A 13-week cash flow forecast projects cash in and cash out, week by week, across one quarter, built from the receipts and payments a business expects rather than from accrual net income. The first build is the easy part. What follows is a weekly production job with a fixed delivery day, a variance report attached to it, and a block of payment dates set by regulation rather than by the client. Scope the engagement against the weeks that follow, not against the first build.
What a 13-Week Cash Flow Forecast Is
A 13-week cash flow forecast is a weekly schedule of expected cash receipts and cash payments, starting from a known bank balance and ending at a projected closing balance for each of the next 13 weeks. It uses the direct method, which builds cash from the money expected to move in and out rather than starting at net income and adjusting for non-cash items. Rolling means the grid travels with the calendar: when a week closes it drops off the front, a new week goes on the back, and the horizon stays the same length every time the schedule is issued.
It is not the statement of cash flows in a year-end package. That statement explains a period that has already closed and ties back to the books. A weekly forecast is a claim about the future, issued before anything has happened, and it gets judged on whether the closing balance it predicted turned out to be the one in the bank.
The audience changes with it. A closing package goes to an owner who wants to know how the year went. A weekly cash schedule is often built for someone outside the business, and where a covenant requires it, that someone is the lender deciding whether to keep lending. That reader checks last week's promise before reading this week's.
Why the Horizon Is 13 Weeks
The horizon is 13 weeks because that is a quarter measured in whole weeks, so the grid lines up with a reporting period without cutting one in half. That is the textbook answer. The working answer is that the number can arrive already written into a loan document, in which case the specification exists before the client asks.
Nortech Systems amended its credit agreement in 2025 and restated the reporting clause to require, no later than 5:00 pm on the first Business Day of each week, a 13-week cash flow forecast in form and substance satisfactory to the lender, covering the 13-week period beginning with that week (Nortech Systems credit agreement amendment). The deadline sits in the covenant, and it does not move because the firm had a heavy week.
A loan agreement amendment filed by bluebird bio the same year went further and named the second deliverable. It required, weekly and delivered on Wednesday each week, a rolling 13-week cash flow forecast together with a comparison report setting out any variance between the cash disbursements actually made and the cash receipts actually received and the projections in the forecast delivered for the preceding period, certified by the chief financial officer, an officer with similar duties, or another duly authorized officer of the borrower acceptable to the agent (bluebird bio loan agreement amendment).
The forecast is only half of that clause. The comparison against what was promised the week before is the other half, and it is the half the certification travels with.
No accounting standard fixes the horizon at 13 weeks. It is a market convention that hardened into contract language, and that has one practical consequence for a firm: ask for the loan document before designing a template, because the client has already agreed to a format, a delivery day and a deadline.
What the Weekly Grid Contains
The grid is built from eight blocks, and they hold across industries.
| Block | What sits in it |
|---|---|
| Opening cash | The bank balance the week starts on, tied to the prior week's closing figure |
| Operating receipts | Collections against open invoices, dated by expected payment rather than by invoice date |
| Other receipts | Deposits that are not sales: refunds, insurance proceeds, owner contributions, draws on a line of credit |
| Payroll | Net pay leaving on each payday in the window |
| Payroll taxes | The deposits that follow payroll, placed on their own statutory dates |
| Trade payments | Vendor runs, grouped by the day the client actually releases them rather than by due date |
| Fixed obligations | Rent, debt service, insurance, equipment leases |
| Net movement and closing cash | The week's change, the balance it leaves, and headroom against any minimum the lender requires |
Operating receipts are where accuracy is won or lost, because collections are not the client's decision to make. Most payment lines are timed by someone at the business, and the statutory ones are timed by rule. Collections are a decision somebody else's accounts payable clerk makes, on their own timetable, which is why collections get built off the aged receivables ledger and each customer's actual payment history rather than off the terms on the invoice.
The Order to Build It In
Sequence matters, because each step constrains the one after it.
- Fix the opening balance to a reconciled bank position, not a book balance.
- Lay in the fixed obligations first, since they are certain in both amount and timing.
- Place the statutory payments on their own dates before anything discretionary.
- Date the receipts by when each customer is likely to pay, invoice by invoice for the largest ones.
- Close each week, carry the balance forward, and mark the weeks that breach any minimum.
- Lock the version and name it, so next week's variance has something to compare against.
Step six is the easiest one to skip and the one the engagement rests on. A forecast that gets quietly overwritten each week can never be shown to be wrong, and it can never be shown to be right either.
The Payment Dates You Cannot Move
Most of the disbursement side is a judgment call about when the client chooses to pay. Employment tax deposits are not. The date is set by regulation, and the balance in the account never moves it. Three things do: weekends, legal holidays, and one accumulation threshold.
An employer is a monthly depositor for the entire calendar year if the employment taxes reported for its lookback period were $50,000 or less, and a semi-weekly depositor if that total exceeded $50,000 (26 CFR 31.6302-1(b)). For an employer filing Form 941, the quarterly federal employment tax return, the lookback period is the twelve month period ended the preceding June 30, so the schedule a client is on this year was fixed by a window that closed last summer, and current cash pressure does not move it.
One event does. An employer ceases to be a monthly depositor on the first day after it becomes subject to the One-Day ($100,000) rule, and immediately becomes a semi-weekly depositor for the remainder of the calendar year and for the following calendar year (26 CFR 31.6302-1(b)).
The two schedules land in different weeks. A monthly depositor deposits the taxes accumulated during a calendar month by the 15th day of the following month. A semi-weekly depositor deposits taxes for Wednesday, Thursday and/or Friday paydays on or before the following Wednesday, and taxes for Saturday, Sunday, Monday and/or Tuesday paydays on or before the following Friday (26 CFR 31.6302-1(c)).
One rule overrides both schedules. If an employer accumulates $100,000 or more of employment taxes on any day within a deposit period, those taxes must be deposited by electronic funds transfer in time to satisfy the obligation by the close of the next day (26 CFR 31.6302-1(c)(3)). A client with a large commission or bonus run can cross that line without anyone noticing until the money is already gone.
Holidays move the date, and they move it into a different week. Where a monthly deposit date falls on a Saturday, a Sunday or a legal holiday in the District of Columbia, the deposit is timely if it is made on the next day that is none of those. A semi-weekly depositor gets an additional day for each legal holiday among the three weekdays following the close of the period, so a Wednesday to Friday period whose following Monday is Memorial Day is not due until the Thursday rather than the Wednesday (26 CFR 31.6302-1(c)).
Why This Breaks a Weekly Grid
In a weekly grid the deposit schedule is the difference between one payroll line and two lines in two different weeks. A client on the semi-weekly schedule with a Friday payday sees net pay leave on Friday and the deposit leave the following Wednesday. Bucket both on payday and the payday week shows a squeeze that will not happen, while the week the deposit actually clears shows no outflow at all. The grid is wrong in both weeks, in opposite directions, and only the closing balance two weeks out agrees with the bank.
That error is invisible in a monthly view and decisive in a weekly one. It is also why a template built for one client may not transfer to the second. The deposit schedule, the payday pattern and the holiday calendar are client-specific inputs, not formatting.
Which Standard Your Firm's Name Sits Under
A weekly cash schedule is prospective financial information, and that puts it inside a defined set of engagements rather than outside the standards altogether.
AT-C section 305 defines prospective financial information as any financial information about the future, and says it can take the form of prospective financial statements or partial presentations, where a partial presentation is one that excludes one or more of the applicable items required for prospective financial statements (AICPA, AT-C section 305). A cash-only weekly grid normally lands in the second category.
The same section splits the work by what the assumptions are. A financial forecast presents an entity's expected financial position, results of operations and cash flows, to the best of the responsible party's knowledge and belief, based on the conditions that party expects to exist and the course of action it expects to take. A financial projection presents the same three things given one or more hypothetical assumptions, the what-would-happen-if version (AICPA, AT-C section 305). The responsible party is the client, whose assumptions the schedule rests on.
The distinction has teeth where the attestation standards reach. AT-C section 305 covers a practitioner examining or performing agreed-upon procedures on prospective financial information, and among its preconditions for an examination it says that because a financial projection is not appropriate for general use, a practitioner should not agree to the use of the practitioner's name in conjunction with a financial projection that the practitioner believes will be distributed to those who will not be negotiating directly with the responsible party (AICPA, AT-C section 305). Model the refinancing that has not closed yet, and the output is a projection with a restricted audience.
The Four Engagements, and the One That Does Not Exist
Four engagements are available for this material and one is not. A firm can be engaged to prepare it, which AR-C section 70 covers when an accountant in public practice is engaged to prepare prospective financial information and is not engaged to compile or examine it. It can be engaged to compile it, since AR-C section 80 applies to compilations of prospective financial information as well as of financial statements. It can be engaged to examine it or to apply agreed-upon procedures to it under the attestation standards. It cannot review it: AT-C section 210, Review Engagements, states that a practitioner should not perform a review of prospective financial information (AICPA, AR-C section 70 and section 80; AICPA, AT-C section 305).
That last point catches firms out. The ladder for historical statements runs audit, review, compilation, and the middle rung is the familiar compromise. On a forecast the middle rung does not exist, so a client asking for something more than preparation is asking for a compilation or an examination, with everything each of those carries.
The assumptions have to travel with the numbers, and preparation and compilation each say so in their own words. On a preparation engagement, AR-C section 70 says at .19 that the summary of significant assumptions is essential to the user's understanding of prospective financial information, so the accountant should not prepare prospective financial information that excludes disclosure of that summary, and should not prepare a financial projection that excludes either an identification of the hypothetical assumptions or a description of the limitations on the usefulness of the presentation. On a compilation, AR-C section 80 sets the matching bar at .24 for the report, and at .25 requires that report to state that the forecasted or projected results may not be achieved and that the accountant assumes no responsibility to update it for events and circumstances occurring after its date (AICPA, AR-C sections 70 and 80). A grid of numbers with no assumptions page attached does not clear either bar.
There is also a rung below all four. AR-C section 70 does not reach merely assisting a client in preparing the schedule, which the standard treats as a bookkeeping service (AICPA, AR-C section 70). Which side of that line an engagement falls on is a judgment the firm makes when it accepts the work and writes the engagement letter, not months later.
Professional judgment made once, in April, under pressure, is not a record. Make it early instead: name the service, name who the responsible party is, and put in writing whether the schedule may travel beyond the negotiation it was built for.
Where the Hours Actually Go
The build is a one-time effort. The week is the recurring one, and it repeats on a fixed day whether or not that week is convenient.
Refresh the actuals. Pull the closed bank week, agree it to the reconciliation, and post the real opening balance. A schedule built on an unreconciled feed inherits every error in it, which is why a clean month-end close sits underneath this work rather than beside it.
Re-age the receivables. Every invoice that did not pay when it was expected moves, and moving it changes the closing balance of every week after it.
Re-cut the disbursement calendar. Payroll dates, deposit dates, debt service and any tax remittance get placed again for the new window, since a rolling forecast picks up a new back week each time.
Write the variance. Separate a timing difference from a permanent one. A customer who paid in week three instead of week two is noise; a customer who is not going to pay is a number that has to come out of the model and stay out.
Re-forecast the tail. Only the weeks the variance actually touched, and only for a reason recorded next to the change.
The objection worth taking seriously is that a competent controller-level person can do all of this quickly. Grant it, and the problem is unchanged. Whatever it takes arrives every week, on the same day, including the weeks when the same person is holding an extension deadline. This is a capacity question long before it is a technical one, and it belongs in capacity planning rather than in a partner's evenings.
When a 13-Week Forecast Is the Wrong Tool
A 13-week forecast answers whether a business can pay what falls due in the next quarter. It does not answer whether the business works.
For anything on a longer decision horizon, such as hiring plans, capital spending or a full-year budget, a monthly forecast running twelve months or more is the right instrument, and the two are not substitutes. Build a weekly grid when the real question is annual, and the client ends up with a precise view of a short runway and no answer to the question they asked.
Skip it, or run it monthly, when there is no covenant requiring it and the client's balance comfortably covers a quarter of obligations. Weekly granularity buys nothing there and costs a recurring deadline.
Fix the ledger first when the books are late or unreconciled. A forecast inherits the quality of the data under it, and a weekly schedule built on a stale bank feed is worse than no schedule, because it looks precise.
And decline it when the client will not give the firm the underlying access. Without the bank position, the aged receivables and the payables ledger each week, nobody is forecasting anything. They are guessing on a spreadsheet with a client's name at the top of it.
Where to Start
Take one client who already reports to a lender. Ask for the loan agreement, build to the format it names, and issue the first schedule with the variance report attached even in week one, when the variance is empty. Then watch what the second and third deliveries take out of the week.
Those hours are the number to know before this becomes a service the firm sells. The first build is a technical exercise and it is not what the client is buying. The recurring delivery day is the product, and it is the part that has to hold in April along with everything else the firm sells as advisory work.
Since 2022 we have placed 30+ trained offshore accountants and tax preparers inside 20+ US firms, working on their software and their SOPs, so recurring production work stops competing with the filing calendar. Don't trust us. Test us. Accountably runs a Free 40-Hour Proof Pilot on a fixed block of your own representative work, graded by your own reviewer before a single client file depends on it, and if someone is not the right fit in the first 30 days we replace them free. Start with the pilot