13-Week Cash-Flow Forecast: Built in Two Weeks, Run Every Monday

A 13-week forecast is not a budget and it is not a projected income statement. Revenue on the P&L is not cash, and a profitable month can end with less money than it started with. The forecast tracks cash only: what customers will pay and when, what the company will pay and when, in weekly columns, 13 of them, because a quarter is long enough to see a problem coming and short enough that the estimates still mean something.

Each column opens with the prior week's closing balance and closes with the new one. Receipts are listed by customer, because a forecast that says receipts of $600,000 is a guess and one that says which 14 customers will pay is a plan. Disbursements are grouped by category: payroll and taxes, subcontractors or cost of goods, rent and occupancy, debt service, insurance, and the discretionary items that get pushed when a week is thin. Line availability sits underneath the closing balance, so the real question, how much room is left, is answered on every line.

The thirteenth week is not the point, and nobody should pretend to know it precisely. The point is week three and week seven, where a shortfall is visible while there is still time to accelerate a receivable, delay a purchase, or talk to the bank on your terms rather than the bank's. A forecast that only confirms this Friday is fine has not told you anything the bank balance could not.

Week one is data. We pull the bank activity, the receivable aging with expected pay dates by customer rather than by due date, the payable aging by vendor and category, the payroll calendar, the debt and lease schedules, the tax deposit dates, and any fixed commitments such as insurance premiums or equipment payments. We interview the people who know which customers pay late and which vendors will wait. Most of the forecast already exists in the company's memory. We write it down.

Week two is the model. Receipts by customer, disbursements by category, weekly buckets, opening and closing cash, line availability, and a variance column that will compare each week's forecast to what happened. The first version is reviewed with the owner and the controller line by line, and the assumptions are documented next to the numbers. The forecast leaves week two as a working instrument, not a draft, and it is issued on the first Monday afterward.

• Reconciled opening cash and line balance for week one
• Receipts by customer with expected dates, not invoice due dates
• Disbursements by category with payroll, taxes, and debt fixed first
• Weekly closing cash and line availability for 13 weeks
• Assumptions page and a variance column from the second week on

Every Monday the forecast rolls forward: last week drops off, a new week 13 is added, actual receipts and disbursements replace the estimates for the week just finished, and the variance is explained in one line per item. The customer who was supposed to pay $85,000 and did not gets a name and a phone call. The vendor payment that ran $20,000 over gets a reason. Then the meeting: 30 minutes, owner, controller, and whoever runs collections, with three questions. What changed, what is tight, and what are we doing about it.

Accuracy comes from the variance, not from the model. A forecast that is off by 15 percent in week one and 4 percent by week six is doing its job, because every miss teaches the model something about how the company's cash actually behaves. The Monday meeting is where the teaching happens. Skip four Mondays and the forecast is a spreadsheet again.

The meeting has one more rule: nothing goes into the forecast that is not tied to a customer, a vendor, or a date. Hoped-for receipts from a proposal that has not been signed do not belong in week four. A vendor payment the company intends to delay is moved, not deleted. The discipline sounds fussy until the first month the forecast is right and the owner realizes the bank balance has stopped being a surprise.

A forecast that is never compared to what happened is a hope with columns.

A bank that receives a 13-week forecast every week from a borrower it is worried about behaves differently from a bank that receives silence. The forecast shows that management knows where cash is going, and the variance history shows whether management can predict it. In a forbearance or an amendment, the forecast is usually the document the lender's terms are written around: minimum liquidity by week, permitted variances, and reporting dates. A forecast that has been kept for two months before the lender asks is worth a great deal more than one built the night before.

Sureties read the same instrument for a different reason. A contractor's forecast built job by job, with pay application receipts, retainage releases, and subcontractor payments tied to those receipts, tells the bonding agent whether the company can carry the next project's front-end cost without draining working capital. We build the forecast in the format the reader needs, and we keep the assumptions visible so it can be questioned rather than merely believed.

The reporting itself is the smaller part. The larger part is that a company producing a weekly forecast has a different conversation with its lender: about the week that is tight and what the company is doing about it, rather than about the certificate that was late. We have sat on the borrower's side of that table with a forecast in hand and without one. With is better, and the difference shows up in the terms.

After six or eight Mondays, something changes. Collections calls happen on Tuesday because the forecast said which customers to call. Purchases get scheduled into the weeks that can afford them. The owner stops checking the bank balance at 6 a.m. because the Monday number has been close for two months. The line of credit is drawn on purpose and paid down on schedule instead of drifting toward its limit. Decisions about hiring, equipment, and taking on a large job get tested in the model before they are made.

At that point the forecast is no longer a project. It is the company's operating rhythm, and it becomes the front end of everything else: the monthly close review reconciles to it, the annual plan starts from it, and the lender package is a printout of it. Companies that reach this stage rarely go back, for the same reason nobody returns to driving without a fuel gauge.

Some companies have a controller or office manager who can run the Monday process once it is built and the procedure is written. For them we build the model, run the first four to six weeks together, document every step with screenshots and a checklist, and hand it off. We check the variance report monthly for a quarter afterward and step in if it drifts. This is the defined build project, priced as a fixed scope.

Other companies do not have that person, or have one who is already at capacity, or are in a situation where the lender wants an outside party on the forecast. For them we build it and keep running it: the Monday update, the variance, the meeting, the lender copy. That is the run option, from $3,500 a month, and it is often the first piece of a fractional CFO engagement. Either way, the model, the assumptions, and the procedure belong to you.

A word on which to choose. If the company is stable and the controller has the time and the temperament to chase a variance on Monday morning, hand off. If the company is in a lender situation, in a growth spurt that is straining working capital, or has an owner who wants a second set of eyes on cash every week, run. Most companies that start with run hand off within a year, which is the outcome we plan for from the first Monday.