From Data to a Manageable Model
Ultimately, an FP&A process is only as good as the information on which it is based. However, it would be too simplistic to conclude from this that FP&A primarily needs more data. In most organizations, there is actually no shortage of data. There is a lack of coherence.
Financial data resides in the ERP system, commercial information in the CRM, personnel data in HR systems, and operational data in production, logistics, project, or service systems. In addition, valuable data regarding orders, invoices, payments, suppliers, customers, and working capital is generated within finance operations. External developments such as commodity prices, interest rates, inflation, and market growth can further complicate matters.
The challenge for FP&A is to build a manageable model of the company from all of this.
From Financial Accounting to Business Model
Financial accounting serves as an important starting point. After all, the income statement, balance sheet, and cash flow statement reflect the financial consequences of the company’s business activities.
But a general ledger does not automatically explain those results.
For example, when revenue declines, the cause may lie in lower volumes, a changed product mix, price pressure, customer churn, or delays in deliveries. When personnel costs rise, this may result from an increase in FTEs, salary increases, a different workforce composition, or higher costs for external hires.
For financial reporting, these causes may ultimately end up in the same general ledger account. For business control, however, they make a fundamental difference.
FP&A must therefore establish the connection between financial results and the operational reality that drives those results.
This changes the way data is viewed. The goal is no longer solely to ensure amounts are posted to the correct general ledger accounts, but to understand the economic relationships underlying those amounts.
Finance operations as a relevant data source
Purchase-to-Pay and Order-to-Cash are particularly interesting data sources in this regard, as they are closely tied to the actual transaction flows.
Purchase-to-Pay can contain information on, among other things, suppliers, orders, liabilities, invoices, payment terms, and actual expenditures. Order-to-Cash includes, for example, customer orders, invoices, credit information, open items, and payment behavior.
For FP&A, this data can provide early warning signs.
A general ledger may not reveal that certain costs are higher until month-end closing. An analysis of open orders and received invoices can make that trend visible sooner. The same applies to the revenue side: order inflow can provide earlier insight into future revenue than the ultimately recorded sales invoice.
Nevertheless, finance operations should not be viewed as the sole data source for FP&A. It constitutes just one relevant layer within a much broader data landscape.
For a manufacturing company, production volume, capacity utilization, and raw material consumption can be just as important. For a service provider, billable hours, staffing levels, and rates play a greater role. In the software industry, ARR, churn, and customer acquisition can be key drivers.
The relevant data sources are therefore determined by the business model.
Drivers Are More Important Than General Ledger Accounts
A good FP&A model therefore does not start with the question of which general ledger accounts are available, but with the question of which factors determine the organization’s financial performance.
We call these factors “drivers.”
Revenue, for example, can be described as:
number of customers × average volume × average price
Personnel costs can be broken down as follows:
number of FTEs × average labor costs
Transportation costs may be related to:
number of shipments × average distance × rate per kilometer
By tracing financial outcomes back to operational drivers in this way, you create a model that not only reports but also explains.
This is essential for planning and forecasting. Those who understand what drives a result can also better model what happens when one of those factors changes.
A five percent change in volume can then be automatically translated into revenue, production costs, logistics expenses, working capital, and ultimately cash flow.
This makes the model useful as a management tool.
Dimensions determine which questions can be answered
In addition to drivers, dimensions are important. These refer to the way in which financial and operational information is broken down.
Think of customer, product, region, business unit, supplier, project, cost center, or sales channel.
The choice of dimensions may sometimes seem like a technical data issue, but in reality, it is a management decision. Every dimension added to a model enables specific analyses.
If revenue is available only at the entity level, management cannot analyze which products or customer groups are driving growth. If personnel costs are recorded exclusively by legal entity, it becomes difficult to assess productivity by department or activity.
At the same time, more detail is not automatically better.
Every additional dimension increases the model’s complexity, the volume of data, and the maintenance burden. The question must therefore always be which breakdown is actually relevant to decision-making.
A manageable model contains sufficient detail to reveal the most important economic relationships, but avoids details that have little impact on decisions.
A Single Definition of Reality
Another important issue is that different systems often use different definitions.
What exactly is a customer? When does an order count toward revenue projections? Which FTEs are included in workforce planning? When is a project considered active?
When each department uses its own definitions, multiple versions of the same reality emerge.
FP&A therefore needs not only a technical data model but also clear definitions and governance.
Management must be able to trust that a KPI in a forecast has the same meaning as the same KPI in a report. A discrepancy must not turn out to be the result of two different definitions.
That may sound like an administrative issue, but it’s a key prerequisite for sound decision-making. If a management meeting starts with a twenty-minute debate over which figure is correct, FP&A hasn’t done its job yet.
Actuals, plans, and forecasts must speak the same language
A common problem arises when actuals, budget, and forecasts are structured differently.
For example, actuals follow the general ledger structure, while the budget is in a separate spreadsheet, and commercial forecasts use yet other product or customer categories.
This makes every comparison labor-intensive.
A mature FP&A environment therefore strives to align actuals, planning, and forecasting with the same model as much as possible.
The same dimensions, drivers, and definitions are used for both historical and future information.
This makes it possible to immediately analyze what actually happened, what was originally planned, what is currently expected, and what is causing the differences.
Technical integration thus supports a managerial need: a single, coherent perspective on the past, present, and future.
Avoiding False Precision
In modeling, there is always the temptation to include as much detail as possible.
A model with a hundred assumptions quickly seems more professional than one with ten. But more detail does not necessarily lead to better decision-making.
On the contrary.
When there is significant uncertainty surrounding a forecast, adding more detail can create the impression of precision without actually increasing reliability.
A revenue forecast of €42,783,219 may be mathematically correct according to a model, while the key volume assumption has a margin of uncertainty of ten percent.
A senior business controller therefore looks not only at the technical quality of the model but also at the material relevance of the assumptions.
Which drivers account for the largest portion of the outcome? Which assumptions are uncertain? Where is a rough estimate sufficient, and where is more detail necessary?
A good model is not the most detailed model. It is the simplest model that describes economic reality well enough to enable better decisions.
From Data Platform to Administrative Infrastructure
Technology can automate much of the work involved in data and modeling. Data can be automatically collected, validated, and consolidated. Models can be maintained centrally, and changes can be reflected immediately in reports and forecasts.
But technology cannot fix a poorly designed governance model.
When the wrong drivers are used, definitions are unclear, or the organization doesn’t know which information is truly relevant, the result is—above all—an ambiguous outcome produced even faster.
That is why a mature FP&A data model starts with the business model itself.
How does the organization make money? Which activities generate costs? Where does cash come from? Which capacity is scarce? Which factors determine growth and margin? Which risks could significantly alter the outcome?
Only then does the question arise as to what data is needed for this.
Data & modeling is therefore not a preparatory IT activity on the periphery of FP&A. It forms the economic architecture upon which planning, forecasting, and performance management are built.
Its quality largely determines how well finance can fulfill its role as a strategic partner to management.
The ultimate goal is not a single central database. The goal is a single, manageable view of thecompany.