TL;DR: Pull data from the ERP. Reconcile it in Excel. Build the model. Check for errors. Update the dashboard. Write the document. Share the answer. A process that takes three days and delivers understanding that should have taken three minutes.
It is Tuesday afternoon. A question arrives from the sales director. Revenue came in ahead of plan last month but margin looks soft. Can finance explain why before the leadership call on Thursday?
It is a reasonable question. The answer is almost certainly sitting in the data. And in most mid-market finance teams, what happens next takes the better part of two days, involves four different tools, and produces an answer that is accurate, well-formatted, and approximately forty-eight hours too late to change anything.
This is the pattern that defines traditional FP&A. Not incompetence. Not lack of effort. A workflow that was built for a different era and has never been fundamentally redesigned for the speed at which the modern business operates.
McKinsey's research on the finance function is direct on this point: CFOs should not be spending eighty percent of their time on reporting and reactive analysis. That is precisely what the current workflow produces. And the extraordinary thing is that every step in that workflow feels necessary, because inside the existing system, it is.
Step One: Find the Data
The first task is locating the relevant data. In a mid-market business running an ERP alongside a data warehouse and operational source systems, this is rarely simple. The revenue data is in the ERP. The cost data may be partially there and partially in an Excel model maintained by a specific analyst. The margin calculation depends on cost allocations that live in a workbook built six months ago.
So step one is not analysis. It is archaeology. The FP&A manager opens the ERP, identifies the relevant reports, exports the data, and cross-references it against the planning model to understand what the baseline comparison should be. This takes between two and four hours depending on how clean the underlying data is and how many systems need to be accessed.
A KPMG survey found that eighty-seven percent of finance teams report spending significant time reconciling data discrepancies rather than generating insight. This is where that time goes. Not on a special project. On the routine first step of answering any business question.
Step Two: Reconcile It
The data from the ERP rarely matches the data in the planning model on the first pass. There are timing differences, reclassifications, intercompany adjustments, and accruals that need to be accounted for before any comparison is meaningful. This is reconciliation, and it is one of the most time-consuming and least value-adding activities in the finance function.
The analyst opens Excel. A new tab is created. The ERP export goes in one column, the plan in another, and the process of explaining the differences begins. Some differences are genuine variances worth investigating. Some are artefacts of how data flows between systems. Distinguishing between them requires judgment, and judgment requires time.
This step alone commonly takes three to five hours for a mid-market business with moderate complexity. By the end of it, the analyst has confirmed that the data is clean and the comparison is valid. They have not yet produced a single insight.
Step Three: Build the Model
With clean, reconciled data, the analysis can begin. The FP&A manager builds a model in Excel that isolates the revenue and margin movements, separates volume effects from price and mix effects, and identifies which cost lines moved relative to plan. This is where the real analytical work happens, and for experienced finance professionals, it is the part of the workflow they are trained for and good at.
It is also the part that happens last, after the data gathering and reconciliation have consumed the majority of available time. The window for genuine analytical depth has already narrowed significantly by the time the model is being built.
The model takes another three to five hours. When it is finished, the FP&A manager has an answer. The margin compression is real. It is driven by a mix shift toward lower-margin product lines and a discounting pattern in one region that started in week two of the month.
Step Four: Check for Errors
Any finance professional who has worked in a mid-market environment knows this step intimately. The model is built. The answer is clear. And before anything is shared with the business, someone needs to verify that the model is correct.
This is not paranoia. Excel models break. Formulas reference the wrong cells. Data exported from the ERP contains a row that was not in the last export. A subtotal includes a line that should be excluded. The consequences of sharing an incorrect analysis with the sales director are significant enough that verification is not optional.
A second analyst reviews the model. Or the FP&A manager reviews their own work with fresh eyes. The Harvard Business Review has documented that companies with fragmented financial data systems spend sixty-five percent more time on report preparation than those with integrated systems. Version control and error checking are a large part of that overhead.
This step takes one to two hours. The answer is now confirmed.
Step Five: Update the Dashboard
The leadership team does not want a raw Excel model. They want something that looks like a source of truth, which means the findings need to be formatted into a presentation or document that can be shared and defended. Commentary is drafted. Charts are created. The output is prepared for a non-finance audience.
This step is largely administrative. It adds no analytical value to the answer that is already known. It takes one to two hours.
Step Six: Write the Document
The sales director needs context, not just a chart. The findings need to be written up in a format suitable for the leadership conversation. A slide or document is created that presents the headline, the key drivers, and the recommendation for what the business should consider doing in response.
This is genuinely valuable work. A well-framed narrative that connects the margin signal to a commercial recommendation is exactly what finance should be producing. The problem is that it is happening at the end of a two-day process, with the Thursday leadership call approaching, rather than being the primary focus of finance's effort.
This step takes two to three hours.
Step Seven: Share the Answer
It is now Thursday morning. The analysis is shared. The sales director reviews it thirty minutes before the call. The margin compression is explained. The commercial team agrees with the diagnosis. Someone asks whether this trend has been building for longer than one month.
The answer to that question requires going back to step one.
What This Workflow Actually Costs
Seven steps. Two to three days. One answer to one question. And the question that prompted it was not complex. It was a routine request for explanation of a variance that the business needed to understand in order to make a decision.
Gartner's research confirms that only fifteen percent of FP&A teams operate with what it defines as a sustainable delivery model, one where the team can maintain consistent decision support without burning out. The seven-step workflow described above is a direct explanation of why eighty-five percent cannot. The work is not hard. The architecture is wrong.
McKinsey's finance transformation research suggests that automating finance processes can free up thirty to forty percent of team capacity. But automation applied to this workflow only makes each step faster. It does not remove the steps. The data is gathered faster. The reconciliation is automated. The model is refreshed more quickly. The answer still arrives two days after the question.
What Three Minutes Looks Like
The alternative is not a utopian vision. It is an architectural one.
When actuals are treated as a continuous signal rather than a periodic input, the explanation of a margin movement does not begin when the question arrives. It has already been forming as the movement was happening. By the time the sales director asks about margin on Tuesday afternoon, the system has been watching the revenue mix and discounting patterns since the movement started in week two. The drivers are already isolated. The context is already built.
The answer is not assembled in response to the question. It is reviewed and refined. Finance confirms what the system has already identified, adds the commercial judgment that only a human can provide, and shares a response that is grounded in three weeks of continuous observation rather than two days of reactive assembly.
Steps one through five collapse. The analyst does not gather and reconcile data because the system has been doing that continuously. The model does not need to be built because the analysis has been accumulating. What remains is the judgment layer: interpreting the signal, framing the trade-off, and recommending what the business should do next.
That is where finance's time and capability should always have been directed. The workflow was just never designed to get there.
Uptio is an AI-native FP&A decision layer that connects to ERPs and transactional source systems, watches actuals continuously, and builds context before finance is asked. The seven-step workflow becomes one step: review, judge, and advise. Learn how Uptio works.