13-week cash flow: what it can tell a client, and what it cannot

A 13-week cash flow forecast is a weekly projection of cash receipts and payments over the next quarter, built from what is actually expected to move through the bank rather than from the income statement. It is the standard short-horizon tool in corporate finance, and most of what is written about it is written for a company running it on itself.

CPA and fractional CFO firms run it differently. The firm does not work inside the business. It runs the forecast for many clients at once, from the outside, on information that arrives in calls and emails as often as in the ledger. That changes what the forecast can be trusted to say, and it is the part the standard guides leave out.

What it is, and why thirteen weeks

The forecast has one column per week. Receipts sit on top: customer payments expected by week, based on open invoices and how each customer actually pays. Disbursements sit below: payroll, rent, vendor bills, loan payments, tax, each placed in the week it will clear. The bottom line is ending cash, week by week, for a quarter.

It uses the direct method. It does not start from net income and adjust. It starts from the bank balance and asks what will arrive and what will leave. That is why it can answer the question a client actually asks, which is whether the money is there in the week payroll clears, when a profit-and-loss view cannot. Where an obligation falls on a specific date and the week is tight, the week’s ending balance can hide a shortfall earlier in that week, and the forecast needs daily detail for that week.

Thirteen weeks is a quarter, and a quarter is the horizon over which near-term weeks are precise enough to manage payment timing and the far weeks are still close enough to act on. A shortfall visible in week ten can be chased, financed or deferred. Longer horizons serve planning questions; the 13-week view serves timing.

Rolling means the horizon stays fixed. Each week that passes is dropped and a new one is added at the far end. That is a property of how far forward the forecast looks, not of what the business has committed to. The budget carries the commitments; the 13-week view carries the timing. The distinction is covered in budget vs forecast: what the difference is for.

What goes in it

Each week has the same shape.

Opening cash. Cash available at the forecast’s starting date, with restricted funds excluded and outstanding payments accounted for once. Future commitments appear in the week’s payment rows.

Receipts by week. Open invoices placed in the week each is expected to be paid, based on how that customer actually pays rather than the due date on the invoice. Plus any other inflows: a loan draw, a capital injection, a tax refund.

Payments by week. Payroll, rent, vendor bills, loan payments, tax instalments and owner distributions, each placed in the week it clears the bank.

Ending cash. Opening cash plus receipts less payments. It becomes the next week’s opening balance.

Around the columns sit three practices that make the forecast worth keeping. A list of the assumptions behind the numbers: which customer is expected to pay late, which bill has been deferred, which hire is pencilled in and for when. A weekly review of what the forecast said last week against what actually cleared, so the assumptions can be corrected. And the roll: each week that closes is dropped, a new week thirteen is added, and the review feeds the new column.

A short example, for a service business with a tight week four:

Week three includes a 50,000 payment from one customer within its 61,000 of receipts. If that payment slips to week five, week four opens at 46,500 and closes at 10,000. The forecast makes that dependency visible before the payment is missed.

Where the standard version stops

Nearly every published guide to the 13-week forecast assumes a finance team sitting inside the company. They know when the big customer said it would pay, because they were on the call. They know the hire was pushed to November, because they made the decision. The forecast is a spreadsheet, and the knowledge that keeps it honest is in the room.

A firm running the forecast for a client is not in the room. It is in a different room, with twenty other clients, and it learns what changed on a monthly call or in an email that arrives between calls. The mechanics of the forecast are the same. Three things about it are not.

A data refresh does not capture a client conversation

The recorded baseline of a 13-week forecast comes from the books and the bank: an invoice was issued, a bill was entered, a payment cleared. Forward-looking assumptions sit alongside it, but refreshing the baseline does not automatically update those assumptions. Automated tools have made the baseline reliable. The model can rebuild every night from live feeds, and the spreadsheet that used to take a day to update now updates itself.

What the rebuild does not include, unless someone puts it there, is what was said. A client mentions on a call that their largest customer asked to pay the March invoice at the end of April. The landlord agreed to defer half of next month’s rent. The Henderson contract was lost on Friday. None of these is a transaction yet. The receivable still shows its original due date. The rent is still scheduled at full. The revenue run-rate still includes a customer who has given notice.

The forecast, rebuilt that night, reflects none of it. It looks exactly like a forecast that is complete, and nothing in it indicates a gap, because from the model’s point of view there is none. For one client, the partner holds the missing items in their head and corrects by eye. Across a portfolio, that correction is applied to the client they spoke with yesterday and not to the one they spoke with three weeks ago.

This is the forecast blind spot. It survives automation because refresh and capture are different acts. What your forecast cannot know covers why the fix is not more discipline, and what has to change instead.

Rolling forward does not automatically preserve a track record

A forecast exists so a decision can be made now about something that has not happened yet. The decision gets made. The month ends. The forecast rolls forward, absorbs the actuals, and unless the earlier version was deliberately kept, the picture that existed on the day the client decided to hire is gone.

Budget-to-actual comparison is standard in every firm, but it explains performance against a plan. It does not compare the forecast, as it stood on a specific date, with what subsequently happened. A rolling forecast that overwrites itself cannot be scored, because by the time the outcome is known, the estimate has already been updated with it. Practices that keep weekly forecast-to-actual variance do exist; the question is whether that discipline holds across a client portfolio when nobody is paid to run it.

So the customer who paid a month late every quarter this year is, in next quarter’s forecast, expected to pay on time. The assumption travels forward unchanged, with the same confidence as everything else in the model, and the client hires against headroom that a calibrated forecast would not have shown.

The cost is not a wrong number. It is a decision the client cannot take back. What your forecast learns from being wrong covers what scoring a forecast actually requires, and why the errors tend to run in a direction nobody has measured.

An ending balance does not explain the assumptions behind it

Accuracy is a property of the forecast. Whether it can be leaned on is a relationship between the forecast and the size of the decision resting on it. The same 13-week view may support a small, reversible spending decision and still be dangerous for two permanent salaries.

Before a client acts, the adviser needs to know three things. Which obligations in the next quarter depend on a single event landing on time, tested by that counterparty’s own typical delay rather than a round number. Which parts of the picture are recorded facts and which are assumptions layered on top, and how old each assumption is. And what the forecast’s track record says about how far off it tends to run, and in which direction.

Most firms can answer the first two by hand, on the call, for the clients they see most. The third requires a preserved forecast history. Without it, forecast error is unmeasured, and the adviser has to assess the decision on available headroom, the downside if the key receipts slip, and how reversible the commitment is.

How to tell whether a cash forecast will bear weight walks through the three tests in the order an adviser would run them, and how to tell a client the picture will not hold without it sounding like the adviser failing.

What this means for the firm

A 13-week forecast is not hard to build. Any of the templates on the first page of a search will do it, and a connected tool will keep the arithmetic current without anyone touching it.

What takes ongoing work, from outside the client’s business and across many clients, is capturing what the client said, preserving what the forecast predicted so it can be checked, and being able to say what the current picture rests on before the client acts. None of that is beyond a well-run process. All of it is work a refreshed feed does not do on its own, and it multiplies by the number of clients. That is where the value of the engagement lives, and where the standard guides stop.

The gap is in the structure, not the firm. A partner who knows every client’s cash by heart is doing the work a system should hold, and the limit of that approach is the number of clients one person can hold in their head.

Where AgentLink fits

AgentLink keeps a rolling 13-week cash forecast current for every client, read from the bank and the ledger daily. Forward-looking information that surfaces in a call, a message or a question is captured as an explicit assumption with its source, held in an overlay on the recorded baseline, and kept open until actual activity confirms or contradicts it. Each assumption is preserved with its forecast date, so when it resolves, the predicted timing and amount can be compared with what happened, and the counterparty’s behaviour profile updates for the next forecast.

Whether a client should make the hire, take the distribution or chase the late payment remains the firm’s judgement. The software keeps the forecast honest about what it knows and what it has been told.

The forecast moment describes how this fits the engagement.

Frequently asked questions

What is the difference between a 13-week cash flow forecast and a cash flow statement?

A cash flow statement is historical reporting: it shows cash movements for a period that has closed, and operating cash flows can be presented by either the direct or the indirect method. A 13-week forecast is a forward projection, week by week, of expected receipts and payments. One is a record of what happened. The other is an estimate that has to be maintained.

What is a rolling 13-week forecast, and how is it different from a static one?

A static forecast covers a fixed calendar period and expires when it ends. A rolling forecast drops each completed week and adds a new one at the far end, so it always looks the same distance ahead. Rolling describes the horizon. It does not by itself mean the assumptions inside the forecast are current.

How often should a 13-week cash flow forecast be updated?

The recorded activity should be updated whenever the bank and the books change, which with a connected tool means daily. The assumptions are a separate question. Each one is as old as the day someone last confirmed it, and a nightly rebuild does not touch them.

Is a spreadsheet template enough?

For building the forecast, yes. A template handles the arithmetic and the layout. What it does not handle is capturing what the client said between calls, preserving what the forecast predicted so it can be scored, or showing which assumptions the current picture rests on. Those are the parts that determine whether the forecast can be relied on, and they are the parts a firm has to solve separately.

Should the forecast be compared with actuals?

Yes, and more precisely than most firms do it. Comparing the current forecast with actuals it has already absorbed tells you nothing. The useful comparison preserves the forecast as it stood on a date, scores timing error and amount error separately, and feeds the verified outcome into the next forecast without erasing the record.

Which clients need a 13-week forecast?

Any client whose obligations could fail on a date: payroll, rent, a loan payment, a tax instalment. Revenue size matters less than timing risk. A seasonal business at eight hundred thousand with a sixty-day collection cycle needs it more than a steady business at three million that collects on receipt.

AgentLink builds controller-layer software sold through CPA and fractional CFO firms. The practitioner observations in the linked guides come from design-partner conversations and are used with permission where attributed. Corrections to nilanjan@agentlink.finance.