Almost everything a 13-week cash flow forecast needs is already sitting in your QuickBooks file. Open invoices carry due dates, open bills carry due dates, recurring templates carry a schedule and an amount, and the last twelve months of cash movement give you a run rate for everything else. What is missing is the arrangement: someone has to sort those commitments into weeks, stack receipts above payments, and roll a bank balance forward. That is the part an assistant does quickly and without complaining. Three things stay with you, and they are the three that decide whether the forecast is any good: how late your customers actually pay, when payroll lands, and which one-off amounts you know about that the ledger does not.
What is a cash flow forecast?
A cash flow forecast is a week by week projection of the cash actually moving through your bank account, a different question from whether you are profitable. It has two halves. Receipts are the money you expect to land. Payments are the money you expect to leave. Each week opens with the previous week's closing balance, adds receipts, subtracts payments and produces a new one. The direct method builds those lines from real dated commitments, not from a percentage of forecast revenue: open invoices and their due dates, open bills, payroll runs, rent, sales tax remittances. Thirteen weeks is the usual horizon because it covers a full quarter, about as far out as a due date can still be trusted, and because it is what a lender or an investor asks for when they want to know if you can make payroll. A profit and loss cannot answer that. An invoice raised in March that pays in June is revenue in March and cash in June.
Why the direct method, and why 13 weeks
There are two ways to forecast cash. The indirect method starts from projected net income and adjusts for non-cash items and working capital changes. It is what the cash flow statement in your accounting software does, and it is the right shape for a full-year plan. The direct method starts from individual receipts and payments with dates on them. It is worse at twelve months and much better at twelve weeks, because over a quarter your cash position is decided by specific invoices from specific customers, not by a margin assumption.
Thirteen weeks is also the horizon where a forecast changes behaviour. A number that says you dip below your comfort line in week nine leaves you time to chase two invoices, delay one purchase, or call the bank. The same number for month eight is interesting and useless.
The ten-minute method
Ask these in one conversation, in this order, so each answer stays in context for the next one. Read the sample output as you go rather than at the end.
"What is the balance of every bank account right now, from the balance sheet?"
"Aged receivables as of today, then the detail: every open invoice with its customer, its amount and its due date."
"Group those open invoices by the week their due date falls in, for the next 13 weeks."
"Aged payables as of today, then every open bill with its vendor, amount and due date, grouped the same way by week."
"List every recurring template in the file, with its amount, its type and how often it runs."
"Cash flow statement for the last 12 months, by month. From that, what do payroll, rent and software cost in an average month?"
"Look at the last 12 months for anything that repeated but has no recurring template. Sales tax remittances, insurance, annual subscriptions, professional fees."
"Over the last six months, how many days after the due date did customers actually pay, on average? Show me the spread as well as the average."
"Now build the forecast. Weekly buckets starting Monday, opening balance from the bank accounts, receivables shifted by the collection lag you just measured, payroll on the 15th and the last business day, everything recurring on its own schedule."
"Lay it out as a 13-week table, one row per line, receipts on top."
The eighth prompt is the one people skip, and it is the one that decides whether the forecast is a forecast or a wish. Due dates tell you when an invoice is supposed to be paid. Payment history tells you when it will be. If your measured lag is fourteen days, every receivable line slides two weeks to the right, and the shape of the whole thing changes.
One detail is worth settling in the first prompt: which accounts make up the opening balance. Use the operating accounts that bills are actually paid from. A savings account you would have to transfer out of is available cash but not this week's cash, so keep it on its own line rather than folding it into the opening balance. A credit card is a payment obligation and belongs in the payments half, on the week the statement is due. Undeposited Funds is the awkward one. It is money you have received and not yet deposited, so it is real and it is close, and it is not in the bank balance the assistant reads off the balance sheet.
Example output, on a fictional company
Studio Nord is a six-person design studio in Montréal. These figures are illustrative, in Canadian dollars.
| Line | W1 8 Sep | W2 15 Sep | W3 22 Sep | W4 29 Sep | W5 6 Oct | W6 13 Oct |
|---|---|---|---|---|---|---|
| Opening balance | 62,400 | 72,050 | 59,950 | 76,370 | 50,020 | 56,670 |
| Invoices collected | 18,200 | 9,400 | 22,800 | 6,300 | 14,900 | 11,200 |
| Retainers (recurring) | 6,000 | 0 | 6,000 | 0 | 6,000 | 0 |
| Total receipts | 24,200 | 9,400 | 28,800 | 6,300 | 20,900 | 11,200 |
| Payroll | 0 | 18,400 | 0 | 18,400 | 0 | 18,400 |
| Rent | 4,800 | 0 | 0 | 0 | 4,800 | 0 |
| Contractors | 3,200 | 0 | 4,100 | 0 | 3,600 | 0 |
| Software | 1,150 | 0 | 380 | 0 | 1,150 | 0 |
| Bills due | 5,400 | 3,100 | 7,900 | 2,400 | 4,700 | 3,300 |
| Sales tax remittance | 0 | 0 | 0 | 11,850 | 0 | 0 |
| Total payments | 14,550 | 21,500 | 12,380 | 32,650 | 14,250 | 21,700 |
| Net movement | 9,650 | -12,100 | 16,420 | -26,350 | 6,650 | -10,500 |
| Closing balance | 72,050 | 59,950 | 76,370 | 50,020 | 56,670 | 46,170 |
The table continues to week 13 in the same shape. Six columns is what fits on a page; thirteen is what you hand to the bank.
What the forecast is telling Studio Nord does not show up in any single week. The balance never looks alarming here. It looks alarming in week 9, where a quiet collection week lands in the same seven days as the next tax remittance, and the closing balance goes under twenty thousand. That is the whole point of the exercise. The problem sits four ordinary weeks away and is completely fixable today, by chasing two specific invoices that are already past their due date.
A forecast is worth building only if you know what you would do about a bad number. There are three levers and they have very different lead times. Chasing receivables is the fastest and the cheapest, which is why the week 9 problem above is a collections problem before it is a financing problem. Delaying discretionary payments buys a week or two and spends goodwill with suppliers. Drawing on credit is the slowest to arrange and the most expensive, and it is the one you want to have arranged in week 1 rather than in week 9.
The three assumptions you have to supply
The assistant can read every dated commitment in the file. It cannot read the three things that actually move the answer.
| Assumption | Why the ledger cannot supply it | What to do |
|---|---|---|
| Collection lag | Due dates record the agreement, not the behaviour. QuickBooks knows when an invoice was due and when it was paid, but nothing in the file says which of your customers will keep doing that | Measure it from payment history, then apply it per customer where one or two big accounts behave differently from the rest |
| Payroll dates and amounts | Payroll usually runs outside QuickBooks and arrives as a summary journal entry, sometimes days later. The forecast needs the actual withdrawal date | State your pay schedule explicitly in the prompt, and use the gross cost including employer contributions, not the net pay |
| One-off items | A contract you signed last week, a tax instalment, a deposit on equipment, a bonus you have decided on. None of it exists in the ledger yet | Add them by hand as extra lines. This is the part of the forecast that stays a human job |
There is a fourth, less obvious one: your credit facility. A line of credit does not appear as an expected receipt anywhere in your books, and a forecast that ignores it will show you running out of cash in a week where you would simply have drawn on it. Say what is available and on what terms, then read the forecast against that number rather than against zero.
Turning it into a template you reuse
Once the numbers are right, the reusable part is the skeleton rather than the run. Copy this into a spreadsheet, or keep it in a document and paste it into the conversation next time with "fill this in for the 13 weeks starting Monday".
| Line | W1 | W2 | W3 | W4 | W5 | W6 | W7 | W8 | W9 | W10 | W11 | W12 | W13 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Opening balance | | | | | | | | | | | | | |
| Invoices collected | | | | | | | | | | | | | |
| Recurring receipts | | | | | | | | | | | | | |
| Other receipts | | | | | | | | | | | | | |
| Total receipts | | | | | | | | | | | | | |
| Payroll | | | | | | | | | | | | | |
| Rent and occupancy | | | | | | | | | | | | | |
| Contractors | | | | | | | | | | | | | |
| Software and subscriptions | | | | | | | | | | | | | |
| Bills due | | | | | | | | | | | | | |
| Sales tax and instalments | | | | | | | | | | | | | |
| Loan and lease payments | | | | | | | | | | | | | |
| One-off items | | | | | | | | | | | | | |
| Total payments | | | | | | | | | | | | | |
| Net movement | | | | | | | | | | | | | |
| Closing balance | | | | | | | | | | | | | |
Rerun it weekly, on the same day, and keep last week's version. The comparison between the two is more informative than either one alone: a receipt that has slid twice has stopped being a timing problem and become a collection problem, and it belongs in the overdue invoice chase rather than in the forecast. If you want the forecast to sit alongside a monthly narrative for a board or a lender, the management report skill packages that reporting pass in the same way. And if you keep several companies under one account, the same ten prompts run against each of them by naming the company in the question, so a firm can produce the identical table for every client in one sitting.
When QuickBooks Online's own cash flow planner is the better tool
QuickBooks Online ships a Cash flow planner, and for a lot of businesses it is the right answer. As of September 2026, it lives under Reports, then Financial planning, then Cash flow planner. It draws on your QuickBooks data plus your connected bank and credit card accounts, it lets you add Money in and Money out items including repeating ones, and those items never touch your books. You can set a threshold you do not want to go under, and export the result as CSV, PDF or XLS.
Two limits from Intuit's own help article settle the choice for many readers. The planner is not available in QuickBooks Online Accountant, and it is not available in any version when Multicurrency is turned on. If you are a bookkeeper working from QuickBooks Online Accountant, or a Canadian company invoicing in USD with Multicurrency on, the planner is not there and the table above is what you have.
Otherwise the split is simple. Use the planner when you want a maintained chart inside QuickBooks that updates itself. Build the 13-week table when you need a document to hand to somebody, when you want the same format across several companies, or when the assumptions need to be visible and argued about rather than embedded in a tool.
Where this is the wrong tool
You need scenario modelling. Best case, worst case, what happens if the big client goes to net 60. The assistant will produce a variant if you ask for one, but it will not maintain three linked scenarios that update together. That is a spreadsheet, or planning software.
Your cash position swings on things outside the ledger. Construction progress billing, inventory purchase cycles with long lead times, revenue recognised over time. The forecast still helps, but the hard part of the problem is not in QuickBooks.
Your books are three months behind. A forecast built on a ledger that stops in June is a forecast about June. Catch up first, then forecast.
QuickBooks Desktop. There is no API to read it from, so this runs on QuickBooks Online only.
And one caution that applies even when the tool is right. A forecast is a set of assumptions with arithmetic wrapped around it. The assistant does the arithmetic reliably and states the assumptions clearly, which is more than most spreadsheets manage, but it has no idea about the contract you are about to sign, the customer whose emails have gone quiet, or the instalment you have not thought about since March. Read the assumption lines first and the closing balance second.
If nothing is connected yet, start with connecting QuickBooks to your assistant. Every prompt in this guide is a read, so the whole forecast runs on read-only access, which is what every connection starts with. If you would like a warm-up before the forecast, asking for a profit and loss is the shortest way to see how the reports come back.