Cash problems are visible weeks before they arrive and get discovered days before. Closing that gap is the whole purpose of a forecast. Everything else about it is implementation detail.
The gap matters because your options collapse as it narrows. Eight weeks out you can accelerate an invoice, move a purchase, defer a hire or have a straightforward conversation with a customer about timing. Eight days out your options are borrowing at whatever rate you can get, paying someone late, or asking your team to wait.
Same problem. Completely different business, depending only on when you saw it coming.
So a cash forecast is not a report. It is an early-warning system, and it has one question to answer: in which week, if any, do I come up short — and how much notice does that give me?
Three questions that make it yours
Most advice on this hands you a template and a set of numbers. The numbers are the part that cannot be handed over. Answer three questions and the shape of your forecast falls out, and all three have different answers depending on how you run.
How far ahead do I need to see?
Thirteen weeks is the common answer and there is nothing magic about it. It is roughly how long it takes an owner-run business to actually change its cash position: collect faster, move a purchase, delay a hire, renegotiate a date, sell something.
So work out yours. Look as far ahead as it takes you to actually do something. If your customers pay on ninety-day terms, thirteen weeks barely covers one cycle and you need longer. If you take payment at the point of sale and your costs are weekly, eight may be plenty.
Far enough ahead that a problem shows up while you can still act on it. Near enough that you are guessing about work you can actually name. Past that, you are forecasting an idea of your business rather than its schedule.
How often do I need to look?
As often as your biggest recurring bill comes due. Run payroll weekly, forecast weekly. If your largest commitment is a monthly stock order, monthly may be enough.
The reason most service businesses need weekly is that a monthly view hides the problem. A month that comes out fine can still contain a week where payroll lands before three big payments do. Rent is monthly, payroll is weekly or fortnightly, customers pay when they pay. The trouble happens inside the month, so you have to look inside the month.
What am I watching for?
This is the one that matters most and varies most.
Which bill would sink you fastest if you missed it? For most service businesses it is payroll, because people leave and they leave immediately. If you buy stock before you sell it, it may be the stock order. If you carry a loan with conditions attached, it may be that.
Whatever yours is, that is what the forecast watches. Not your total balance. A total balance is one number doing several jobs, and it can look fine in a week where the money is already spoken for.
Then find your own number for it:
Look back over the last three months, find the worst single week of that bill, and add a bit for the weeks you know run heavier. That is what you are protecting.
For one business that is a few thousand dollars. For another it is sixty. It means nothing out of context, which is exactly why nobody else’s number helps you.
The one rule that trips everybody
Enter money by the week it arrives, not the week you earn it.
That is the whole conceptual difficulty. Everything after it is arithmetic.
If you invoice on completion with thirty-day terms, work delivered in week two shows up in week six or seven. The labor for it was paid in week two. The gap between those two dates is precisely what you are trying to see, and a forecast built on earned revenue rather than expected receipts will show you a healthy business right up until the week it does not.
Which means you need to know how long people actually take to pay you. Not your terms. What happens.
Take your last thirty or forty invoices and count the days from sending to being paid. Then take the middle one rather than the average. A single customer who took four months will drag an average upward and tell you your terms are worse than they are. If the middle number is thirty-eight days against thirty-day terms, forecast thirty-eight.
Your terms are what you wrote down. Your lag is what actually happens. Forecast the second one, and that alone will improve your forecast more than anything else you could change.
Four numbers you already have
All four are sitting in records you already keep. An afternoon gets you every one.
What share of the money goes to people. Add up last quarter’s wages, including the tax and insurance you pay on top of them, and divide by what came in. Once you know that share, revenue tells you payroll: if you expect a given week’s money and you know the share, you know roughly what wages follow and when, given your pay cycle.
What leaves no matter what. Rent, insurance, software, loan payments, salaried wages. This is your floor, and it tells you what a month with no revenue actually costs you.
What you take out before the business spends anything. More on this next.
Your busy and quiet months. If you have a season, last year’s monthly revenue as a share of the year gives you the pattern. It is rough. It is much better than pretending every month is the same.
Take it out at the start, not the end
Here is the idea that changes how the forecast behaves.
Whatever is left at the end is never anything. Profit you plan to keep once everything else is paid is not a plan, it is a hope, and there is never anything left over.
So take it out at the start. The moment money arrives, move profit and tax somewhere the business cannot spend them. Then build the forecast on what is left. Which means the ending balance is not money in minus money out. It is:
(Money in × the share you actually run on) − Money outSet those percentages yourself. How much to hold back for tax depends on your business structure, your income and where you are, and the only sensible source for that number is your accountant. How much to keep as profit depends on what you are building: a business built to pay you now and a business built to be worth more later should not use the same number.
The split is not the point. The point is that the business has to run on what is left, and if it cannot, that tells you something. A share you cannot operate on means you are not selling enough, not delivering efficiently, or not charging enough.
The part that confuses people: some months the operating balance sits flat and you conclude you made nothing. You did. It went where you sent it. Flat operating cash next to a growing profit balance is a business working properly.
Find out how wrong you usually are
Your first several forecasts will be wrong, and that is the useful part.
Every week, write down what you expected and what actually happened. After a couple of months you will know how far off you tend to be and in which direction. Almost everyone is wrong the same way: money arrives later than you thought it would.
Knowing how wrong you usually are is worth more than the forecast. A forecast you know runs about twelve percent optimistic is far more useful than one you argue with, because you can adjust for a bias you have measured and you cannot adjust for one you have not.
When a week comes up short
The levers, in the order I would use them:
- Accelerate money in. Invoice today rather than at month end, ask for a deposit, call your two largest outstanding receivables. Cheapest, fastest, most under-used.
- Move money out. Delay a discretionary purchase, shift a payment date, defer an owner draw.
- Draw on reserves, if this is what they are for.
- Draw on credit, if this is a timing gap whose end you can name.
If you reach the fourth lever regularly, you do not have a cash-timing problem. You have a pricing or profitability problem wearing a timing costume, and the forecast has just told you so. That distinction is worked through in profitable on paper, broke in the bank.
What it will not do
Worth stating, because a forecast can create false confidence.
It will not make your estimates true. The most common error is assuming money arrives sooner than it does, which is exactly why you check your guesses against what happened.
It will not fix profitability. It tells you when you run out. It has no opinion about why. If the underlying economics do not work, it will show you a deteriorating trend accurately and repeatedly.
It will not replace bookkeeping. It looks forward, and it is only as good as your understanding of what already happened.
Build yours
- Decide how far ahead to look — as far as it takes you to actually do something
- Decide how often to look — as often as your biggest recurring bill comes due
- Pick the bill that would sink you fastest, and find its worst week from the last three months
- Count how long people actually take to pay you — the middle number, not the average
- Work out what share goes to wages, and what leaves no matter what
- Decide what comes out before the business spends anything, with your accountant
- Every week, write down what you expected and what happened, until you know how wrong you usually are
Download the 13-week forecast template →
The template is one workable layout with example figures filled in so the structure is visible. Replace every one of those numbers with yours. They show the format. They are not recommendations.
It will take a month or two to get everything genuinely in order and running smoothly. That is normal. It is also the work you should be doing more of, and scheduling and quality control is the work you should be doing less of.
Frequently asked
Why thirteen weeks? Because a quarter is roughly the lead time on the actions available to most owner-run businesses, and it aligns with quarterly reporting. If your customers pay much more slowly than that, look further out. Your circumstances set this, not a convention.
Do I need separate bank accounts? Not to build the forecast. It works considerably better with them, because account balances then give you a real-time read for free. See the separate accounts system.
What if my revenue is genuinely unpredictable? Forecast conservatively and treat upside as variance, or forecast a range. Unpredictable revenue is an argument for a bigger cushion and a longer view, not against forecasting.
How accurate should this be? Early on, not very. Directionally accurate is enough to change a decision, and changing decisions is the entire job. It gets more precise as you learn your own bias, not by trying harder at the start.
Should my accountant build it? They can. Build the first one yourself anyway, because most of the value is in understanding how money actually moves through your business. Hand over maintenance afterwards if you want.
What if I am already in a crisis? Build it today. You need to know how many weeks you have before you can decide what to do, and the answer is frequently more than it feels like from inside the panic.
The forecast and the owner hourly-rate calculation are usually the first two things I build with an owner, because almost every other decision depends on them. Here is how I work.