A dashboard shows a variance. What was decided afterwards, whether it was carried out and whether it worked sits beside it — usually in a second, manual steering world. This chain closes the loop and feeds it back into planning.
It is in the report, with an amount and a sign. The reason is not, and an owner certainly is not.
It was agreed in a meeting and noted in minutes. Whether it is running is known by whoever asks.
Which assumption was planned against gets reconstructed at year end. That is why the same planning error repeats.
The way back from effect to assumption is the part that is almost always missing in practice. Without it, planning is an annual ritual.
Four steps, and the same cross-functional capabilities across every one of them.
Every single step exists in some system today. What is usually missing is the column in the middle: the evidenced connection between them.
01 · Strategy and targetsFinance Control Tower
What is to be achieved, measured against which quantity, at which priority relative to other initiatives.
02 · PlanFinance Control Tower
Driver-based, integrated across P&L, balance sheet and cash flow — and with the assumption it was planned against recorded rather than reconstructed later.
03 · ForecastFinance Control Tower · Cash Forecasting
Rolling, with lineage per contribution and stated uncertainty. Forecast accuracy is a capability of its own, not a side analysis.
04 · SignalFinance Control Tower · Margin & Profitability
The variance with its driver: scope creep, concession, rework, an escalation not claimed. A variance without a driver is a number.
05 · DecisionFinancial Case Management
Options with an expected effect, approval by authority, the reasoning recorded. From here the variance has an owner.
06 · ActionFinancial Case Management · Investment, Cost & Value
What is done, by whom, by when, with what expected effect against a recorded baseline.
07 · ResultAccounting Operations · Margin & Profitability
The realised effect, from the same data basis, in the same period boundary as the plan.
08 · LearningFinance Control Tower
Expected against realised, and the way back into the assumption from station 02. Was the error in the assumption, in the action or in the execution?
The way back from station 08 into the assumptions of station 02 is why this is called a control loop and not a reporting layer. Without it the same planning error repeats every year.
Every module is a standalone entry point. This chain shows what comes together once several of them run on the same finance data model.
In active development
Driver-based planning with a recorded assumption, scenarios and approval paths. The assumption is recorded while planning, not reconstructed later — that is the only reason the way back is possible at all.
Pilot / validation
The rolling forecast with lineage per contribution and stated uncertainty, on the same data basis as plan and actuals.
In active development
The named driver between the margin stages. Only with it does a variance become a decision rather than a discussion.
In active development
The variance as a case with an owner, a deadline, approval and verification. Closure is a statement about the effect, not a status change.
In active development
The recorded baseline that the measurement runs against, and the lock against counting the same saving twice.
Renegotiate the price, raise a change order, redirect cost, or deliberately do nothing — within the authority that applies to this entity and this amount. Deliberately doing nothing is also a decision, if it is recorded.
Actuals come from the ledger and stay there. FinanceOS does not plan against a truth of its own.
Where one is in use it can keep its model and calculation logic. What FinanceOS contributes is the assumption, the decision, the action and the measured effect.
A report stays a report. This chain begins where it ends: at the question of who does what by when, and whether it worked.
What is not marked, we do not claim. The full state per capability is on the module pages and under what we can evidence today.
No. The model, the calculation logic and the input screens can stay where they are. What is added here is the recorded assumption, the decision with an owner and the measured effect — exactly the stations that sit beside the planning tool in emails and minutes.
The assumption planned against is recorded while planning. After the effect is measured it is therefore possible to say whether the error was in the assumption, in the action or in the execution. Without a recorded assumption that is an opinion.
Then that is recorded, with a reason. A chain that only knows successes is not effect measurement. Across several cases patterns become visible — which kind of action holds for which kind of variance.
Bring this case as it looks in your organisation. We work through it on your example and say where a common finance layer holds and where it does not.