How to forecast cash for the next ninety days
Build a rolling ninety day cash forecast from your own bank balance and dated commitments so you can see the week a shortfall arrives before it lands.
The week you run short is knowable in advance
Most owners discover a cash problem on the morning it becomes one. The year is profitable. Salaries go out on the first. A statutory payment falls due mid month. Then the largest customer pays forty days late and the decision becomes which vendor to hold back. A ninety day forecast does not create money. It tells you which week the gap arrives so the decision gets made with four weeks of notice rather than at the counter.
This guide covers a rolling forecast built by hand in a spreadsheet and reviewed weekly. It does not cover audited projections for a lender or an investor. Those are a different document with a different standard of proof and they need a chartered accountant.
What it costs and what it will not tell you
A cash forecast excludes anything you cannot name and date. It will not tell you whether a customer honours a promised payment date. It will not value a pipeline deal that has no signed order. It will not catch a cost you have never recorded. Everything it shows is a restatement of information you already hold.
The direct cost is time. Expect two or three hours to build the first version and roughly half an hour each week to update it. No software purchase is required. What you must supply is the real data: balances across every current account, the invoice ageing list with actual payment behaviour per customer, payroll, loan instalments, rent, vendor commitments and dated statutory outflows such as GST and TDS deposits. Statutory dates and amounts follow rules that change. A qualified professional should confirm them for your situation rather than you working from last year's pattern.
Who needs this and who is wasting a morning
A ninety day model earns its keep when money arrives and leaves on different clocks. Agencies with credit customers qualify. So do manufacturers buying stock ahead of a season and anyone carrying an instalment against equipment.
Skip it if you collect at the point of sale and have no credit customers and no buying cycle. A single salon or a small café settling by UPI daily has three fixed outflows and same day inflows. For that business a weekly look at the bank balance is the whole exercise and a ninety day grid is false precision dressed as control. Skip it also if you cannot currently state what you are owed and by whom. Build the receivables ledger first. A forecast assembled on guessed inputs produces confident numbers that are wrong and that is worse than no forecast.
How to build it
Use weeks as columns and not months. A month that closes positive can still contain a week where the balance goes below zero. Open with the combined balance of every current account less any cheque already issued and not yet presented.
Enter inflows by named source only. Each unpaid invoice goes in at the date that customer actually pays rather than the date printed on the invoice. If a buyer historically pays in fifty days then fifty days is the input. Then list outflows in order of how hard they are to move: statutory dues first, then payroll, then loan instalments and rent, then vendor bills, then discretionary spend. Closing balance rolls forward as the next week's opening balance.
Finally build a second version where every receivable shifts later by the average lateness you have observed. That pessimistic case is the one worth acting on.
How to tell whether the forecast is any good
The metric that matters is weekly accuracy on the closing balance. Set a tolerance band in advance and count how many of the past eight weeks landed inside it. Watch the direction of the error too. A forecast that is consistently optimistic is reporting a collections problem you have not accepted yet.
The metric that flatters is the ninety day total. Inflows comfortably exceeding outflows across the quarter says nothing about the third week of next month. A forecast you have never revised also flatters. It means you stopped comparing it against the bank statement.
The common mistake
The expensive error is entering receivables at the due date instead of observed behaviour. The forecast then shows comfort in precisely the weeks the business is thin. On the strength of that comfort an owner commits to an advance to a supplier or raises ad spend. Three weeks later the money has not arrived and short notice borrowing gets arranged at whatever rate is available that day. The cost is not the shortfall. It is the price of funding it without warning.
The second version of the same mistake is keeping one optimistic case and calling it the plan.
What to do with four weeks of notice
The forecast is only worth the actions it unlocks early. Ranked by cost those usually run as follows: chase and incentivise early collection on the largest two invoices, defer discretionary spend, renegotiate a vendor date in writing before the date passes, then arrange a facility. An overdraft limit is cheaper to negotiate in a month that looks healthy than in the week you need it.
Statutory dues are not a flexible line in that ranking and should never be used as working capital.
This guide explains how something works. It is not legal, tax or financial advice, and rules change. Confirm anything regulatory with a qualified professional before you act on it.