A practical direct-method model turns near-term bank movements into a weekly operating discipline rather than a spreadsheet refreshed only during a crisis.
A 13-week cash flow forecast is a weekly direct-method schedule of opening cash, expected receipts, planned disbursements, financing movements, closing cash, and available liquidity. Build it from operational evidence, reconcile forecast to actual every week, roll the horizon forward, and preserve human review for uncertain timing, one-offs, and funding decisions.
The question the model must answer is operational: what cash is expected to enter or leave each bank-controlled pool, in which week, and what headroom remains afterward? That requires the direct method. Customer collections, payroll, supplier runs, taxes, interest, principal, capital expenditure, and financing drawdowns are forecast as cash movements. Each material line should be traceable to a source or named assumption.
An indirect forecast starts with profit and reconciles through noncash charges and balance-sheet movements. It is useful for explaining statutory cash flow and connecting a long-range plan to financial statements. It is less useful for deciding whether Thursday’s AP run can proceed because net income and monthly working-capital assumptions conceal exact payment timing. The two methods can coexist, but the weekly liquidity model should not be a compressed indirect cash flow statement.
Thirteen weeks is a planning convention, not a claim that uncertainty disappears. It spans enough payment cycles to expose recurring obligations and gives management time to change collections, spend, or funding. Weekly columns retain enough precision to act. The early weeks may use transaction-level evidence; later weeks rely more heavily on ageing patterns, purchase commitments, and run rates. Confidence should be visible rather than implied.
Start with unrestricted opening bank cash, not the general ledger balance without review. Restricted cash, trapped balances, uncleared items, and overdrafts may require separate treatment. Add receipts by meaningful source and disbursements by controllable category. End with closing cash, facility usage, remaining revolver availability, and a minimum-liquidity comparison. The structure must be detailed enough to diagnose a miss but stable enough to compare week after week.
| Line | Primary driver | Typical source |
|---|---|---|
| Opening cash | Prior week closing cash | Bank balances reconciled to treasury records |
| Customer receipts | Invoice due dates and expected payment behavior | AR ageing, collections notes, billing schedule |
| Other receipts | Contractual dates or confirmed events | Agreements, grant schedules, asset sale support |
| Payroll and benefits | Pay calendar, headcount, taxes, bonuses | Payroll register and HR calendar |
| Supplier payments | AP due dates, terms, approved payment runs | AP ageing, purchase ledger, procurement commitments |
| Tax | Filing and remittance dates | Tax calendar and approved computations |
| Debt service | Interest and principal schedule | Facility agreements and lender notices |
| Capital expenditure | Approved milestones and invoice timing | Capex register, purchase orders, project owners |
| Financing movements | Draw, repayment, and covenant decisions | Treasury plan and facility terms |
| Closing cash and availability | Opening cash plus net movement and facility limits | Calculated output with covenant or reserve constraints |
Avoid a giant “other” line. If an item can change a decision, give it a category or a one-off line with an owner. Conversely, excessive vendor-level presentation makes review impossible. Keep transaction detail in supporting schedules and summarize it into decision-relevant output lines. For a broader view of how finance workflows can be redesigned, see AIM’s finance transformation services.
A durable workbook follows FAST-style principles even when it is not formally certified. Separate inputs, calculations, and outputs. Use consistent time columns. Keep one formula pattern per row so reviewers can inspect logic horizontally. Never hide a hardcoded adjustment inside a formula; place the assumption in a labeled input cell with source, owner, and date. These habits reduce key-person dependency and make review faster.
AR ageing supplies invoice amounts and contractual due dates, but the contractual date is not always the expected receipt date. Collections notes and customer behavior provide the judgment layer. AP ageing and terms establish obligations, while the approved payment-run plan determines timing. Payroll calendars, tax deadlines, debt schedules, leases, insurance renewals, and capex milestones should feed known payments. Maintain a one-off register for litigation, acquisitions, annual bonuses, refunds, and other nonrecurring items.
Every manual override should answer four questions: what changed, why, who approved it, and when it expires. Without that discipline, adjustments accumulate and become invisible model logic. Scenario cases should change explicit drivers, such as collection delay or capex deferral, rather than overwrite output cells.
Freeze the prior forecast before importing actuals. Replace the completed week with reconciled bank activity, investigate cut-off items, and compare actual line by line with what was forecast. Then add a new thirteenth week, refresh source data, obtain owner updates, and publish a new approved version. The horizon always remains thirteen weeks.
The opening balance of each week must equal the preceding closing balance. The first opening balance must reconcile to usable bank cash. Version control should preserve the original forecast, actual result, adjustment history, preparer, reviewer, and publication timestamp. A forecast that is repeatedly overwritten cannot teach the team anything.
The bridge converts misses into process improvement. Classify each difference as timing, amount, omitted item, classification, or actuals cut-off. A customer receipt that arrives one week late is a timing variance, not necessarily a permanent shortfall. A supplier invoice absent from AP is an omission. A payment booked to the wrong category is classification. The bridge should explain the completed week and indicate whether the miss shifts later weeks.
Repeated timing misses may reveal that due dates are being treated as expected dates. Repeated omissions may expose a broken purchasing or invoice-intake process. Amount differences can reveal stale payroll or tax assumptions. This diagnostic loop is why review quality matters more than producing a cosmetically precise first forecast.
Model legal entities at the level where cash can actually be moved. A consolidated group may look liquid while a subsidiary cannot meet payroll because balances are trapped, transfer approval is pending, or a facility cannot be used by that borrower. Each entity schedule should show its bank cash, receipts, payments, local facilities, and proposed intercompany funding.
Intercompany movements must appear symmetrically: an outflow for the sender and an inflow for the receiver in the same period, then eliminated on consolidation. Add controls that flag unmatched amount, currency, or week. Do not eliminate the movement from entity views because the transfer still affects local liquidity.
For multiple currencies, retain transaction currency and local-functional views, then translate to a reporting currency with a documented rate source and convention. Separate operating cash movement from foreign-exchange translation. Test sensitivity where rate movement can affect headroom, and recognize that a reported currency total is not the same as immediately transferable cash.
Automation is well suited to repeatable, inspectable work: importing bank balances and open-item ledgers, applying mapping tables, generating recurring payroll and debt schedules, translating currencies, rolling formulas, checking intercompany pairs, and producing variance reports. AI implementation services can also support exception classification and draft commentary when source references remain visible.
People must still own assumptions about disputed receivables, strategic supplier payments, discretionary capex, covenant responses, financing draws, and unusual events. A system cannot infer commercial consequences from a ledger alone. The controller or treasury owner should approve actuals completeness; business owners should confirm uncertain timing; and an accountable leader should approve actions arising from a liquidity shortfall.
Agentic workflows can coordinate these steps, but authority should remain bounded. AIM’s guide to agentic AI solutions for finance explains how tool permissions, validation, and human escalation fit together.
Thirteen weeks gives treasury and operating leaders a quarter-like view while retaining weekly payment detail. It is long enough to expose payroll, tax, debt, and supplier pressure, yet close enough for invoice-level evidence. The horizon should roll forward every week rather than shrink toward a fixed quarter end.
Use the direct method: forecast actual cash receipts and payments by week. The indirect method starts with accounting profit and adjusts noncash items, which suits financial statement cash flow reporting but obscures payment timing. A liquidity forecast needs bank-oriented categories that owners can trace to invoices, calendars, and commitments.
Review forecast versus actual every week, before publishing the next roll-forward. Separate timing differences from amount differences, omissions, and classification errors. The purpose is not merely to score accuracy; it is to identify which assumptions, source feeds, or owner inputs must change so later weeks become more decision-useful.
Data extraction, mapping, recurring schedules, currency conversion, arithmetic, and variance reports can be automated. Human review remains necessary for disputed invoices, customer payment behavior, discretionary spending, financing choices, and unusual events. Automation should produce traceable exceptions and proposed updates, not silently replace accountable treasury judgment.
Model each legal entity first, including its own bank cash and funding constraints. Record intercompany transfers symmetrically in sending and receiving entities, then eliminate them in the consolidated view. This preserves entity-level liquidity while preventing group receipts and payments from being overstated. Flag transfers that require approvals or currency conversion.
Discuss the model, source data, controls, and automation needed for a forecast that remains explainable.