Why Most Financial Analysis Stops Too Soon
A finance team closes the month. Revenue was down 8% versus budget. Gross margin compressed 2.4 points. A report goes to leadership with the variances highlighted in red.
And then, nothing changes.
Not because leadership ignored the report. Because the report answered the wrong question. It answered what happened. What leadership needed was why it happened, and what to do about it.
This is the gap that separates reporting from analysis, and analysis from genuine financial intelligence. It can be described as three distinct layers of FP&A work. Most finance functions operate almost entirely in the first. The ones that create real business value move through all three.
Layer 1: Reflection: What Happened?
Reflection is the foundation. It is the accurate, structured recording and presentation of financial results, revenue, costs, margins, cash, versus budget and versus prior period. Done well, it is essential. Done alone, it is insufficient.
Reflection answers questions like: Revenue was $2.1M versus a budget of $2.3M, what were the main variances? Gross margin was 34% versus a budgeted 37%, where did the gap come from? Cash balance at month end was $180k versus the $240k expected, why?
The output of Layer 1 is typically a management report: a P&L variance summary, a budget-versus-actual comparison, a cash flow statement. These are necessary. Every serious finance function produces them.
The problem is that Reflection creates the illusion of analysis. The numbers are organised, the variances are labeled, the report looks thorough. But the reader still has to guess at the cause, and most of the time, they do.
Your reports describe the outcome in detail but leave the cause unexplained. Leadership reads the numbers and asks follow-up questions. Those questions should have been answered in the report.
Layer 2: Diagnosis: Why Did It Happen?
Diagnosis is where analysis actually begins. It takes the variances identified in Layer 1 and works backward, decomposing them into their underlying drivers to find the real cause.
This is harder than it looks. Most variances have multiple contributing factors, and attributing the whole gap to a single cause is almost always wrong. A revenue miss might reflect lower volume, worse pricing, a shift in product mix, or some combination of all three, and the management action for each is completely different.
The standard tool for this is variance decomposition: separating the effects of price, volume, mix, and cost from each other. A gross margin shortfall, for example, might look like a cost problem when it is actually a mix problem, the business sold more of its lower-margin products, and unit costs stayed exactly on budget.
A mid-size manufacturer finishes the quarter with gross margin 2.4 points below budget. The Layer 1 report says: "Gross margin of 34.1% versus budget of 36.5%, unfavorable variance of $186k."
Layer 2 asks: what drove that $186k? A proper decomposition reveals:
| Driver | Variance | Cause |
|---|---|---|
| Price effect | ($74k) unfavorable | Sales team discounting to protect volume, average prices 1.2% below budget |
| Mix effect | ($89k) unfavorable | Shift toward lower-margin capacity variants, invisible in the top-line revenue figure |
| Cost effect | $23k favorable | Unit costs running marginally ahead of budget, partially offsetting the above |
| Total variance | ($186k) unfavorable | The problem is pricing and mix, not cost |
The diagnosis changes everything. The problem is not cost, costs are actually running slightly ahead of target. The problem is pricing discipline and product mix. Those require a sales conversation, not an operations review. Without Layer 2, leadership has a number. With it, they have a direction.
You can explain what caused the variance, but the conversation ends there. No one asks: given what we now know, what should we do, and what happens to the numbers if we do it?
Layer 3: Decision Modeling: What Should We Do?
Decision Modeling is where FP&A becomes genuinely strategic. It takes the diagnosis from Layer 2 and builds forward, modeling the financial impact of the choices available, so that decisions are made on analysis rather than instinct.
This is the layer that most finance teams rarely reach, not because they lack the capability, but because Layers 1 and 2 consume all available time. The month-end close, the variance commentary, the board pack: by the time those are done, there is nothing left for forward-looking modeling.
The output of Layer 3 is not a report of what happened. It is a structured answer to a specific decision question, with numbers attached to each option.
"A good decision model ends with a recommendation, or at minimum, with the information needed to make one."
The diagnosis is clear: the margin problem is driven by pricing and mix, not cost. Layer 3 asks: what are the realistic options, and what does each one look like financially?
The decision is now a financial conversation, not a gut-feel one. Leadership can see the tradeoffs, the risks, and the magnitude of each option, and make a choice with their eyes open.
Your analysis ends with a recommendation, not a description. Leadership leaves the meeting knowing what the decision is, not just what the numbers are.
How to Connect the Layers in Practice
The three layers are not separate workstreams. They are a sequence, each one builds on the last, and skipping a layer produces a gap that shows up as a poor decision downstream.
Start with clean Reflection
A well-structured budget-versus-actual comparison, with variances clearly labeled and nothing buried. This is the input to everything else. If the numbers are wrong or incomplete, the layers above collapse.
Move immediately to Diagnosis
Do not stop at the variance number. Ask what drove it. Decompose it into its components, price, volume, mix, cost, timing. The goal is to get from "margin was down $186k" to "margin was down $186k because of X, Y, and Z, and Z is the one that matters."
Build the Decision Model around a specific question
Not "what should we do about margins?", that is too broad. "Should we restore price floors on Product Line A, and if so, what volume loss is acceptable?" That is a decision model. It has a specific question, defined options, and a financial output for each.
Triage, do not apply all three layers to every variance
Most months, two or three variances deserve the full three-layer treatment. The rest get a concise Layer 1 description and move on. The skill is knowing which two or three. Material variances with clear management levers attached are the ones that earn the full analysis. Timing differences and minor line items do not.
The Tools That Support Each Layer
Each layer has its own analytical demands. The right model makes the difference between an analysis that takes a week and one that takes a day.
For Reflection and Diagnosis
A complete annual budgeting model with central assumptions hub, monthly P&L, actuals input, and automatic budget-versus-actual variance reporting. The foundation of Layer 1.
Takes you from the Layer 1 variance number to the Layer 2 answer, separating price, volume, mix, and cost effects with a full reconciliation check and waterfall-chart-ready output.
For Decision Modeling
A driver-based 5-year forecast with three scenarios and built-in sensitivity analysis, showing which assumptions move the outcome most. The structure for Layer 3 thinking.
Addresses one of the most common Decision Modeling questions directly, where is pricing power in the product line, and what does the optimal strategy look like given margin and volume tradeoffs?
Extends Decision Modeling to the liquidity question, not just what the P&L looks like under different scenarios, but what happens to cash, and when does the business need to act?
A comprehensive forecasting tool for manufacturing businesses, integrating backlog, lead times, capacity constraints, and scenario planning into a single decision-support model.
The Practical Checklist
When reviewing any financial analysis, yours or someone else's, ask these questions:
Layer 1: Reflection
- Are the variances clearly stated in dollar and percentage terms?
- Is the comparison meaningful, budget vs actual, or prior period, or both?
- Is nothing buried, are the significant numbers visible at the top, not hidden in footnotes?
Layer 2: Diagnosis
- Has the variance been decomposed into its drivers?
- Is the root cause identified, not just labeled, but explained?
- Would a reader know why the number moved, not just that it moved?
Layer 3: Decision Modeling
- Is there a specific decision question attached to the analysis?
- Are the realistic options modeled with financial outputs for each?
- Does the analysis end with a recommendation, or at minimum, with the information needed to make one?
If the answer to all of these is yes, the analysis is doing its job. If the answers stop at Layer 1, the numbers are being reported, but not used.
Related Insights
What a financial analyst actually does in a company, and what really happens when the function disappears.
A worked quantitative example of Layer 2 diagnosis and Layer 3 modeling, using price elasticity and product-mix analysis to find the optimal pricing strategy.
Decision Modeling applied to the liquidity question, how a structured cash flow forecast reveals the gap between what a business expects to pay and what it actually pays.
The Diagnosis layer depends on causal commentary — explaining WHY variances happened, not just WHAT changed. This benchmark tests exactly that capability across Claude models, with a scoring rubric built around the same analytical standard.