Why Cash Forecasts Fail: Five Structural Problems (and How to Fix Them)
July 22, 2026
Ask a group of treasurers whether they trust their cash forecast and the honest answers are rarely encouraging. The forecast exists, it is produced on schedule, and it is quietly discounted by everyone who uses it. In our experience, this is almost never a failure of effort or intelligence. Cash forecasts fail for structural reasons: fragmented data, misaligned incentives at subsidiary level, a forecast model asked to do too many jobs at once, the absence of any feedback loop, and a process that lives in a spreadsheet rather than in the organisation.
Over more than three decades of running treasury operations for multinational groups, we have reviewed hundreds of forecasting processes. The symptoms vary. Persistent variances, idle cash sitting alongside expensive borrowings, month-end surprises that force unplanned drawdowns. The underlying causes are remarkably consistent. This article sets out the five structural problems we encounter most often, and the fixes that work in practice.
Problem One: The Forecast Is Assembled, Not Produced
In most multinational groups, the consolidated forecast is a patchwork. Several ERPs, a dozen banking portals, and subsidiary submissions arriving by email in templates that have drifted apart over the years. By the time the figures are consolidated, the underlying data is already days old, and nobody at the centre can trace a number back to its source with any confidence.
We saw the cost of this vividly during a post-acquisition integration for a mid-market industrial group. The acquired business ran on a separate ERP and reported cash by emailed spreadsheet, typically a week in arrears. A substantial tax payment surfaced in the consolidated view with only days of notice, forcing an unplanned facility drawdown at an unattractive rate. Nobody had made an error. The information had simply travelled too slowly to be useful. Visibility problems become funding costs.
The fix does not begin with the ERP. It begins with the banks:
- Prioritise bank-level visibility first. Direct bank connectivity, whether through SWIFT, host-to-host channels or APIs, delivers reliable balance and transaction data long before any ERP integration completes.
- Impose a single submission template with standard flow categories across every entity. The categories matter far more than the tool.
- Use the structured data that ISO 20022 messages now carry. Automated categorisation of flows is considerably more reliable than it was under legacy formats, and forecasting is one of the places where that pays off first.
Problem Two: Subsidiaries Forecast Defensively
This is the problem treasurers discuss least, because it is behavioural rather than technical. A local finance manager who under-forecasts inflows and pads outflows will never be caught short, and will never be criticised. Multiply that quiet conservatism across twenty entities and the group forecast shows a liquidity shortfall that does not exist.
One group we worked with drew on its revolving credit facility repeatedly over several quarters to cover forecast shortfalls that never materialised. When we compared submissions against actuals entity by entity, the pattern was unmistakable: subsidiaries were systematically building in buffers. The group was, in effect, paying commitment fees and margin to insure itself against its own reporting behaviour.
| From the field. The clearest signal of defensive forecasting is directional bias. Random forecast error scatters in both directions. Sandbagging shows up as a persistent one-way variance: actual closing cash consistently above forecast, entity after entity, quarter after quarter. Once you measure it, you cannot unsee it. |
The remedy is transparency rather than technology:
- Measure forecast accuracy by entity as a standing KPI, and publish it. Visibility alone changes behaviour within two or three cycles.
- Keep the exercise blame-free but open. The objective is a realistic number, not a safe one.
- Separate forecasts from targets. The moment a forecast is used to judge performance, it stops being a forecast and becomes a negotiation.
Problem Three: One Model Asked to Do Three Jobs
Many companies attempt to serve every purpose with a single forecast: daily funding decisions, quarterly covenant headroom, annual planning. The result satisfies none of them. It is too coarse for daily cash positioning and too noisy for strategic decisions.
Mature treasuries layer their forecasting instead. A daily-to-weekly view built on the direct method, four to six weeks out, drives cash positioning and short-term funding. A 13-week rolling forecast, also direct, is the workhorse for liquidity management, covenant monitoring and the timing of FX exposures. Beyond one quarter, a monthly indirect forecast derived from the projected P&L and balance sheet supports capital structure and planning decisions. Each layer has its own owner, its own cadence, and its own tolerance for error. Forcing one model to carry all three is among the most common design flaws we encounter, and it is usually invisible to the people inside the process because the single model has always been there.
Problem Four: The Forecast Is Filed, Not Tested
Ask a treasury team what their forecast error was last quarter and the silence is often the answer. If nobody compares forecast to actual in a structured way, the process cannot improve. The same errors are simply repeated with fresh dates.
The discipline that fixes this is unglamorous and effective. Compare forecast against actuals at every cycle. Classify each material variance by root cause: timing, amount, or omission. Track the error trend over time, by entity and by category. The objective is not a perfect forecast. It is a forecast whose errors are understood, shrinking, and concentrated where they matter least. That is a standard any treasury team can meet, and it costs nothing but routine.
Problem Five: A Spreadsheet Task Instead of a Governed Process
In many organisations the entire forecasting process lives in one workbook, maintained by one person, documented nowhere. When that person is on leave, the process degrades. When that person resigns, it collapses. This is key-person risk dressed up as efficiency, and internal audit teams are increasingly unwilling to accept it.
Forecasting deserves the same governance as any other treasury process: a defined owner, a documented methodology, a published calendar, and clear escalation rules when projected liquidity breaches agreed thresholds. Technology, whether a TMS module in Kyriba or SAP Treasury or a specialist forecasting tool, belongs at the end of this journey rather than the beginning. Automating a broken process merely accelerates it.
A Quick Diagnostic: Six Questions for Your Next Treasury Meeting
| ☐ | Can we see all group bank balances from a single point, on a same-day basis? |
| ☐ | Do we measure forecast accuracy by entity, and do the entities know it? |
| ☐ | Do we run separate short-term (direct) and medium-term (indirect) forecasts, each with its own owner? |
| ☐ | Is variance analysis a standing agenda item, with root causes classified and tracked over time? |
| ☐ | Is the methodology documented well enough that a new joiner could run the process unaided? |
| ☐ | Do forecast outcomes actually drive decisions on funding, investment and hedging, or are they produced for their own sake? |
If the answer to two or more of these is no, the issue is structural, and no amount of spreadsheet refinement will resolve it.
Forecasting Is an Operating Capability
Reliable cash forecasting is less a modelling exercise than an operating capability. It combines data infrastructure, behavioural design, layered methodology, feedback discipline and governance, and it has to run every week whether or not anyone is on holiday. That is precisely why it is difficult to sustain with a lean in-house team, and it is one of the reasons many mid-market multinationals choose to anchor the process within a centralised structure, whether an in-house bank, a shared treasury function or an outsourced operating model. The organisational form matters less than the principle: someone must own the process end to end, with the data, the mandate and the routine to keep it honest.
Frequently Asked Questions
How accurate should a cash forecast be?
There is no universal benchmark. Acceptable error depends on the horizon, the volatility of the underlying flows, and the decisions the forecast supports. A short-term positioning forecast should be materially tighter than a 13-week view. What matters most is that accuracy is measured consistently and that the error trend is improving over time.
What is the difference between direct and indirect cash forecasting?
The direct method builds the forecast from expected receipts and payments and suits short horizons where transaction-level data exists. The indirect method derives cash flow from projected P&L and balance sheet movements and suits medium- to long-term planning. Mature treasuries use both, in layers, rather than choosing one.
How often should the forecast be updated?
Short-term forecasts should roll weekly at a minimum, with daily updates to the cash position. The 13-week view is typically refreshed weekly on a rolling basis, so the horizon never shortens as the quarter progresses.
Can AI fix a poor forecasting process?
AI and machine learning can genuinely improve pattern recognition in receivables behaviour and payment timing, but only where the underlying data is complete, categorised and timely. Applied to fragmented or defensive inputs, automation simply reproduces the structural problems at greater speed. Fix the process first, then automate it.
FTI Treasury has provided cash reporting, forecasting and treasury outsourcing services to multinational corporations for over 30 years. Our operating model is built around the disciplines described in this article, including consolidated bank-level visibility, standardised forecast processes and systematic variance analysis. If you would like to discuss how your forecasting process compares to current best practice, contact our team.
Related Services: Cash Reporting & Forecasting | Treasury Outsourcing | In-House Banking