Month end in a regulated firm has a rhythm to it, and a lot of that rhythm is people checking whether two numbers agree. One set of figures goes to the regulator and the auditors. Another set goes to the executive team and the board to run the business. They are supposed to describe the same company over the same period. In my experience they rarely do without help, and the help is a group of finance staff spending days making them line up.
So here is my thesis in one line. Regulated firms end up maintaining two reporting stacks, one built for regulators and auditors and one built for running the business, and the manual work of keeping them aligned is hiding a much larger risk than the hours it consumes.
I work on Power BI and Microsoft Fabric reporting for UK organisations, and this pattern is not unique to financial services. Regulated firms feel it hardest, though, because one of the two stacks has a regulator on the end of it. Let me explain where the two stacks come from, what they actually cost, and what closing the gap looks like in practice.
Nobody designed this
Ask a finance director why the firm has two reporting stacks and the honest answer is usually… “it just happened”.
The regulatory stack tends to come first. It exists because a regulator or an auditor asked for specific figures in a specific format on a specific date, and someone built a process to produce them. That process takes a copy of the data from the core systems at a fixed point in the cycle, applies a set of rules, and produces the submission. It is usually careful, slow and hard to change, which is roughly what you want from something a regulator reads.
The management stack grows up alongside it because the people running the business need answers faster than the regulatory cycle can give them. Different team, different copy of the data, often taken on a different day. Definitions get set independently, because the person building the management report is solving a management problem and has no reason to open the regulatory rulebook.
Then the drift starts. A product gets reclassified for a regulatory return and nobody updates the management view. A new fee line appears in the business dashboard because a commercial director asked for it, and the regulatory team never hears about it. Each change is reasonable on its own. Nobody has to make a bad decision. It accumulates quietly, one sensible edit at a time.
Sit with a team walking through why two reports disagree and the striking thing, in my experience, is that most differences have a perfectly good explanation. Timing. A definition. An adjustment that lives in one stack only. The two sets of numbers are both “right” by their own rules. They just are not the same rules.
The hours are the smaller problem
The obvious cost is the reconciliation itself. Often days of skilled people’s time every month, explaining why a figure appears twice with two values. Firms tend to measure this cost, because it is easy to measure, and conclude that it is annoying but manageable.
I think that misses the real cost, and I’ll put it plainly. When a business spends its month end reconciling numbers, it has no time left to interrogate them.
Think about what the finance team is doing during those days. They are proving that the management figure and the regulatory figure can be bridged. Nobody is asking whether the management figure tells the executive team anything useful, or whether the trend underneath it has changed. The energy goes into agreement… and the questions that would actually move the business get pushed to next month, then the month after.
There is a second cost, and in my view it is the more dangerous one. The knowledge of how the two stacks map onto each other usually lives in the heads of two or three people. Sometimes it is one. In our experience the “bridge” between regulatory and management reporting is often a spreadsheet, maintained by the same individual for years, with tabs nobody else opens. That person is the process.
You are probably thinking, we have controls for that, we have handover procedures. In my experience the handover works for the mechanics and fails for the judgement. A successor can rerun the spreadsheet. What they cannot do is explain, under questioning, why line 14 is treated differently from line 15, because the reason was never written down. It was known. Nothing broke. The figures were eventually explained. But for those weeks the firm was running on numbers nobody could fully defend, and the people who mattered did not know it.
Closing the gap without pretending the stacks are the same
The fix is simpler to describe than to do.
One governed set of definitions, agreed once, feeding both purposes. Revenue means one thing. A customer means one thing. A product category means one thing. Those definitions live in a place both the regulatory team and the management reporting team draw from, so that when a figure appears in both places it has been calculated the same way, from the same source, at the same point in time.
Now, the objection I always hear. “But regulatory reporting genuinely is different. The regulator wants things treated in ways the business does not.” That is true, and it is the important bit. Some divergence is legitimate. The point is that it should be deliberate and written down. If the regulatory return has to exclude a category that the management view includes, that exclusion should exist as a documented rule everyone can see, applied automatically, with the difference between the two figures visible on demand rather than rebuilt by hand every month.
In practice this is a data modelling job before it is a reporting job. When we work on Power BI for financial services at Metis BI, a UK Power BI and Microsoft Fabric consultancy, the first thing I go looking for is that bridging spreadsheet, because it tells me where the definitions actually live today and how far the two stacks have drifted. What replaces it is a shared semantic model, which is the term Power BI uses for a governed layer that holds the business definitions and calculations in one place, so that every report, regulatory or management, reads from the same set of rules. The regulatory view and the management view become two windows onto the same numbers, with the differences between them named and explainable.
I like to say the whole job comes down to being able to point at the difference.
The trap
Here’s the thing that usually goes wrong at this point.
A firm recognises the problem, commissions a new reporting project, and a year or two later it has three stacks instead of two. The new dashboards look better. The old regulatory process still runs because nobody dared touch it, the old management pack still runs because the board is used to it, and the new solution sits alongside both, with its own copy of the data, its own timing and its own definitions.
It is more common than it sounds. In one financial services estate review we carried out, more than ten redundant semantic models ended up being consolidated, each one a slightly different answer to the same questions. It happens because the project was framed as “build better reports” when the actual problem was “we have two definitions of the same thing”. More reporting on top of unsettled terms multiplies the reconciliation rather than removing it. If the definitions are not agreed first, every new report is another version of the truth waiting to drift.
Summary
Regulated firms did not choose to run two sets of numbers. It accumulated, one reasonable decision at a time, and now month end is spent proving that figures which should already agree can be made to. The hours are real, but they are the small cost. The larger one is a business running on figures nobody has time to question, held together by a mapping that lives in two or three heads. Closing the gap means one governed set of definitions feeding both purposes, with the legitimate differences made deliberate and written down. Until that happens, adding more reporting only gives you another set of numbers to reconcile.