How to forecast cash flow for a small business

What a cash flow forecast actually is

A cash flow forecast is a simple projection of the money moving in and out of your business over a future period. You start with the cash you have today, add the money you expect to receive, subtract the money you expect to pay out, and you get your projected balance for each week or month ahead. That projected balance is the whole point: it tells you in advance whether you can cover what's coming.

It helps to be clear about how this differs from your profit and loss statement. The profit and loss statement looks backward and tells you whether your business made money over a period that has already happened. A cash flow forecast looks forward and tells you whether you'll have money in the account on a date that hasn't arrived yet. One is a report card; the other is a weather forecast.

This is why a profitable business can still run out of cash. Imagine you invoice a client $20,000 in March for work you've finished. Your profit and loss statement records that $20,000 as revenue in March, so the month looks great. But if the client pays on net 30 terms, the actual cash doesn't land until late April. If your rent, payroll, and supplier bills all fall due in early April, you can be profitable on paper and short of cash in reality at the exact same moment. The forecast is what shows you that gap before you hit it, while there's still time to chase the invoice, delay a purchase, or arrange cover.

So the job of a forecast is not to predict the future perfectly. It's to give you enough warning. A forecast that flags a tight week three weeks out has done its job, because three weeks is enough time to do something about it.

The four inputs every forecast needs

Every cash flow forecast, no matter how simple or how fancy, is built from the same four inputs. Get these four right and the rest is arithmetic.

The first three are reasonably easy. Your starting balance is a fact you can read off your bank account. Your expected money in is mostly the invoices you've already sent that haven't been paid yet, plus any regular income you can count on. Your expected money out is your bills and your fixed monthly costs, which tend to be stable and predictable: rent, software, payroll, insurance.

The fourth input, timing, is where forecasts live or die, and it gets its own section next. The reason is simple. A forecast that knows you'll receive $20,000 but assumes it arrives a month earlier than it really will is worse than useless, because it will tell you everything is fine in exactly the week things are tight. Totals without timing are just a profit estimate wearing a forecast's clothes.

Timing is the assumption that makes or breaks it

The hardest part of forecasting cash flow isn't knowing the amounts. It's knowing when each amount actually moves. A bill dated the 1st might be due the 30th. An invoice you send today might not be paid for six weeks. The dates, not the dollars, decide whether your forecast matches reality.

Start with money coming in. The date that matters is not when you send the invoice or when the work is done. It's when the client's payment actually clears your account. If you bill on net 30 terms, the realistic assumption is that the money arrives 30 days after the invoice date at the earliest, and often later, because plenty of clients treat net 30 as a starting point rather than a deadline. If you have a client who reliably pays at day 45, forecast day 45 for them, not day 30. Honest timing beats optimistic timing every time.

This is also where slow payers quietly wreck an otherwise sensible forecast. If a large invoice is overdue, it should not sit in your forecast as money arriving this week just because you'd like it to. It belongs on the date you actually expect it, and chasing it is part of keeping the forecast real. Having a clear process to follow up on unpaid invoices does double duty here: it brings the cash in sooner and it tells you when to stop assuming an overdue invoice is about to land.

Now do the same for money going out, but with the opposite instinct. For bills you owe, forecast the latest reasonable date you'll pay, which is usually the due date, not the invoice date. You don't pay a bill the moment it arrives; you pay it when it's due. Stretching your own payables to their due dates (without going late) is one of the few levers you fully control, and a good forecast shows you exactly how much that helps in a tight week.

Put those two habits together, money in dated when it truly arrives and money out dated when it's truly due, and your forecast stops being a wish list and starts being a plan. Get the timing wrong in either direction and even perfect amounts will point you at the wrong week.

Short-term vs rolling forecasts

There are two ways to think about how far ahead to forecast, and the right answer for most small businesses is a bit of both.

The workhorse of cash flow forecasting is the 13-week rolling forecast. Thirteen weeks is one quarter, broken down week by week rather than smoothed into three monthly numbers. The weekly view matters because cash problems are weekly problems. A month can show a healthy net inflow while still containing one week where payroll lands before a big invoice clears. Monthly totals hide that week; weekly buckets expose it. Thirteen weeks is also a sweet spot for accuracy: it's far enough ahead to give you room to act, and near enough that your assumptions about who pays when are still grounded in reality.

The "rolling" part is what keeps it useful. Each week, you drop the week that just finished and add a new week at the far end, so you always have a fresh 13 weeks in view. You also update the near weeks with what actually happened: the invoice that got paid, the bill that came in higher than expected. A rolling forecast is a living document, not a thing you build once and file away. The first time you build it takes an hour; each weekly update takes a few minutes.

Beyond 13 weeks, you can forecast further, monthly out to six or twelve months, but be honest about what that buys you. The further out you go, the more your numbers depend on revenue you haven't won yet rather than invoices you've already sent, so the longer forecast is more of a planning aid than an early-warning system. Use the near-term weekly forecast to avoid running short, and the longer monthly one to think about hiring, big purchases, or seasonal swings. They answer different questions, and trying to make one forecast do both jobs is how you end up trusting it for the wrong thing.

Where doing it by hand falls apart

You can absolutely build a cash flow forecast by hand in a spreadsheet, and for a brand-new business with a handful of invoices, that's a perfectly good place to start. The trouble isn't building the first one. It's keeping it true.

Three things wear a manual forecast down. The first is stale data. A forecast is only as good as its last update, and a spreadsheet doesn't update itself. The week you're busiest is the week you skip the update, and a skipped update is exactly when the forecast quietly drifts away from your real bank balance. The second is the maintenance burden. Re-typing paid invoices, rolling the weeks forward, and adjusting dates every week is dull, and dull work gets dropped. The third, and the one almost nobody does by hand, is seasonality. Most small businesses have a rhythm, a slow January, a busy fourth quarter, a summer dip, but spotting that pattern in your own numbers and folding it into next year's forecast is genuinely hard to do manually.

This is the case for letting your accounting tool do the forecasting from data it already holds. Because your invoices and bills are already in the system, the forecast can be built from them automatically and refreshed every time something is paid, with no re-typing. That solves the stale-data and maintenance problems at a stroke.

Argo Books takes this a step further with its predictive analytics, which are part of the Premium tier (basic real-time analytics are free; the forecasting, trend detection, and confidence scoring are Premium). Instead of a fixed rule like "assume everyone pays in 30 days," it applies real time-series techniques, Holt-Winters triple exponential smoothing and Singular Spectrum Analysis, to your own history. In plain English, those methods separate your numbers into three parts: the underlying level (roughly where your cash sits), the trend (whether it's drifting up or down over time), and the seasonal pattern (the repeating yearly, half-yearly, or quarterly rhythm). It tests for those cycles automatically, so a business with a strong fourth quarter gets a forecast that expects the fourth quarter, rather than a flat line that's wrong every December.

Two things make it practical rather than just clever. First, it shows a confidence score from 0 to 100 next to each forecast, built from how much data you have, how stable that data is, how strong the seasonal pattern is, and how accurate its own past forecasts turned out to be. A low score is a feature, not a flaw: it tells you to treat the number as a rough guide rather than a promise. It tracks that accuracy by backtesting itself against your real history, and averages around 88% accuracy in backtesting. Second, it all runs locally on your own computer using ML.NET, so your financial data stays on your machine and is not sent to a cloud service. There's nothing to set up; the forecasts and plain-language insights appear from the data you already have.

None of this means a spreadsheet is wrong for everyone. If your business is small and steady, a simple sheet you update every Friday is fine. But the moment forecasting starts eating real time each week, or you suspect there's a seasonal pattern you can't quite pin down, automatic forecasting earns its place. You can download Argo Books and try it on your own numbers to see what the forecast and confidence score look like for your business.