- Forecast receipts and payments, not revenue and costs.
- Separate committed from discretionary outflows — they behave differently under stress.
- Roll it forward weekly and record variances against the previous version.
- Model a downside case; a single scenario is a wish, not a plan.
- Simplicity beats precision — a model nobody maintains forecasts nothing.
Why thirteen weeks
Thirteen weeks is one quarter of operating reality. It is long enough that a developing problem becomes visible with time to act, and short enough that most inputs are known rather than estimated.
Beyond a quarter, forecasting accuracy falls sharply for most operating companies, which is why longer-horizon planning is usually done monthly and at a lower level of detail.
Structure
Rows are categories of movement; columns are weeks. Keep the row count small enough to maintain.
| Section | Contents |
|---|---|
| Opening balance | Actual cash at the start of the week, across all accounts |
| Receipts | Customer collections, funding, refunds, other inflows |
| Committed outflows | Payroll, taxes, rent, debt service, contracted vendors, card settlement |
| Discretionary outflows | Marketing, hiring costs, equipment, anything deferrable |
| Net movement | Receipts less all outflows |
| Closing balance | Opening plus net movement, carried into the next week |
Where the inputs come from
Good inputs matter more than modelling technique. Most of them already exist in systems you run — the work is connecting them rather than estimating.
- Receipts from the receivables ledger, adjusted for each customer's actual payment behaviour.
- Payroll from the payroll calendar, including taxes and employer contributions.
- Committed vendor payments from the payables queue.
- Card settlement from the programme's fixed cycle.
- Annual and quarterly items from a calendar of known one-offs.
Scenario modelling
Build a base case and a downside case, and make the downside genuinely uncomfortable — a large customer paying thirty days late, or a collection failing entirely.
The output that matters is not the closing balance but the week in which the balance would breach your minimum. That date is what triggers action.
Variance review is the whole point
Each week, compare what you forecast against what happened, and record why they differed. Over a couple of months this produces something more valuable than the forecast itself: a calibrated understanding of how your inputs actually behave.
Forecasts that are never compared with outcomes drift into fiction quietly, because nothing corrects them.
Frequently asked questions
Weekly for the short horizon, because monthly buckets hide timing problems inside the month — payroll on the fifteenth against collections arriving on the twenty-eighth looks fine monthly and is not.
Whoever can see both receivables and payables, usually finance. It needs one owner rather than being assembled from several disconnected views.
Accurate enough to identify the week a problem appears. Precision to the currency unit is wasted effort; identifying the right week is the entire value.
Keep reading
Sources and further reading
Every factual statement on this page is checked against primary documentation. Terms change frequently, so confirm details with the provider before acting on them.
- FASB Accounting Standards Codification Reference point for accrual, expense recognition and close-process statements.
- Nacha — ACH Network rules and resources Used for statements about ACH timing and payment rails.
- FDIC — deposit insurance and pass-through coverage Reference for how deposit insurance applies, including through third-party arrangements.