What a Business-Unit Finance Team Actually Does All Day

The Function Nobody Explains

FP&A stands for financial planning and analysis and it's the corner of finance that students hear the least about despite being by staff the largest finance function within most operating companies. While investment bankers advise on deals and accountants keep the official books FP&A looks ahead: budgeting and forecasting analyzing how the business is performing in front of them and partnering with operational leaders to decide what to do next. At a company the size of Amazon theWork is distributed across business unit finance teams embedded within each part of the business cloud finance device finance operations finance store finance with a central planning group that consolidates everything into a single corporate view. The job description for an Amazon finance position reads almost like a definition of FP&A: quantify new business ideas identify capital investment requirements and cost reduction opportunities create detailed reports and communicate results to business partners

Why It Matters More Than Ever in 2026

This job has rarely had so much at stake. Amazon has committed to spending roughly $200 billion in capital in 2026 and reported about $44 billion in the first quarter alone. Each of those dollars starts as a business case that someone in finance built tested and defended. When a company is deploying capital at that speed the analyst who can tell which investment clears the return hurdle and which only looks attractive until the assumption of return is isolated.That depends is the person who is silently shaping the direction of the company

The Planning Cycle

The heart of FP&A is the operating plan and its regular updates. Amazon runs a famously structured planning process built around its operating plan cycles in which each team builds a plan from the bottom up: the revenue and cost drivers it expects the headcount it needs the capital it requests and the specific objectives it agrees to measure itself against. Once established that plan becomes the yardstick for the year. As actual results come in Finance compares them to the plan in what's called performance analysis.variation and then explain the gap: was the error due to volume price mix or cost?

A Worked Example: Decomposing a Miss That Was Actually a Beat

That legend describes the skill. Here is the skill in the smallest possible example because the mechanics are the central point and are rarely shown

The plan. Sell ​​100,000 units at $50 to obtain $5,000,000 in income

Real. It sold 95,000 units at an average of $52 for $4,940,000

Revenue was $60,000 lower than expected down 1.2 percent. That's what the summary report says and if that's where the analysis ends it tells the business unit that it had a slightly disappointing period

Now divide the variation into its two causes. The convention is to keep one variable in the plan while flexing the other

The volume change is the change in units valued at the planned price: 95,000 minus 100,000 is negative 5,000 units multiplied by $50 which gives negative $250,000

The price change is the change in price valued in real units: 52 minus 50 is $2 multiplied by 95,000 units which gives positive $190,000

Add them up: 250,000 negatives plus 190,000 equals 60,000 negatives which exactly reconciles the reported error

The 1.2 percent revenue loss is actually a 5 percent volume shortfall largely masked by a 4 percent price increase. Those are completely different situations with completely different answers and the headline number describes neither

Now take it to the benefit which is where the diagnosis is reversed. Suppose the contribution margin is 40 percent that is $20 per unit at the planned price

The lost volume costs 5,000 units multiplied by $20 which represents a $100,000 contribution

The price increase adds $2 for each unit sold and a price increase carries no additional cost so it all comes down to $95,000 times 2 equals $190,000 contribution

The net contribution is $90,000. up flat

The period reported as a revenue loss was a profit beat. The company sold fewer units at better prices and made more money doing it which may be exactly what it should continue doing and the summary report would have told the trader to go look for volume

That investment is the most useful thing an FP&A analyst produces and it requires four lines of arithmetic. These are illustrative figures and the actual decompositions add variations in mix currency and costs but each of them is the same measure: keep everything else in the plan and isolate one factor at a time

Driver-Based Models

A good FP&A is not last year's spreadsheet plus five percent. It is based on drivers: revenue is modeled as units times price costs are modeled as volume times cost per unit and each line is tied to an operating factor that the company can genuinely influence. That structure is what allows a financial partner to respond to a leader who asks what happens if volume falls ten percent below plan and respond in seconds instead of days. It's alsoThat's why fluency in Excel from pivot tables and lookups to the occasional bit of automation is just a game for the position. The tools aren't the point but you can't think without them and the analyst whose model can adapt to a new question in real time is the one invited back into the room

Business Partnering

The highest-value FP&A work is partnering directly with the business. That means sitting down with a leader to quantify a new idea building the business case for an investment looking for a cost-reduction opportunity and translating all that into a recommendation that leadership can actually act on. The person in finance is the one in the room who can say how much a given decision is really worth and what assumption it depends on. It's influence without authority: you don't run the business but you shape its decisions through the credibility of your analysis. If doneWell the financial partner becomes the person without whom an operator will not be able to make an important call

Case Study: What Happens When Nobody Pushes Back on the Target

Planning is often described as a technical exercise. Wells Fargo is the reason to take it seriously as a governance function because that's what a planning process looks like when the goal survives contact with reality and nothing else does

The bank had built its strategy around cross-selling the practice of getting each customer to have more products. It set an internal goal of eight products per household promoted it internally with a slogan built around the number eight and dispersed that goal across the branch network in individual employee quotas with daily monitoring and consequences for not meeting them

The target was chosen because the metric correlated with profitability. What no one seemed to check was whether it was possible to achieve it through the behavior the bank really wanted or whether the average household even needed eight banking products

The employees achieved the goal in the only way available. For several years accounts and credit cards were opened in huge quantities without customer authorization. Approximately 5,300 employees were fired for this before the practice became public

In September 2016 regulators fined the bank $185 million. The CEO resigned the following month. In February 2018 the Federal Reserve imposed an asset limit effectively prohibiting the bank from growing a nearly unprecedented penalty that remained in place for years and cost far more than any fine. In February 2020 the bank agreed to pay $3 billion to resolve criminal and civil investigations

Read that sequence as an FP&A failure and the lesson will land differently than the usual ethical framework. A number was established cascaded and measured. No one in the goal-owning chain asked the two questions that the previous section describes as working: what has to be true for this goal to be achievable and what behavior this metric actually incentivizes

The analysis of variance would have shown that the objective was met. That is exactly the problem. A planning process that only asks if the number was reached and never if the number was correct is a mechanism for making anything that measures

Deciding Under Ambiguity

Real FP&A questions almost never come with clean data. Should the company increase capacity in a region with three-quarters of a messy incomplete history? Is a sudden increase in costs structural or temporary? The defining skill is making a defensible recommendation despite gaps explicitly stating your assumptions limiting the likely range and being clear exactly what new information would change your answer. That comfort in operating under ambiguity is what separates a financial analyst from a calculator and it's the quality that topsLeaders remember long after they have forgotten any forecast

Where the FP&A Self-Image Overreaches

I've described this feature in flattering terms which is how it describes itself.Four honest ratings

Forecasts are often wrong and the process is expensive. Large companies spend enormous amounts of time on annual planning cycles that produce figures replaced in a quarter. There is a real argument seriously made by professionals that the effort is disproportionate to the accuracy and that rolling forecasts and scenario ranges would provide more decision value for a fraction of the work

Driver-based models create false precision. A model with forty inputs appears more rigorous than one with four and is often less precise because each additional assumption adds error and gives the appearance of care. The confidence produced by a detailed model in a room is not evidence about the world

Business partnerships can mean streamlining. The role is described as shaping decisions through analysis. The awkward version is that often a senior manager comes in having decided and the finance partner's real job is to build the case. Everyone in the function knows this happens and almost no role description recognizes it

Routine parts are being automated rapidly. Variance reporting consolidation and standards analysis are precisely the tasks that systems now do well. That doesn't eliminate the feature but rather moves the value entirely to the judgment layer meaning that the arithmetic in the example above is becoming the input price rather than the contribution

My opinion is that diagnosis and backtracking are the lasting parts of this job and that anyone entering it should develop these rather than spreadsheet speed

How I Would Approach the Work

I'm more of a student than a practitioner that's how I would try to operate rather than an insider report

I would always break down before explaining. The worked example above is discipline: never report a variance without splitting it because splitting changes the recommendation about half the time and adding almost always disguises two things moving in opposite directions

Second I would write down what would change my answer before submitting it. A recommendation with an overriding condition declared is much more useful than a safe one and makes the update a normal event rather than an embarrassment

Third I would model in ranges and start with the sensitivity instead of the point estimate. No one in an operational meeting needs three decimal places. They need to know which two assumptions the answer really depends on

Fourth I would address each goal I helped set with the Wells Fargo question attached: If someone had to hit this number and couldn't do it honestly what would they do instead? A goal that no one tests like that is a risk created by the finance function

Fifth I would keep a written record of my own forecasts and check them later because the only way to know if your judgment is good is to rate it and almost no one does that

Why It Is a Real Career, Not a Consolation Prize

FP&A and business unit finance are places where you learn how a company really makes money at the line-item level where you work shoulder-to-shoulder with operators and senior leadership rather than on the outside and where analytical skills accumulate quickly because you see the consequences of your recommendations on next quarter's results. For anyone who would rather understand companies than shape them from a distance it's one of the best jobs in finance and increasingly a launching pad into operational and strategic roles that you rarely reachpure advisory work

The Bottom Line

FP&A is the largest finance function in most operating companies and the least explained. It makes the plan measures the business against it and partners with operators to decide what to do next and in a company committing roughly $200 billion of capital in 2026 every single one of those dollars starts as a business case that someone in finance had to defend

The main skill is divided into four lines. A period selling 95,000 units at $52 against a plan of 100,000 at $50 reports a revenue loss of $60,000 which breaks down into a volume deficit of $250,000 and a price increase of $190,000. Take it to a 40 percent contribution margin and the volume loss costs100,000 while the price gain totals 190,000 so the reported loss was a profit of $90,000. The number on the summary described the opposite of what happened

Wells Fargo is a reminder that setting a number is as important as measuring it. A target of eight products per household cascaded in installments produced approximately 5,300 layoffs a $185 million fine in 2016 a Federal Reserve asset limit in 2018 and a $3 billion settlement in 2020 and variance analysis would have shown the target was met every time. Prepare the reportIt's the easy part. The diagnosis and the willingness to question the goal is the real work