Sales growth isn’t a cash event. It’s an accrual event first, a receivable second, and cash only when cleared funds land in your bank. That order is the whole problem.

Call the control point expected clearing: the date an open invoice is expected to become usable cash. It isn’t the invoice date. It isn’t automatically the due date. And it certainly isn’t the date a customer says, with majestic confidence, that the payment run is “being processed.”

You can put revenue up 20%, grow receivables right alongside it, and still miss payroll before month-end.

The income statement records the sale. The balance sheet holds the timing gap in receivables, inventory, payables, and accrued liabilities. The cash-flow statement shows the reconciliation.

Your bank only cares about the last one. It has a very narrow reading list.

The Control Point

Start at the bank. Your 13-week direct-method forecast starts with reconciled cash on a specific cutoff date, not the cash number in your general ledger.

Outstanding checks, deposits in transit, processor holds, and unposted bank fees can make the two balances disagree before your forecast has even had a chance to be wrong.

Keep three figures separate:

Recorded cash is the accounting-ledger balance.
Bank cash is the cleared balance confirmed by the financial institution.
Available cash is what can actually be used after restrictions, settlement delays, frozen accounts, and segregated funds.

IAS 7 draws the boundary. Cash equivalents are short-term, highly liquid investments convertible to known cash amounts with insignificant value-change risk, and noncash financing and investing transactions stay outside the cash-flow statement.

So credit you haven’t drawn is potential liquidity, not cash. Show that line below your operating position with its expiry, covenant status, borrowing base, and lender discretion.

Then build receipts before payments. Put every open invoice into the week it’s expected to clear. Put payroll and payroll tax at their fixed dates, then supplier runs, rent, subscriptions, and debt service at the dates they’ll actually be released.

Net them by week, not month. A positive month can contain a negative week when payroll clears before a customer receipt. That’s the week that matters.

Past 13 weeks, roughly 90 days, named invoices get less reliable because many of them don’t exist yet. That horizon is a different forecast built a different way, and it belongs to the monthly cycle rather than this one.

The Two Clocks

Every credit account needs two controls. A payment term tells your customer when an invoice is due. A credit limit tells you how much exposure may stay open.

Net 30 without a limit is a clock with no wall around it.

Set your payment clock from a provable event: accepted delivery, milestone approval, or receipt of a valid invoice.

Use 7–15 days for easily verified repeated deliveries, 30 days for an ordinary business account, and 45–60 days only where the buyer’s approval cycle genuinely requires it.

Those are calibration bands, not magic amulets. The rule is that the starting event must be documentable.

Your forecast then replaces the contractual clock with the realized days to pay for that customer. A Net 30 customer that’s paid in 38 days on average over the last 4 quarters is forecast at 38 days.

Across the cited U.S. B2B material, 55% of invoiced sales run overdue against stated terms, and payment lands 43 days after invoicing on average.

Neither figure predicts one account, and both explain why a terms-only forecast models a business you don’t have.

Set your limit against peak exposure rather than average monthly sales. Include open invoices, released but unbilled work, authorized orders awaiting shipment, tax, and temporary exceptions.

A verbal promise doesn’t increase available credit. Cleared funds do.

Customer type	Credit limit	Payment-clock trigger	Realized days to pay	Escalation point	Cash exposure
Easily verified, repeated delivery	Set to peak open exposure, including orders and tax	Accepted delivery or valid invoice receipt	Customer’s own cleared-payment history	Missed documented promise or an invoice moving overdue	Open invoices, unbilled work, orders, tax, exceptions
Ordinary business account	Set to peak exposure and customer concentration	The contract’s provable start event	Actual days from invoice to cleared funds	Due-date aging plus a named obstacle and next owner	Open exposure less only cleared funds or formal credits
Buyer with a genuine approval cycle	Set to peak exposure, not average monthly sales	Milestone approval or accepted invoice	Actual behavior for that approval path	Approval, purchase-order, or acceptance defect is unresolved	Invoices plus work released before billing

The table doesn’t give you one escalation day, because the corpus doesn’t. That isn’t a missing spreadsheet cell.

It’s a control decision tied to value, promise reliability, friction, contractual rights, and the trading relationship. A high-value invoice waiting on a routine payment run needs a pre-due confirmation. An invoice with a missing purchase-order number needs the route fixed, not a louder reminder.

The Monday Loop

Run your 13-week forecast as a weekly rebuild. Don’t recreate it from scratch, and don’t overwrite the old view into oblivion.

Replace the completed week with actual bank-categorized receipts and disbursements.
Classify each miss as a timing error, amount error, or assumption error.
Roll the remaining 12 weeks forward, correct the assumptions, and add a new week 13.

Keep three columns: original forecast, latest estimate, actual. Grade your forecast at 1 week, 4 weeks, and 13 weeks ahead.

Original-to-actual shows whether the plan supported the decision you actually made. Latest-estimate-to-actual shows short-term control. Original-to-latest shows how far the outlook moved before settlement.

Measure receipts and payments separately. A small net miss can hide two large offsetting ones, with receipts below forecast and payments below forecast at the same time.

Keep signed error for bias, absolute error for the size of the miss, and ending-cash error for the liquidity consequence. Forecast accuracy that can’t explain your payroll trough is a decorative percentage.

On Monday, list every open invoice by expected clearing date rather than due date. Put the next collection action beside the 10 largest exposures: named payer, named obstacle, named owner, dated next event.

The only acceptable outcomes are payment expected on a stated date, documentation requested, dispute opened, settlement proposed, or escalation scheduled.

“Customer contacted” isn’t an outcome. It’s a diary entry.

The Aging File

Age every invoice from both invoice date and contractual due date. Invoice-date age shows elapsed collection time. Due-date age separates a long agreed term from genuine lateness.

Then read the aging as movement, not colored buckets.

For each starting band, show what was paid, stayed, rolled forward, moved backward on a correction or credit, entered dispute, or was written off.

“Current to overdue” is an early warning. “61–90 days to paid” is a cure.

Backward movement isn’t automatically collection success. Rebilling, a changed due date, or an unposted credit can make an old exposure look young without producing a cent.

Use invoice count and currency value together. Ten overdue invoices worth $500 tell you a different operational story from one overdue invoice worth $120,000.

Then pair that with concentration. If you’re leaning on one customer for half of next week’s receipts, show that dependency, even when DSO looks respectable. DSO moves with sales volume, seasonality, payment terms, customer mix, tax treatment, and the sales denominator. It’s a headline, not a diagnosis.

UK reporting guidance makes the point cleanly, because its payment measures separate average time to pay, late invoices by number, and late invoices by value.

Count, value, and timing answer different questions. Report one of the three and call it collections performance, and you’ll get ambushed by an invoice you were already measuring.

Reading the Symptom

Use the shape of the problem to choose the stage that needs repair.

Sales rise while cash falls, and receivables grow with revenue → forecast timing is broken. Open invoices are forecast at contractual terms rather than actual payer behavior. Rebuild expected clearing dates from realized payment history and inspect the 13-week trough.
Older bands fill while unapplied cash or approved-credit backlogs rise → ledger integrity is broken. Match cash, post credits, and check duplicate invoices and the legal billing entity before treating the balance as customer delinquency.
One customer moves the entire next-week position → concentration control is broken. Show that customer’s open exposure, payment behavior, and effect on available cash. A nominal limit doesn’t remove correlated risk.
Calls and emails increase while deposits don’t → the collection queue is broken. Separate attempted contact, right-party contact, promise, kept promise, and cash received. Name whether the obstacle is a purchase order, proof of delivery, tax document, acceptance, pricing, quantity, or ability to pay.
Month-end cash is positive but payroll still pinches → the measurement point is broken. Record the daily or intraday liquidity trough, not only the week-end or month-end balance.
Aged balances fall after disputes, credits, or write-offs → the outcome code is broken. Don’t call the decline a collection win until cash has actually cleared.

Separate Lanes

An invoice dispute is not ordinary delinquency, and it isn’t demonstrated credit loss either.

Put the contested amount into a resolution queue with the allegation, supporting documents, disputed and undisputed amounts, owner, and decision date. Keep collecting the amount the customer acknowledges.

Preserve the original due date. Reissue a whole invoice to accommodate a partial dispute and you erase the history you need for escalation.

Keep promises, disputes, credits, write-offs, and collected cash in separate fields.

A promise is a future event. A credit reduces the claim. A write-off is an accounting event that doesn’t by itself forgive the customer or stop collection, and a late recovery posts against the written-off account rather than quietly reopening the original invoice.

Under IFRS 9’s simplified approach, qualifying trade receivables without a significant financing component carry lifetime expected credit losses, and the provision matrix has to adjust historical outcomes for current conditions and reasonable, supportable forecasts.

That accounting estimate is not a substitute for the cash forecast.

For card disputes, don’t borrow your invoice cadence. The acquirer notice controls the response deadline, and Mastercard’s May 2025 guide runs 120-calendar-day windows for many conditions with outer limits of 365 or 540 days in specified delayed-delivery cases.

Those network clocks aren’t commercial-invoice rules. They’re just a reason to retain your evidence and read the actual notice.

A receivable is an asset in the ledger. Cleared funds are an asset in the bank. A collection promise is activity. Cash matched to the invoice is an outcome.

Forecast the clearing date. Control your exposure. Treat everything else as evidence until the bank says otherwise.