There is a version of the budget meeting that exists in the calendar, and there is a version that exists in reality. The first is ninety minutes long. The second was four minutes, last Tuesday, between the elevator and the kitchen. The first will produce a deck. The second produced the answer.

If you have ever sat through a budget meeting and wondered why nothing seems to be being decided, this is why: it has already been decided, and the meeting is the announcement, served as a discussion. It is theater. Acknowledging that it is theater is the first move of a person who is going to be useful in it.

I. The audience of one.

Every budget meeting has, somewhere in the room, the person whose head needs to nod. Usually the CEO. Sometimes the chair of the board's finance committee, on a Zoom rectangle in the corner. Occasionally a founder who is no longer operationally involved but who funds the company personally, in which case the entire meeting is for them, regardless of who is speaking.

Identify this person before slide one. Every line item, every chart, every gentle correction is being delivered to that person and only that person. Everyone else in the room is performing for them too, and pretending not to. The CFO who understands this is calm in budget meetings. The CFO who doesn't is the one talking to slide 14 about cohort retention while the audience-of-one is checking their phone.

It has already been decided, and the meeting is the announcement, served as a discussion.

II. The decision is upstream.

The number that lands at the top of the budget — the headline growth assumption, the EBITDA target, the headcount cap — was set somewhere else. Usually a one-on-one between the CEO and the CFO, four to six weeks before the meeting. Sometimes a board call. Almost never a budget meeting.

What the budget meeting does is wrap that number in evidence. The forty-three slides are the evidence-wrap. The graphs are evidence. The cohort triangles are evidence. The benchmark callouts are evidence. The argument is not being made; the argument is being decorated. People who attend their first few budget meetings assuming the argument is live tend to make two kinds of mistakes — they argue back, which is loud and embarrassing, or they fall silent, which reads as agreement and gets quoted in Q3.

The third option, the one experienced operators use, is to ask one specific question about a number on a non-headline slide. "On slide 27, the EMEA pipeline coverage — is that the new methodology?" This signals you have read the deck, you respect the work, and you are not trying to relitigate the number. The audience-of-one notices. You have earned a chip.

III. What is actually happening.

The meeting is doing four things at once, and the deck is only doing one of them. The deck is presenting the plan. The other three things — the alignment of the senior team, the political laundering of decisions that were made privately, and the calibration of who in the room is going to push back on what — are happening in body language, in side-eye, in the order in which people speak.

If the VP of Sales nods early on the growth slide, the number is safe. If they look down and adjust their notebook, the number is going to come up again in a one-on-one, and the CFO will eventually agree to soften it, and the deck will be updated, and nobody will mention that the deck was updated. This is what is meant by "a productive budget meeting."

IV. Why we keep doing it.

If the meeting is theater, the obvious question is why anyone bothers. The answer is that it produces three things you cannot get any other way: a shared artifact (the deck, now blessed), a shared timestamp (the budget exists as of today, regardless of what was decided before), and a shared narrative (we discussed it; we agreed).

The artifact, the timestamp, and the narrative are useful in roughly that order. The artifact protects you in diligence. The timestamp protects you in board minutes. The narrative protects you in the all-hands, when the engineer asks why headcount was cut, and the CEO can say, with sincerity, that the team aligned on it.

None of this is dishonest. It is the difference between deciding and committing. The decision is upstream. The meeting is the commitment. The deck is the receipt.

A footnote, for the babies.

If you are new and you are reading this and you are thinking, "this seems cynical," we would like to gently point out that you are correct and also that it does not matter. Theater does not mean fake. It means that the form is doing work the form is not advertising it is doing.

The hallway conversation could not have happened without the meeting on the calendar. The meeting on the calendar could not have happened without forty-three slides of evidence. The evidence could not exist without an FP&A analyst who has been awake since 5:14am. You are part of this. Welcome.

Janine has been the controller at the same Series C SaaS company for four and a half years. She has closed sixteen quarters there. The first ones were rough. The last ones have been quiet. Quiet, in finance, is the highest compliment a function can be paid. It means the numbers arrived on time, in the right buckets, and nobody had to be woken up.

This is the story of the quietest quarter she ever almost ruined. She did not, in the end, ruin it. She fixed it, in fourteen lines, on a Sunday evening in October, by herself, and went to bed. Nobody at the company will ever know it happened, with two exceptions: Janine, and her CFO, who learned about it two months later, by accident, in a one-on-one that had been scheduled to talk about headcount.

I. The error.

Three quarters earlier — Q4 of the prior year — the company had restructured its cost-center hierarchy. There were good reasons. The board wanted product-level P&Ls. The new CFO, six weeks in the seat, wanted to be able to answer the question "how much do we spend on Platform vs. Apps," and the old chart of accounts could not.

Restructuring a chart of accounts is a thing that takes two days to design, three weeks to implement, and approximately fourteen months to actually finish, because every recurring journal entry, every accrual schedule, every payroll mapping has to be re-pointed, and the recurring entries are buried in seventeen different systems and four different people's heads.

One of the recurring entries was the bonus pool accrual. It is monthly. It is large. It had been mapped, in the prior structure, to a parent cost center called "G&A Compensation," which made sense in the old world. In the new world, it should have been split across each product P&L, weighted by department headcount, so that the platform team's bonus accrual hit the Platform P&L and the apps team's hit Apps.

It wasn't. For three quarters. The entire bonus accrual — a number with seven figures — landed in G&A every month. The product P&Ls looked artificially lean. The G&A line looked artificially fat. The board, looking at the product P&Ls, concluded that Platform was a better business than Apps. They were not wrong about that conclusion, in the end. They were wrong about how much better.

The fix is not the story. The noticing is the story.

II. The noticing.

Janine noticed it on a Sunday. She was not working. She was paying her own bills, on her own laptop, on her own time, which is when many controllers do their best thinking. She had been mentally previewing the Q3 close, the way some people mentally preview a long drive. She remembered that the new product P&Ls were going to be presented to the board for the first time, in detail, in two weeks.

And she remembered, with a quality of memory that is hard to describe to anyone who has not done a close, that the bonus accrual had been touched in the chart-of-accounts migration, and that nobody on her team had ever confirmed the mapping after the fact, and that the only way to confirm it would be to look at the journal entry detail for the past nine months and check what cost center the offsetting credit was hitting.

She closed her laptop. She made tea. She opened her laptop again.

III. The fix.

It took her two and a half hours. The first hour was confirming the error. The second hour was modeling the correction — nine months of restated accruals, six product P&Ls, the cumulative true-up. The half hour was writing the journal entry itself. Fourteen lines. Debit each product P&L compensation account by the appropriate share of the cumulative true-up. Credit G&A Compensation by the total. Memo: "Q1-Q3 reclass — bonus accrual cost-center realignment, ref CoA migration Dec 2024."

She booked it as a Q3 adjustment, with the cumulative catch-up isolated on a single line so the variance commentary could explain it cleanly. She left a note for her senior accountant to review the entry on Monday morning. She closed the laptop. She went to bed.

The Monday review took twelve minutes. The senior accountant agreed with the fix. The variance commentary was updated. The Q3 product P&Ls, the ones the board saw, were correct.

IV. What was saved.

If Janine had not noticed, several things would have happened. The board would have seen product P&Ls that overstated Platform margin by a meaningful amount. They would have made strategic decisions on that basis — probably to lean further into Platform and slow hiring in Apps. The Apps general manager, who was already being quietly squeezed, would have been squeezed harder.

Some time later — months, maybe four quarters — the error would have been caught, almost certainly by an auditor preparing the year-end financials. The fix at that point would have been larger, more public, and would have required restating numbers the board had already used. There would have been a memo. The CFO would have written it. The board would have read it. Confidence in the finance function would have absorbed a small, permanent dent.

None of that happened. Instead, a single line in a Sunday-evening journal entry caught nine months of drift, and the close went out on time, and the board saw clean numbers, and the strategy meeting two weeks later was the right strategy meeting.

V. What this means.

There is a kind of work in finance that is the work nobody mentions. It is not in the deck. It is not in the all-hands. It is not in the quarterly business review. It is in the journal entry detail, in the chart-of-accounts mapping, in the recurring schedules that someone has to be paying attention to or they drift, slowly, in a direction nobody intended.

The people who do this work are, on average, the most under-thanked people in your company. They are also the reason your numbers are real. If you are a CFO, find your Janine. Ask her, by name, what she fixed last quarter that you don't know about. If she tells you, that is the most important fifteen minutes of your week.

If she tells you, also: do not put it in a deck.

There is a recurring fantasy among newly arrived heads of finance, particularly those who joined from a top-tier consulting firm or a recent MBA program, that the existing model can be replaced. The new model will be elegant. It will be one tab. It will have a beautiful dashboard. It will have dynamic arrays and the kind of formula architecture that produces compliments at conferences. It will, the new head of finance is convinced, finally let everyone see the numbers.

We are writing today, with appropriate gentleness, in defense of the four-tab model. The boring one. The one with the named ranges from 2017 nobody can find the definitions for. The one that has four tabs because four tabs is what is necessary, and not five, and not seventeen, but four.

I. What the four-tab model actually is.

The four-tab model is the format that emerges, over years, when a finance function actually has to use a model rather than present it. It has, in order: a tab called Inputs (or Assumptions, in some houses, or just Drivers); a tab called Calc (or Model, or, in older specimens, Sheet2); a tab called Output (or Dash, or simply P&L); and a fourth tab that exists because there is always a fourth tab, and it is usually called Scratch or Working or, in deeply lived-in models, just a person's first name.

The four-tab structure is not designed. It is precipitated. The first tab exists because you need a single place to change a number. The second exists because the calculations got too long for the first. The third exists because someone, at some point, needed to send the CEO a summary without sending the whole file. The fourth exists because finance work has a residue.

The four-tab structure is not designed. It is precipitated.

II. Why the elegant one-pager fails.

The new head of finance arrives, looks at the four-tab model, and proposes a single-tab replacement. Inputs at the top. Calculations below. Outputs to the right. It will all fit on a single laptop screen, they say. It will be self-documenting. They have seen a version of it in a screenshot on finance Twitter.

Three things will happen in the next four months. First, the single-tab model will be built. It will be beautiful. The new head of finance will use it in their first board meeting and the board will compliment them on it. Second, the single-tab model will encounter a real-world ask, such as "what does the P&L look like if EMEA pipeline closes 18% lower in Q4," and the answer will require a sensitivity that does not fit on a single tab. A fifth tab will be added, called Sensitivity, which the new head of finance considers a temporary exception.

Third, six months in, the model will have ten tabs. The new head of finance will, by this point, have stopped commenting on the architecture. They will have learned what the previous four heads of finance learned, which is that the four-tab model is not the form that finance models take when they are well-built. It is the form they take when they are used.

III. The aesthetic objection.

There is an aesthetic case against the four-tab model, and we want to take it seriously. The aesthetic case is that good thinking produces clean artifacts, and a model with named ranges from 2017 and a tab called Janine is not, in any standard sense, clean. We agree that it is not clean. We disagree that this is a problem.

The clean spreadsheet is a spreadsheet that has not been used yet. Every spreadsheet that has done real work in a company has been opened by twelve people, each of whom needed a slightly different cut of the numbers, and each of whom left a small piece of themselves in it. The tab called Janine is not a mess. The tab called Janine is evidence that Janine, three years ago, needed to know what the Q3 numbers looked like if the Series B was delayed by a quarter, and the model accommodated her.

A model that has not accommodated anybody is a model that has not been used. We would rather have a model that has been used.

IV. What we are not saying.

We are not saying that the four-tab model cannot be improved. It can. We are not saying that legacy named ranges, broken links, and circular references are virtues. They are not — they are debts, and they should be paid down on a schedule that the rest of the function can absorb without losing a close.

What we are saying is that the impulse to throw out the model and start over is, more often than not, an aesthetic impulse dressed as an analytical one. It feels like rigor. It is, usually, the opposite. Rigor is sitting with the existing model, understanding what each tab is for, and adding the fifth tab only when it is the smallest possible change that answers the question.

V. A note for the new head of finance.

If you are the new head of finance, and you are reading this, and you are looking at the four-tab model on your desktop right now, we would like to make one suggestion. Before you replace it, use it. For a quarter. Open it every day. Touch each tab. Find out which formulas are load-bearing and which are vestigial. Find out who knows what.

Then, at the end of the quarter, if you still want to replace it, replace it. We will not stop you. We will, however, gently predict that you will end up with four tabs.