By: Eleanor Hill
The first warning sign of a material revenue or cash movement rarely appears in the ERP. By connecting upstream operational events to their future financial consequences, through new data technology, CFOs can move beyond periodically rebuilding the forecast and start acting while there is still time to change the outcome.
Eight weeks before a quarterly cash forecast misses its target, the numbers may still look perfectly healthy – somewhere else in the business, however, the position has already begun to change. Perhaps a specialist needed for a major implementation is unavailable, or a project milestone has slipped by two weeks. Or perhaps the customer simply can’t approve the completed work on the expected date, which means the revenue can’t be recognised, the invoice can’t be raised, and the cash won’t arrive as planned.
Every link in that sequence is entirely logical. Yet finance may see none of it until the quarter closes and the expected receipt fails to materialise. The problem is neither poor arithmetic nor a genuinely unforeseeable event, but the distance between the first operational evidence and the financial view.
“Something that happened eight weeks earlier may only impact the financials at the end of the quarter,” explains Ben Reid, Chief Executive of Intelligent Lagoon. “It’s not that nobody in the business knows about it – it’s because it sits upstream of finance, in systems that were never connected to it, so finance experiences it as a surprise, after the fact.”
Closing that distance between operational change and financial awareness could reshape more than the accuracy of the forecast. It gives CFOs a longer window in which to reprioritise resources, resolve customer issues, renegotiate billing arrangements or adjust liquidity, before the consequences crystallise.
When the work obscures the answer
Traditional forecasting processes can make that early intervention extraordinarily difficult, particularly in decentralised organisations where numerous finance leaders produce submissions for their own parts of the business. Each contributor applies a different interpretation of risk, a different level of conservatism, and a different degree of diligence – and the consolidated forecast moves at the speed of the last person to complete it. The result is a point-in-time snapshot that can conceal both the assumptions behind the numbers and the risks already developing beneath them.
“So much effort goes into producing the forecast that there’s often not much time left to analyse whether it’s right,” observes James Kelly, former group treasurer at Pearson and co-founder of Your Treasury.
The limitations become even more obvious once actual results are compared with the forecast. A report showing that accounts receivable in one country came in £5m better than expected offers little insight into which receipts arrived, which assumptions proved correct, or what changed beneath the headline number – and net figures can be especially deceptive.
What’s more, a seemingly modest £1m variance may conceal several much larger movements that happen to offset one another. And until those gross movements are separated out, finance cannot tell whether the underlying view of the business remains sound and the cash has simply moved between periods, or whether a significant payment was omitted altogether.
Automation removes some of the manual labour without necessarily solving the explainability problem. An aggregate model may predict that £5m will arrive over the next few days – if £8m turns up instead, the answer can’t simply be that the algorithm was wrong. Finance still needs to understand which transactions created the difference, and whether the same pattern is likely to recur.
A better process, in other words, needs more than a faster way of producing another number. It needs enough granularity to reveal the events behind it.
A surprise with a long lead time
Of course, business uncertainty cannot be entirely eliminated, and no forecast should pretend otherwise – deliveries move, customers change their minds, projects run into difficulty, and resources become constrained. The more damaging surprises are usually the ones that develop gradually, while the financial forecast continues to assume the original plan will be delivered exactly as written.
One response is to forecast more frequently, but repeatedly asking the organisation to reconstruct a complex submission is neither practical nor sustainable. Another is to build large buffers into every expected timeline, which may reduce the number of apparent surprises, but only by making the entire business slower and more conservative. Dr Jamie Ballin, Chief Innovation Officer and Founder of Intelligent Lagoon, describes those buffers as a “silent killer of business opportunity”.
The temptation, then, is to compensate with more advanced algorithms, but Ballin questions what they can achieve if the underlying information remains unchanged.
“We don’t really need more sophisticated algorithms looking at the same data sets,” he argues. “The signal does often exist in the business – it’s just not visible to the people who are making the forecast when they need to make it.”
According to Ballin and Reid, the opportunity lies in following the chain of causality before it ever reaches the ledger: a delayed resource affects project completion, this then determines customer acceptance, which controls revenue recognition and invoicing – in turn, shaping the timing and probability of cash. Viewed individually, those relationships can look fairly obvious. Across hundreds of contracts, operating processes, and systems, however, informal communication alone is not enough to keep the financial picture reliable. So, what can technology offer to help?
Read the signal, not just the variance
Earlier visibility into upstream operational events and processes helps finance interpret the nature of a potential miss, rather than treating every delayed receipt as the same problem. A major invoice might be held up because somebody omitted a purchase order number, while another customer may be withholding payment because delivery has fallen short. The immediate cash effect can look identical either way, but the CFO’s response should not be.
An administrative obstacle may need little more than a focused intervention to complete the missing process and accelerate payment. A performance dispute is a different matter, potentially affecting revenue recognition, future receipts and the wider customer relationship. Yet the distinction is rarely apparent from the financial data alone.
“If a client has already made three complaints to customer services and is unhappy, there’s a heightened chance they won’t pay the invoice when it falls due,” Kelly points out. “The human still has to interpret whether that’s an irritating process failure or something fundamental to the contract – but the signal means the business can investigate before the payment is missed.”
Reaching that conclusion after the problem has surfaced can mean searching through project documentation, customer service records, CRM notes, and correspondence, by which point the picture may have been further distorted by reluctant reporting, fragmented ownership, and each function’s natural tendency to treat the issue as somebody else’s responsibility.
Making the connection earlier turns the cash forecast into a shared operational concern rather than finance’s alone: project leaders, customer teams and finance can all see not just that an activity has moved, but what the movement is likely to mean further along the chain.
Not every cash flow deserves a spotlight
That said, building that level of intelligence around every transaction would be unnecessary – many cash flows are regular, high-volume and sufficiently consistent for conventional statistical or machine-learning methods to handle perfectly well.
Your Treasury’s Cash Flow Diagnostics work helps companies distinguish those flows from the irregular transactions that deserve closer attention: its regularity and outlier power indices examine the size, frequency and timing of unusual movements, on the basis that the same variance can carry very different consequences depending on when it lands. A payment that slips during a period of abundant liquidity may be a minor inconvenience – but the same delay immediately before a dividend, debt repayment or major capital investment can demand urgent action.
Kelly recalls two large state customers from his time at Pearson that each paid only once a year, with individual receipts of around $110m and $140m, against annual operating cash flow of roughly £400m to £500m – either payment landing in January rather than November or December could materially alter the year-end position. There is little historical data from which a conventional model can learn when a transaction happens only once a year, and no way to dilute the risk across thousands of similar receipts. Each payment is individually significant, and depends on its own specific set of operational, contractual, and customer events.
“Those are the transactions where you either appoint somebody to man-mark every stage of the process, or put technology in place to provide robust signals,” Kelly explains. “The sensible starting point is the handful of flows that genuinely move the needle.”
Concentrating on those outliers also stops an improvement project turning into another vast transformation programme. Routine activity can stay automated and relatively light-touch, while richer data and closer scrutiny are reserved for the contracts, counterparties, and cash movements capable of changing an important decision.
From snapshots to a living portrait
Until now, a major constraint to taking this approach has been the lack of a cohesive view on operational information. But new data technologies are beginning to address that gap by giving operational data a sense of state, sequence, and probability – within a common model. Intelligent Lagoon’s timeplane®, for example, applies that structure alongside the systems companies already use, organising disparate events into a continuously updating view of how a process is unfolding and how today’s operational activity is likely to affect future financial performance.
The underlying structure is a graph connecting the people, decisions, activities, and dependencies that shape the eventual result. Ballin contrasts that approach with static planning tools that assume each milestone will land on a predetermined date. “Gantt charts are all very well, but they simply can’t deal with the uncertainty,” he notes. The further out the view extends, the less useful those fixed dates become for forecasting purposes. What finance needs instead, Ballin argues, is to see what is likely to happen in three, four, or five weeks’ time, what that outcome depends on, and how the position shifts as new information arrives.
Operational data can be drawn from established systems such as CRM, project management, or resource planning platforms; where processes are still managed through contracts, spreadsheets, or more informal arrangements, timeplane® can provide the structure through which that information gets captured instead. Each operating team stays responsible for the activity it understands, while the financial implications update automatically as those activities progress.
“The notion of forecasting becomes much less important,” Reid suggests. “Once the upstream activities have been modelled, the probabilities attached to the financial metrics remain forward-looking and up to date, by definition.” In other words, rather than assembling the forecast through a periodic, resource-intensive exercise that may already be out of date by the time it is finished, the organisation gains a forward-looking view that moves with the business.
Foresight changes the decision
Reducing the effort required to prepare forecasts and explain variances is an immediate efficiency gain, but the larger prize is the ability to act differently, believes Reid. Returning to the delayed implementation example, finance could see that resource constraints are putting a material milestone at risk, and help the business decide whether scarce expertise should be redirected before the quarter closes.
If full delivery simply can’t be recovered, the company might negotiate partial acceptance or staged billing, rather than let the entire invoice slip unnoticed into the next period. Treasury and finance, meanwhile, can adjust liquidity plans earlier, concentrate collection activity on the most vulnerable receivables, or review the use of credit facilities – and management can work out whether a prospective miss is a temporary timing issue or evidence of a more substantial performance problem, before deciding what to say externally.
Operational signals can support decisions well beyond working capital, too. Kelly points to an example in which an organisation treated the purchase of insurance for a new site as evidence that a planned data-centre investment had become sufficiently certain to act on – treasury could then begin arranging funding and hedging the associated foreign exchange (FX) exposure, rather than waiting for the expenditure to show up in a finance system months later.
The same principle applies across construction projects, private equity-backed businesses and any organisation managing large, irregular commitments: in each case, a seemingly peripheral operational event can turn out to be the most useful early indication of future financial activity.
“Every single day, you’re making decisions about how to run the business that will ultimately play out in its financial performance,” Reid reflects. “The more you can understand how those decisions affect that performance, the smarter the strategic decisions you can make.”
Such foresight could eventually support experimentation as well as prediction. Once the causal relationships are mapped, management can test how changing a contractual process, an allocation decision, or an operating model might affect future revenue, margin, or cash – before committing to the change itself.
Start with the problem child
Thankfully, none of this requires an immediate replacement of the ERP, TMS, or other operational systems already in place – timeplane® is designed to sit alongside them, giving existing data a forward-looking structure rather than demanding a wholesale technology overhaul.
The most practical first step is simply to identify the activity that repeatedly creates material surprises: one large customer payment, perhaps, a milestone-led revenue stream, an uncertain procurement process, or a group of contracts whose performance can’t be understood through historical snapshots alone.
Mapping that process can reveal value even before the full model is running. Contractual milestones may sit in one system, customer information in another, and financial dates in a third, with no single function able to see the whole chain – bringing those dependencies together clarifies ownership and enables finance colleagues to start asking better questions earlier. From there, the scope can expand as the initial use case proves its worth, connecting more processes, stakeholders, and financial outcomes without trying to boil the ocean from day one.
In summary, forecasts will never remove uncertainty from a business, but they shouldn’t leave finance discovering material change only once its consequences have already arrived. By the time a receipt is missed, a covenant comes under pressure, or a quarter-end number moves, much of the opportunity to intervene has usually already gone.
The real advance here isn’t another attempt to predict one perfect number. It’s empowering the finance team to see the business changing before the numbers move – and making sure the CFO is no longer the last to know.
