Most cash forecasts are wrong by week two. Not because finance teams are careless, but because the model is built on how the business is supposed to behave — and businesses don't.
That gap shows up differently depending on who you are.
Start-ups. Sheer novelty makes forecasting a best-guess exercise until traction builds. … read more show less
It's also the best-case trap. Early on, relationships with suppliers and clients are close — often running directly through the founders — so every invoice lands within 30 days and the pipeline feels tight. When the biggest customer quietly moves to paying on day 45, or the sales team expands and the pipeline inevitably loosens, the model has to adjust. You also have to plan for the inevitable lack of accuracy.
Mid-size corporates. Established relationships and regular payment patterns help — but new complexity arrives. … read more show less
Longer-term contracts and regular payment patterns make the forecasting process easier. But new markets, sprawling sales operations and more exposure to fluctuating market prices — FX, inflation and possibly interest rates — give rise to new challenges.
Large corporates. The sheer complexity becomes the challenge. … read more show less
Numerous entities, each with their own bank accounts and currencies; treasury teams fragmented across geographies and time zones. Forecasting sales is the least of your problems — that variance fades towards background noise. Sweeping enterprise initiatives, integrating major M&A deals, geopolitics and regulatory change become the bigger threats to forecast accuracy.
Who owns the forecast
That split is where most forecasts struggle. Sales teams are notorious for managing expectations, and often can't give clarity on the timing of receipts. Procurement owns the payment calendar. If the people supplying the inputs never see the variances, the inputs never improve. The single most effective forecasting reform costs nothing: when a number misses, the person who supplied it explains why.
Core framework
Whatever your stage, a cash forecast is built from the same five parts:
- Starting cash position — bank balances, restricted cash, short-term investments and debt availability.
- Operating inflows and outflows — collections, payroll, supplier payments, taxes, subscriptions and recurring overheads.
- Non-operating items — capex, fundraising, financing flows, dividends, FX settlement, debt drawdowns and repayments.
- Forecast horizon and cadence — stage-dependent, but a 13-week forecast is the most common starting point, balancing detail against horizon. Supplement it with a longer view, generally 12–36 months. Review short-term forecasts daily or weekly; update longer horizons monthly or quarterly.
- Review and update loop — continuously test and refresh assumptions rather than "set it and forget it".
Improving forecast accuracy
Accuracy is earned, not bought. Four habits do most of the work:
- Visibility — make sure every contributor understands where their numbers feed in, what they drive, and who the audience is. Understanding the process generates buy-in.
- Accountability — costs nothing and improves accuracy fastest: when a number misses, the person who supplied it explains why.
- No blame — the aim is to improve accuracy, not apportion blame. Don't give stakeholders a reason to withhold an important update.
- Forecast, compare, explain, adjust — weekly, daily, monthly. Accuracy isn't a property of the model; it's a product of that loop running enough times. The explaining step matters most, because it separates bias from noise. Bias is systematic (sales forecasts always under-egged, receipts always optimistic by a week) and correctable. Noise is genuine surprise, and tells you how much buffer you need. Confuse the two and you'll either over-correct or learn nothing.
One model can't answer three questions
- Daily positioning — can we pay what's due today?
- The 13-week forecast — can we get through the quarter, or will we need to lean on facilities?
- The 12-to-36-month view — when do we refinance, and how much FX and rate exposure should we hedge?
Different horizons, different data, different owners. Companies get into trouble stretching one model across all three, usually by asking their 13-week sheet questions about a longer horizon that it was never built to answer.
A forecast is one path; liquidity needs the cautious ones
Whatever your model says will happen is a single line through a range of outcomes. The liquidity decisions — buffer size, facility headroom — belong to the bad lines: the biggest customer at 60 days, the supplier demanding prepayment, another point on the floating-rate stack.
At three sizes
Start-ups. Founders own the forecast — and it's geared towards fundraising, so it skews optimistic. … read more show less
In a forecast designed to track runway and liquidity, that optimism is a dangerous trap. At this level the longer forecast has one primary function: to pinpoint when the company runs out of money, and to time when the next fundraising cycle needs to begin. Honest inputs and caution matter more here than model sophistication.
Mid-size corporates. The 13-week discipline, connected to real bank data and reviewed weekly. … read more show less
Most importantly, at this level — with a finance or even treasury team in place — someone owns the misses. Longer term, a rolling 12- or 24-month view estimates beyond contracted payables and receivables, and a process for setting budget rates feeds directly into a policy for hedging market exposures.
Large corporates. Statistical methods and system feeds do the mechanical work — so the risk moves elsewhere. … read more show less
With consolidation across entities automated, the danger becomes a forecast nobody challenges: forgotten assumptions driving the models, fed by subsidiaries with their own reasons for optimism. The machine is only as honest as its inputs.
Key numbers to watch
The numbers that matter change as you scale. A rough map of what to watch, by stage:
| Start-up | Mid-size | Large enterprise | |
|---|---|---|---|
| Liquidity | Runway (months), cash burn | Cash-conversion cycle, operating cash buffer, DSO, DPO | Global liquidity coverage, trapped cash, entity liquidity requirements, pool utilisation |
| Forecast quality | Cash-burn variance, runway variance, 13-week accuracy | Forecast accuracy by entity/business unit, DSO/DPO variance | Accuracy by region/entity, exception rate, forecast confidence bands |
| Working capital | Cash-burn rate, receivables timing | DSO, DPO, inventory days, late payments | Net working capital by entity, intercompany balances, cash concentration |
| Control | Bank balance, reconciliations, payment visibility | Decision escalation, budget-to-cash variance, funding approval cycle | Policy compliance, segregation of duties, exceptions management, control breaches |
| Strategy | Time to next fundraising, hiring capacity, milestone funding gap | Capex affordability, covenant headroom, capacity | Repatriation capacity, debt headroom, strategic liquidity allocation |