In short
- Start from a reconciled bank position, then model receipts on how customers actually pay rather than on invoice terms.
- Layer payables, payroll, tax, fixed commitments, debt service and the inventory cycle onto their real dates.
- Mark the minimum acceptable balance, run realistic scenarios, and compare forecast to actual every week.
Step 1 — Fix the opening position
Start from the actual bank balance across all accounts and facilities, reconciled, on a defined day of the week. Everything downstream is a movement from this number, so an unreconciled start makes the rest cosmetic.
Step 2 — Build receipts from behaviour, not terms
Take the receivables ledger and age it by how customers actually pay, not by the terms on the invoice. If nominal terms are 45 days and effective collection runs longer, model the longer number. Then add forecast new sales and convert them to cash using the same behaviour profile, offset by the sales cycle.
This single step is the most common reason forecasts fail. A model built on nominal terms will show comfort that does not exist.
Step 3 — Layer in payments
- Payables from the ledger, scheduled by supplier payment run.
- Payroll and related statutory payments on their fixed dates.
- Sales tax, income tax and other statutory obligations by due date.
- Rent, utilities, insurance and other recurring commitments.
- Debt service, lease payments and any planned capital expenditure.
Step 4 — Add inventory and the working capital cycle
If the business holds stock, purchasing decisions consume cash weeks before the related sale returns it. Link purchase commitments to the production or buying plan rather than treating them as a smooth run rate — that is where lumpy weeks come from.
Step 5 — Set the trigger, then run scenarios
Define the minimum acceptable balance, including facility headroom, and mark it on the chart. The forecast is not there to predict a number; it is there to show whether and when the line is crossed.
Then test the realistic risks: a key customer paying two weeks late, a slow sales month, a supplier shortening terms, an unplanned repair. Each scenario should produce an action and a date by which the action must be taken.
Step 6 — Run the weekly loop
Each week, compare forecast to actual, explain the difference, and correct the assumption that caused it. After a handful of cycles the forecast becomes accurate enough to make decisions on — which is the entire point. A model that is never compared with reality stays a spreadsheet.
Talk this through with Finviq
Tell us where your reporting stops short. In 30 minutes we'll identify the first two or three areas worth investigating in your business.