The 13-week rolling cash-flow forecast: the model every growth-stage CFO should run

Ask a Series A or Series B founder what their cash runway is and you will usually get one of two answers. Either an exact number ('14.2 months') derived from current cash divided by a flat monthly burn, or a vague one ('about a year, give or take'). Both are wrong, in different ways.

The exact number is wrong because it averages out the lumpiness of real cash movements: quarterly GST payments, annual insurance renewals, advance tax instalments under Section 211, ESOP exercise proceeds, customer milestone payments, year-end vendor settlements. The vague number is wrong because it is not actionable — a founder who knows runway to the nearest month cannot make capital-allocation decisions to the nearest week.

The 13-week rolling cash-flow forecast (often called the TWCF, or treasury working capital forecast) solves both problems. It is the direct-method weekly cash forecast that every working capital line, every leveraged transaction, and every disciplined growth-stage CFO runs as a matter of course. Done well, it is the highest-leverage hour the finance team spends each week.

What the 13-week forecast actually is

The 13-week forecast lists every expected cash inflow and outflow, by week, for the next 13 weeks. That is one quarter forward. Each line is at the granularity of the underlying transaction: not 'AR collections — Rs. 80 lakh this week' but 'TCS Foundation invoice 2026-1142, due 18 May, expected 22 May, Rs. 12.4 lakh; Nykaa settlement 19 May, Rs. 38.6 lakh; Razorpay net settlement weekly, estimated Rs. 22 lakh average'.

The week is the unit, not the month. Within a week, the cash position can swing by 20 to 30 percent — payroll falls on a specific day, GST on the 20th, TDS on the 7th. A monthly forecast misses every one of those swings. A weekly forecast catches the dip and lets the treasury team manage to it.

The 13 weeks is not arbitrary. It is far enough out to see the next funding gap, the next quarterly tax payment, the next compliance bill. It is close enough that each week's estimates can be grounded in real expected events rather than statistical averages. Anything beyond 13 weeks is a budget exercise, not a cash forecast.

The inputs

Six inputs feed the model.

AR aging multiplied by collection probability. Pull the AR aging report from the ERP. For each invoice, assign a collection probability based on customer history: a customer that has paid every invoice within 5 days of due date gets a 95% probability for that week and a 5% probability for the next; a customer with a track record of paying 30 days late gets 0% in week 1 and 60% in week 5. Roll these probabilities into expected collections by week.

AP aging with payment plan. Pull the AP aging. Apply your payment policy: if the policy is net-30 for vendors over Rs. 5 lakh and immediate payment for utilities, the AP forecast follows that policy, not the invoice due date. For vendors on extended terms or contested invoices, model the actual expected pay date.

Payroll dates. Net payroll, employer PF, ESIC, professional tax, all by the date they hit the bank. For most companies this is two outflows per month: net salary on the 1st or 5th, statutory dues on the 15th or 20th. ESOP exercises that result in TDS under Section 17(2)(vi) go in the week they happen.

Statutory dues. GST payment on the 20th of each month. TDS on the 7th of each month for the prior month. Advance tax instalments on 15 June, 15 September, 15 December, and 15 March. GST annual return (GSTR-9) by 31 December if your turnover is above the threshold. ROC filing fees in the months they are due.

Debt service. EMIs, interest on working capital facilities, principal repayments on term loans, all on the day they leave the account.

Capex and board-approved items. Anything the board or the CEO has committed to that has not yet been paid out: equipment purchases, security deposits, advisory fees, advance payments on long-lead items.

Why 13 weeks is the sweet spot

Forecasts shorter than 13 weeks (an 8-week forecast, for example) miss the quarter-end events that cluster: advance tax, GST annual return, year-end vendor settlements. A finance team running an 8-week view in early March will not see the 15 March advance tax payment.

Forecasts longer than 13 weeks (a 26-week or 52-week forecast) degrade fast because the customer-level collection assumptions become guesses. Beyond 13 weeks, the forecast is no longer a cash model — it is a budget restated in cash terms. Budgets and forecasts are different tools.

The five things that move forecast accuracy from 70% to 95%

Most companies that start a 13-week forecast settle into 65 to 75 percent accuracy on the next-week cash position. The gap from there to 90 to 95 percent comes from five operational disciplines.

Customer-by-customer collection forecasting

Not 'AR collections this week — Rs. 80 lakh'. 'Invoice 2026-1142 from TCS Foundation, due 18 May, expected 22 May (TCS has historically paid 4 days late), Rs. 12.4 lakh.' This forces the finance team to look at every open invoice every week, and it surfaces the invoices that should have been chased but were not.

Weekly review with the sales head

Half an hour every Monday with the sales head, walking through the top 20 open invoices and the new contracts being signed that week. The sales head knows things the finance team does not: which customer is about to release a payment, which contract is held up at the customer's finance department, which deal closing is at risk of slipping. That knowledge belongs in the forecast.

Supplier negotiation room for payment extension

Knowing on a Tuesday that you have a Rs. 1.2 crore cash dip in week 8 means you have time to call your top three vendors and ask for a 10-day extension on the largest invoice. Knowing on the morning of the dip means a missed payment and a damaged supplier relationship.

Line of credit availability tracked weekly

Most growth-stage companies have a working capital line they barely use. The 13-week forecast should flag every week where the projected closing balance dips below a threshold, with the action being 'draw Rs. X from the WC line in week N, repay in week N+2'. This is what working capital lines exist for, and most companies do not use them this way.

Variance review the week after

Every Friday, compare last week's forecast against actuals. Where was the forecast wrong, and why? A customer paid 5 days late — was the probability assignment too generous? A vendor invoice came in unexpectedly — was the procurement team not in the loop? The variance review is how the forecast accuracy improves over time.

What this looks like at a real company

A Series B SaaS company we run finance for had a 75% forecast accuracy when we started. After two quarters of the disciplines above, accuracy on the next-week cash position is 94%, on week 4 it is 88%, and on week 13 it is 78%. The founder no longer worries about runway in monthly chunks; she reviews the TWCF every Tuesday morning, looks at the next 4 weeks in detail, and makes capital-allocation decisions on a weekly cadence.

The same company has avoided two short-term cash crunches in the last six months that a monthly forecast would not have caught. In both cases, the action was simple — delay a discretionary capex by three weeks, accelerate one customer's collection by a week — and the decisions were made on the Tuesday before the crunch, not the morning of.

The 13-week forecast is not a finance team's deliverable. It is the founder's operating dashboard.

What it is not

The 13-week forecast is not your budget. It is not your funding model. It is not the model you take to the board to approve next year's spend. Those are different artefacts that live on a longer horizon and answer different questions.

The 13-week forecast is the operational model. It tells you, on Tuesday morning, what your cash position will be on the morning of payroll, on the morning of GST payment, on the morning of advance tax. If you run it well, it does not eliminate cash surprises. It moves them from Friday morning to the Tuesday three weeks before, which is where finance can do something about them.

Every growth-stage company we work with builds this model in the first quarter of the engagement. Within two quarters, it is the most-used file in the founder's drive.