How Do I Forecast Cash Flow for a Small Business?

Build a 13-week cash forecast. Start from your bank balance, not your profit. Lay out every dollar landing and leaving, week by week, for the next 13 weeks: money in when invoices actually pay, money out for payroll, rent, taxes, principal, and draws. The running balance is the forecast, and its low point is the week that can bounce payroll. Profit and cash aren't the same number, which is why you start from the bank, not the P&L.

Start with a 13-week forecast, not a yearly budget

A budget tells you what a good year looks like. Cash forecasting answers something narrower: whether you can make payroll on the 15th. Thirteen weeks is the sweet spot, long enough to see a tax payment or a slow month coming, short enough that you can actually guess when each invoice pays. Push the window out to a year and it's fiction by March. Keep it weekly and you can trust it.

Monthly forecasting hides the thing that kills you. Rent, payroll, and a tax deposit can all land in the same week while your biggest client pays the following Monday afternoon. On a monthly view that month nets out fine. In real life you bounced payroll on Thursday. Cash problems live inside the month, so your forecast has to live there too.

How do you build a 13-week cash forecast?

One spreadsheet, thirteen columns across the top, two blocks of rows down the side. The whole build is six steps.

1. Start from today's bank balance. Not net income, not the budget. The actual cleared number in the account this morning. Everything else builds off that one true starting point.
2. List money in, by the week it actually lands. Walk your open invoices and drop each one into the week you really expect payment, not the week you sent it. If a client pays in 45 days, the invoice you cut last week lands in week five, not week one.
3. List money out, by the week it leaves. Payroll on its real dates, rent, software, loan payments with the principal included, estimated taxes, owner draws. Put the bills that don't wait at the top.
4. Run the balance forward. Each week is start balance, plus money in, minus money out, equals ending balance, which becomes next week's start. Thirteen rows, one running line.
5. Find the low point. Scan the ending-balance column for the smallest figure. That week is your real constraint, and it's almost never this week.
6. Roll it every week. Drop the week that passed, add a fresh week 13 at the back, swap estimates for actuals. Built once, it's a document. Rolled every Friday, it's a system.

The two row blocks are simpler than they sound:

Two of those lines trip people up. If you're a pass-through (S-corp, LLC, or sole proprietor), your income tax isn't a company bill. It leaves as an owner draw sized to cover your personal estimate, so park it there and skip any separate "business taxes" row, or you'll count the same dollars twice. And sales tax you collect was never your money: it rode in when the customer paid and rides back out on the remittance date, so net it, don't book it as revenue. Many service firms sell nothing taxable and drop that row altogether.

The whole model rests on one number you have to earn: your real days-to-pay. Pull your last 20 to 30 paid invoices, and for each one subtract the invoice date from the day the cash actually cleared the bank. Average those, per client if a few big accounts drive your revenue. That average is the timing your inflows ride on. If your books are too messy to show clearing dates, fix that first, because every dollar of timing rides on it.

In the service businesses I clean up, the gap between the invoice date and the deposit hitting the bank runs 30 to 60 days. That's a practitioner range, not a published stat. But it's the whole reason the money-in column almost never lines up with the month you booked the sale. Time the inflows wrong and the forecast is worse than useless. It lies to you with a straight face.

What a filled-in forecast actually looks like

Recipes are easy to nod at and hard to picture. Below is a six-week slice of one I'd actually build, with the numbers rounded and the shape kept exactly as it lands over and over in real service firms.

Read down the ending column. Week 4 dips to negative $21K, and that's the number the whole exercise exists to surface. The $60K client payment that pulls you back to positive doesn't land until week 5. So the forecast isn't telling you you're broke. It's telling you that you have three weeks to cover a two-week hole, which is a problem you solve on a quiet Tuesday instead of discovering on a payroll Friday.

Where do first forecasts go wrong?

Three mistakes, almost every time.

• Booking money in when you invoice, not when you collect. This is the big one. Your forecast says the cash is here in week two. Your client's habit says week seven. Time every inflow to when it clears, and if you don't know, use your real days-to-pay history, not your payment terms. Terms are a wish. History is data.
• Forgetting the invisible outflows. Loan principal and owner draws never show on your P&L, so a forecast built off the income statement leaves them out, and then the account comes up lighter than the model every single week. Those are real cash gone. Profit and cash aren't the same number, and a projection built on profit inherits every blind spot in that gap.
• Making it too precise. You're not predicting to the dollar. You're trying to spot the week you go tight. Round to the hundred, get the timing roughly right, and move on. The first build takes an afternoon, but after that a rolling update that eats three tedious hours never gets done, and a forecast nobody updates is just a screenshot of a moment that already passed.

You found the low point. Now what?

A negative week isn't a verdict. It's a to-do list with weeks of lead time. You have four levers, roughly least painful to most: pull collections forward (call the client whose $60K lands in week 5 and ask for half now), delay a discretionary outflow (your own draw, a software renewal, a hire you were about to make), lean on a line of credit to bridge the two-week trough, or stretch a vendor you have the relationship to stretch. The forecast's whole job is to hand you that choice while it's still a choice, not after the payroll run fails.

How tight is too tight? The rule of thumb I use: if your lowest week doesn't hold at least one full payroll cycle in the account, you're running without a cushion, and one late client turns a tight week into a missed one. Two to four weeks of operating expenses at the low point is real breathing room. On a firm burning $40K a week, that puts the floor somewhere around $80K to $160K, never zero. That's a practitioner benchmark, not a formula, but it's the line I watch.

How far out can you actually trust it?

Your near weeks are firmer than your far weeks, but don't oversell them. Payroll, rent, a scheduled tax draw: those you know cold. The money coming in is the loose part even up close, because clients pay on habit, not on your terms. Everything from week five on gets looser the further out it sits. Near weeks catch the emergency. Far weeks catch the trend. Week 13 doesn't need to be accurate. It needs to warn you.

Open invoices only fill the first few weeks, though. Two-thirds of a 13-week window is work you haven't sold or invoiced yet, so you estimate it: take a normal month of collections, strip out anything unusual, and drop a conservative weekly run-rate into weeks five through thirteen. Under-book it on purpose. A far week that comes in high is a happy surprise. One that comes in low is a bounced payroll. Replace the estimate with real invoices as the weeks approach, and the guess firms up on its own.

That decay is exactly why you roll it. Every week that passes, week five slides into week four and firms up. A cash forecast is never finished and was never meant to be. Pair it with your monthly financial reports and you get both eyes at once: the reports say whether the business is healthy, the forecast says whether it clears the next 90 days. One looks back. One looks forward. Keep both open on the same desk.

The bottom line: a forecast is a decision tool, not a crystal ball

Predicting the future isn't the job. The job is answering one question: can I hire, take the draw, or make the tax payment without bouncing payroll three weeks from now? The first build takes an afternoon, and after that rolling it forward runs about an hour a week. The point is never the number sitting in week 13. It's the decision you make today because you saw week 13 coming.

And if you can't tell when last month's cash actually cleared, your forecast is built on sand no matter how neat the spreadsheet looks. Clean books first, or you're forecasting fiction.