A monthly income statement tells you whether the year is profitable. It will not tell you that payroll and three subtrade payments land in the same week eight weeks from now. A 13-week forecast will.
Key takeaways
STEP 01 OF 10
The forecast's week-zero starting point is the actual, reconciled bank balance on the day the forecast is built — not the general ledger's cash balance, which can differ from the bank by outstanding cheques, deposits in transit, or timing differences that do not matter for accrual accounting but matter enormously for a cash forecast. Reconcile the two before building week one, or every subsequent week inherits the starting error.
Rebuild this starting reconciliation every time the forecast rolls forward, not just the first time it is built. A forecast that trusts last week's ending balance without re-verifying it against the bank compounds any small error week over week.
STEP 02 OF 10
Break inflows into their actual sources: progress billing collections by project, holdback releases as they come due, any financing draws already committed, and anything else specific enough to attach a week and a dollar figure to. A single line item labelled “collections” hides which specific project's payment is actually driving the forecast's accuracy or its risk.
Where a collection date is genuinely uncertain, forecast it at the client's typical payment behaviour rather than at the contract's stated payment term — a client that reliably pays 10 days past terms should be forecast on that pattern, not on an optimistic contractual date the forecast will keep missing.
STEP 03 OF 10
Payroll, subtrade payments, material supplier terms, equipment lease payments, and loan covenant-driven payments each get their own line, by week. Payroll in particular is worth its own explicit line even though it feels automatic and predictable — it is exactly the outflow most likely to collide with an unexpected trough if it is not sitting visibly on the same forecast as everything else.
Include statutory holdback releases the firm itself owes to subcontractors as an outflow line, not just the holdback it is owed as an inflow. A forecast that only tracks holdback coming in, and forgets holdback the firm owes downstream, is only telling half the story.
STEP 04 OF 10
A monthly cash forecast smooths exactly the kind of week-to-week volatility a 13-week forecast exists to catch. Update the forecast every week: drop the week that just closed, extend the horizon by one new week thirteen weeks out, and true up every remaining week against what actually happened versus what was forecast.
Track the variance between forecast and actual for each closed week, not just the ending balance. A forecast that is consistently wrong in the same direction — collections always land a week late, say — is telling you something systematic to correct in how future weeks are built, not just noise to shrug off.
STEP 05 OF 10
The entire value of a 13-week forecast over a monthly one is lead time on a trough. Scan the full 13 weeks for the lowest projected ending balance, and treat that week as the one the forecast exists to manage — not the week that happens to be closest on the calendar. A trough eight weeks out is still eight weeks the firm has to arrange a bridge, delay a discretionary payment, or accelerate a collection, none of which is available once the trough has already arrived.
STEP 06 OF 10
BDC frames a healthy target as a working-capital ratio — current assets divided by current liabilities — of roughly 1.5 to 1.75. This is not a substitute for the week-by-week forecast, but it is a useful outside check: a firm whose 13-week forecast looks comfortable every week while its underlying working-capital ratio sits well below that range should treat the comfortable-looking forecast with some scepticism, not as confirmation everything is fine.
Recalculate the ratio from the same balance sheet the forecast's starting cash position was pulled from, so the two checks are actually looking at the same underlying numbers, not two different snapshots in time.
STEP 07 OF 10
Not every outflow line carries the same flexibility. Payroll and statutory holdback releases are effectively fixed; a planned equipment purchase or an early supplier payment taken for a discount is discretionary. Tag each outflow line accordingly, so that when a trough week appears, the forecast itself already shows which lines are actually available to delay and which are not.
STEP 08 OF 10
A trough identified eight weeks out is a planning conversation with a lender; a trough discovered the week it happens is an emergency conversation with a lender, and the two produce very different terms. Bring the forecast itself to a financing discussion — a lender responds differently to a firm that can show a specific, dated trough and a plan for it than to one asking for money because cash is currently tight.
STEP 09 OF 10
A 13-week cash forecast is only as accurate as the project-level billing and cost data feeding its inflow and outflow lines. See building a job cost reporting pack for the underlying project data this forecast should be pulling from — a forecast built on stale job cost data will be wrong in exactly the ways the job cost report itself is out of date.
Reconcile the forecast against the monthly close figures once a month, even though the forecast itself rolls weekly, to confirm the two are not quietly drifting apart from each other.
STEP 10 OF 10
A forecasting process that never checks its own historical accuracy cannot improve. Keep every rolled-forward 13-week forecast, and periodically compare a given week's forecast, made eight or ten weeks before it arrived, against what the bank balance actually was that week. Use the gap to correct whichever assumption — collection timing, a specific client's payment pattern, a systematically underestimated outflow — is producing the recurring error.
Building the forecast from the accounting system's book balance instead of the reconciled bank balance. Book and bank balances diverge for reasons that don't matter for accrual accounting but matter enormously for a cash forecast. Reconcile before week one, every time the forecast rolls.
Forecasting collections at the contract's payment terms instead of the client's actual payment behaviour. A client that reliably pays 10 days late should be forecast on that pattern. Forecasting the contractual date and being surprised every time is not a forecasting error, it's a refusal to use the data already available.
Updating the forecast monthly instead of weekly. A monthly cadence smooths out exactly the week-to-week volatility the 13-week forecast exists to catch. If it isn't rolling weekly, it isn't doing the job a 13-week forecast is for.
Treating every outflow line as equally fixed. Payroll and statutory holdback are not the same kind of commitment as a discretionary equipment purchase. Tag each line's flexibility so a trough week's options are visible on the forecast itself, not discovered under pressure.
Using national sector-wide data as a substitute for the firm's own forecast. Monthly national investment figures are real signal for the sector's direction, but they move at a scale and cadence too coarse to tell a specific firm anything about its own week 6 or week 9.
A simplified three-week slice of a 13-week forecast, with figures for demonstration only.
Week 1. Starting cash $85,000. Inflows (progress billing collected) $120,000. Outflows (payroll plus subtrade payments) $95,000. Ending balance: $85,000 + $120,000 − $95,000 = $110,000 — comfortably positive, and the kind of week that can make a monthly view look fine all on its own.
Week 2. Inflows drop to $78,000 (a large progress billing has not yet cleared) while outflows rise to $200,000 (payroll plus three subtrade payments landing the same week). Ending balance: $110,000 + $78,000 − $200,000 = −$12,000 — a real shortfall. Week 3. The delayed billing clears: inflows $165,000, outflows $88,000. Ending balance: −$12,000 + $165,000 − $88,000 = $65,000, positive again.
A monthly view spanning these three weeks would show net inflows of $363,000 against net outflows of $383,000 and call it a mildly tight month. It would never show the specific week-2 trough at −$12,000 — the one number that actually needed a decision, made with enough lead time to arrange a short bridge or shift one discretionary payment, rather than discovered the week the account actually went negative.
StatCan's Daily releases on building-construction investment break out by province and by residential versus non-residential — real numbers, not a single national trend line.
A firm bidding work in more than one province should treat the national monthly figure as background only, and watch its own province's line in the same release for whether the sector direction it is actually exposed to matches the headline number.
Thirteen weeks is roughly one calendar quarter — long enough to see a trough with real lead time to act, short enough that the far end of the forecast still relies on data close to the firm's actual known commitments rather than open-ended speculation.
No — they answer different questions. The annual budget and monthly close, covered in the monthly close guide, tell you whether the year is profitable. The 13-week forecast tells you whether the cash is actually going to be there on the specific week a specific payment is due.
BDC's own guidance suggests roughly 1.5 to 1.75 in current assets for every dollar of current liabilities as a target range, with the ratio ideally trending toward the higher end during a growth period rather than the lower one.
As early as the trough is visible on the forecast — typically 6 to 8 weeks, given a 13-week horizon. A lender presented with a specific, dated shortfall and a plan responds very differently from one approached the week cash actually runs short.
A 30-minute call is enough to tell you whether it is worth building.
