends on purpose, data and organizational scope.
The direct structure is simple: opening cash plus receipts minus disbursements equals closing cash, and one week’s closing cash becomes the next week’s opening cash. AFP’s receipts-and-disbursements methodology also stresses that schedules should be prepared on a cash basis and adjusted for when payments actually clear. The control burden sits around that arithmetic.
| Layer | Minimum content | Primary evidence | Control before use |
|---|---|---|---|
| Opening cash | Available balances by entity, account and currency at the stated cut-off | Bank reporting, cash-position record and reconciliation items | Reconcile scope, timestamp and material in-transit items |
| Receipts | Expected customer collections and other inflows by cash date | AR detail, collection notes, remittance evidence, contracts and recurring schedules | Challenge due-date assumptions, disputes, concentration and double counting |
| Disbursements | Expected payroll, tax, supplier, debt, rent, capital and intercompany payments | AP, payroll, tax calendars, contracts, approved commitments and local submissions | Separate obligation timing from payment authorization or proposed deferral |
| Scenarios | Explicit changes to timing, amount or probability | Documented scenario assumptions and approval record | Do not overwrite the base case or count an action before approval |
| Variance and sign-off | Prior forecast, actual cash, explanations, model changes and reviewer decision | Bank actuals, forecast snapshot, variance log and review record | Resolve material exceptions before adding the new thirteenth week |
The model should carry a version date, cut-off time, currency basis, scenario label and owner. A reader should be able to tell whether a number is an actual balance, a committed cash flow, an expected cash flow, an uncertain estimate or a proposed action. Without those distinctions, the same cell can be interpreted several ways.
Fix the cash perimeter and weekly cut-off
Before collecting inputs, define the cash perimeter. It answers which legal entities, bank accounts, currencies and cash states are included. The same perimeter must be used for the opening balance, forecast flows, actuals and variance analysis.
Define available opening cash
Opening cash should be tied to a named cut-off and reconciled to the cash-position process. Treasury should decide how to treat restricted balances, trapped cash, overdrafts, zero-balancing structures, sweeps, deposits in transit, unpresented payments and accounts that report late. The accounting balance and the cash that can be used operationally may not be the same. The forecast should show that difference rather than hide it in an unexplained adjustment.
The underlying bank-connectivity and cash-reporting design should make account coverage, source timestamps, pending-item treatment and late-reporting behaviour visible.
Set one cut-off rule
A weekly cut-off should state:
- the time and time zone used for bank balances and actual cash movements;
- the last accepted source-system extract or departmental submission;
- how late information is treated after the model is frozen;
- which foreign-exchange rates translate non-base-currency cash flows;
- who may reopen a signed-off forecast and how the change is recorded.
Late data should not be silently inserted into a prior snapshot. Record it as a post-cut-off change, assess whether it is material, and preserve the locked version used for the original decision. This makes the next forecast-to-actual review meaningful.
Forecast receipts by expected cash date
Customer due dates are a starting input, not proof of when cash will arrive. ACT’s cash-forecasting guidance points to customer payment history, internal invoice cut-offs and recurring payment behaviour as relevant forecast evidence. Treasury and AR should convert those signals into an expected cash date and keep the source visible.
| Receipt status | Typical evidence | Forecast treatment | Required owner action |
|---|---|---|---|
| Committed | Confirmed remittance, settlement advice or contractually fixed receipt | Place on the expected clearing date | Confirm amount, account, currency and any settlement lag |
| Expected | Open invoice plus current collection status and reliable payer history | Use the evidence-based expected date, not the invoice date by default | Record the timing basis and update when collection evidence changes |
| Uncertain | Dispute, unbilled milestone, conditional sale or weak payment history | Keep visible with an uncertainty flag or scenario treatment | Name the condition that would move or remove the receipt |
| Non-recurring | Asset sale, insurance recovery, tax refund or other one-off item | Show separately from ordinary collections | Verify the approval, condition and expected settlement path |
Material receipts should carry a source, owner, last-updated date and reason for the expected week. Aggregated receipt lines still need a documented rule, such as a customer-specific collection curve or a cohort assumption. The rule should not mix customers with materially different behaviour merely to make the model easier to populate.
Collections teams should also identify concentration. A total may look stable while depending on one large receipt. Treasury does not need to list every invoice in the executive view, but the working file should allow a reviewer to trace a material weekly total back to its drivers.
Forecast disbursements from obligations and payment decisions
Disbursements should begin with known obligations and expected clearing dates. The source mix commonly includes payroll and benefit calendars, tax dates, debt-service schedules, rent and lease commitments, approved supplier invoices, purchasing commitments, recurring services, capital expenditure and intercompany settlements. Large project disbursements may also depend on readiness: the data-center financing readiness case shows how permits, power, community conditions and lender requirements can affect when committed capital is available to draw.
Separate four states that are often blended:
- Contracted or statutory obligation: the organization owes the amount under a contract, payroll rule, tax requirement or financing agreement.
- Approved for payment: the payment has passed the applicable approval process.
- Scheduled or released: a payment instruction has an expected execution and clearing date.
- Proposed timing change: management is considering a deferral, acceleration, cancellation or negotiation.
The forecast records expected cash movement; it does not replace payment authority. A proposed deferral should remain outside the base case until the responsible owner confirms that it is permitted and approved. The same rule applies to accelerated supplier payments, discretionary spend reductions and other management actions.
The Vistry credit-limit procurement watch separates a reported insurer action from a confirmed supplier-term or order change before treasury moves it into the base cash forecast.
AP data also needs a clearing assumption. A payment issued on Friday may not reduce the relevant bank balance in the same forecast week. The model should apply the organization’s observed payment-rail and banking cut-offs rather than treating issue date and cash date as identical.
Keep the base case, downside case and management actions separate
A scenario is useful only when the reader can see what changed. AFP’s cash-forecasting practices call for agreement on purpose, acceptable output and accuracy measures, together with documented assumptions and timely variance review.
Apply that discipline to external data too. The June 2026 Japan current-account treasury test separates dividend timing, seasonal data and import costs before a headline changes JPY hedge or liquidity assumptions.
| Layer | What it represents | What must be recorded | Control rule |
|---|---|---|---|
| Base case | The best current estimate using approved assumptions | Source, owner, timing basis and current evidence | Do not add hidden conservatism or unapproved actions |
| Downside case | A defined adverse change, such as later receipts or higher required payments | Changed variables, effective weeks and rationale | Apply each change once and preserve the unchanged base case |
| Upside case | A defined favourable change supported by evidence | Changed variables, probability basis and dependency | Do not treat aspiration as expected cash |
| Management actions | Potential responses to a projected constraint | Owner, approval status, lead time and cash effect | Keep separate until approved and executable |
Each scenario should start from the same locked base version. If a downside case moves a customer receipt and the base case is also edited, the reader cannot tell whether the difference came from the scenario or from a general model update. A short scenario-change log prevents this loss of traceability. The Dolce & Gabbana covenant-waiver case shows why cash already raised, still-unspecified financing and possible asset proceeds require separate forecast layers.
Run a fixed weekly preparation and review cadence
The weekly process needs named deadlines, not a vague instruction to “update the forecast.” AFP’s treasury policy guidance recommends setting the forecast’s goals, frequency, format, update schedule, acceptable methods and variance-analysis requirements. It also identifies Payroll, AR, AP, tax and FP&A as important contributors.
| Stage | Primary action | Owner | Required output | Control gate |
|---|---|---|---|---|
| 1. Freeze actuals | Capture opening cash and actual movements through the cut-off | Treasury cash-position owner | Reconciled opening position and actuals file | Scope and timestamp agree with policy |
| 2. Refresh inputs | Update receipts, payments and assumption evidence | AR, AP, payroll, tax and other line owners | Submitted changes with explanations | Late, missing and duplicate inputs are flagged |
| 3. Consolidate and challenge | Apply model rules, review material movements and test scenario separation | Treasury forecast preparer | Draft forecast and exception list | Every material change has an owner and source |
| 4. Review variance | Compare the locked prior forecast with actual cash | Treasury plus accountable line owners | Classified variance log and corrective actions | Policy thresholds and recurring bias are addressed |
| 5. Approve and roll | Confirm the decision view, sign off and add the new thirteenth week | Treasurer, cash manager or delegated reviewer | Locked version, review record and distribution | Open exceptions are accepted, escalated or resolved |
The day names are local choices. What matters is that the cut-off, submission window, challenge meeting and sign-off happen in a stable order. The forecast should not roll merely because the calendar advanced. It rolls when the current version has passed its control gates.
Turn forecast-to-actual variance into a control loop
Variance analysis should compare a locked forecast with actual bank cash, not a forecast that was edited after the outcome became known. ACT’s forecast-refinement note describes line-item forecast-to-actual reconciliation and recommends involving the people who know the underlying cash flows when explaining differences.
Use a small set of variance categories that point to different fixes:
| Variance type | Example | Likely response |
|---|---|---|
| Timing | The amount was correct but cleared in another week | Adjust the cash-date rule, banking lag or customer behaviour assumption |
| Amount | The cash flow occurred in the expected week at a different value | Correct quantity, price, tax, foreign-exchange or deduction logic |
| Missing item | An actual cash flow had no forecast line | Add the source, owner or recurring schedule that failed |
| Classification | The total cash was correct but assigned to the wrong category or entity | Repair mapping and preserve comparability across periods |
| Management action | An approved acceleration, deferral or cancellation changed cash | Record the approval and distinguish the action from forecast error |
| Cut-off or data failure | Late or stale information distorted opening cash or a weekly flow | Fix extraction timing, submission discipline or escalation |
Do not rely on one blended accuracy percentage. A net closing-cash variance can look small because receipt and payment errors offset each other. Review gross receipts, gross disbursements, closing cash, timing shifts and directional bias by horizon. A recurring optimistic bias in weeks five to thirteen may need a different response from a one-off bank cut-off error in week one.
For every material variance, record the affected week, amount, category, root cause, accountable owner, correction and whether the forecasting rule changed. The next run should show whether the correction worked. The ACT guidance cited earlier states that forecasts should use reliable base data, be updated for known changes and be checked against actuals over time.
Responsibility, control and handoff matrix
| Role | Primary responsibility | Evidence supplied | Control performed | Handoff |
|---|---|---|---|---|
| Treasury process owner | Set scope, calendar, methods, thresholds and distribution | Policy, model rules and prior signed-off version | Confirm consistency and resolve cross-functional exceptions | Signed decision view to the treasurer or CFO |
| Cash-position owner | Prepare available opening cash and actual movements | Bank balances, statements, feeds and reconciliation items | Verify perimeter, timestamp and material in-transit cash | Reconciled actuals to the forecast preparer |
| AR and collections | Forecast customer and other operating receipts | Open receivables, disputes, promises, remittances and payer history | Challenge due dates and explain collection changes | Receipt schedule and exceptions to treasury |
| AP and procurement | Forecast supplier and purchasing-related payments | Approved invoices, commitments, terms and payment runs | Separate obligations, approvals, release dates and proposed changes | Disbursement schedule and approval status to treasury |
| Payroll and HR | Provide payroll, benefit and headcount-related cash dates | Payroll calendar, approved changes and funding requirements | Confirm completeness and exceptional runs | Weekly cash requirement to treasury |
| Tax, debt and intercompany owners | Provide specialist cash schedules within their remit | Filing calendars, agreements, notices and settlement plans | Confirm dates, currencies, conditions and approvals | Controlled schedules to the forecast preparer |
| Business unit or FP&A contributor | Explain operational changes not yet visible in transaction systems | Approved plans, milestones and current business evidence | Reconcile operational assumptions with submitted cash effects | Documented changes to treasury, not an untraceable plug |
| Treasurer, cash manager or CFO reviewer | Challenge projected liquidity, scenarios and unresolved exceptions | Draft forecast, variance log, scenario log and exception list | Accept, reject or escalate the version and its limitations | Approved version for controlled distribution |
Segregation should match the organization’s size. In a smaller team, one person may prepare several inputs, but the same person should not silently change material assumptions and approve the final version without an independent challenge. Where full segregation is impractical, record the compensating review.
Sign-off checklist before the horizon rolls
The reviewer should be able to answer each question before the oldest week is removed and a new thirteenth week is added:
- Does opening cash reconcile to the approved perimeter and cut-off?
- Are restricted, trapped, swept and in-transit balances treated consistently?
- Do material receipts use expected cash dates with current evidence?
- Do material payments distinguish obligation, approval, release and clearing?
- Are scenario changes visible and free from double counting?
- Are proposed management actions separate from the base case until approved?
- Was the prior forecast compared with actual cash using the locked snapshot?
- Do material variances have a cause, owner and corrective response?
- Are late or missing inputs disclosed rather than hidden by a plug?
- Has the reviewer accepted or escalated every open exception above policy thresholds?
If one of these checks fails, the forecast does not always need to stop. It does need an explicit status. Mark the version provisional, state what is missing, identify the decision that could be affected and assign a deadline for resolution. That is more useful than presenting an unqualified number whose limitations are known only to the preparer.
Frequently asked questions
Should receivables in a 13-week cash flow forecast use invoice due dates or expected cash dates?
Receivables should use expected cash dates when the available evidence supports them, rather than treating contractual due dates as automatic cash dates. Treasury should consider payment history, payer cut-offs and certainty, then retain the assumption owner and evidence so later variance review can distinguish timing error from a changed collection outcome.
How often should a 13-week cash flow forecast be updated?
A 13-week cash flow forecast should run on a fixed weekly preparation and review cadence, with known changes incorporated and the prior locked version compared with actual cash. The policy should define cut-off, input timing, review and roll-forward responsibilities so a late or missing input becomes a visible exception instead of a silent plug.
How should scenarios and management actions be shown in the forecast?
The base case, downside assumptions and proposed management actions should remain separate layers. A reviewer needs to see what changed, who owns the assumption and whether an action is approved; combining them can double count effects or make an unapproved response appear to be part of the underlying forecast.
