Financial Planning and Analysis

While a company is small, one finance question covers everything: how much is in the bank right now? You look at the balance, decide whether this month survives, and you're done.

As headcount grows, decisions start showing up that the balance can't answer. When a request to hire five more engineers meets the argument that there's no room for it, and both sides have nothing but the bank balance behind them, the louder person wins. The monthly close that accounting produces tells you what happened last month; it says nothing about where five extra engineers leave you eight months out.

Building a plan once isn't enough either. The spreadsheet made in January has already drifted from reality by March, and if nobody reopens it after that, it may as well not exist. What's needed is for two jobs to sit in the same pair of hands: producing the forward numbers, and checking every month why they came out wrong so the next version is closer. Financial planning and analysis (FP&A) is the name for that seat.

Roughly half the work runs on a monthly cycle and half is answering questions that arrive whenever someone has one. The cycle half usually goes in this order.

1. Collect last month's actual numbers at the start of the month. Nothing starts until accounting closes the books, so a late close pushes the entire schedule back.
2. Line up plan against actual, line item by line item, and calculate the gap. Marketing was budgeted at $23,000 and spent $32,000, so the gap is $9,000.
3. Ask the owning team about every large gap. This is where most of the hours actually go. On paper it reads as overspending, but the answer is often that advertising planned for next month got pulled forward — which means it isn't an overrun this month, it's a hole next month.
4. Recalculate the next twelve months with whatever you learned. A financial forecast isn't a document you finish; it's a table that gets repainted every month.
5. Write the conclusion up as a single page. Burn rate, the amount of cash disappearing each month, and runway, how many months the balance lasts at that speed, land on that page almost every time.

The ad hoc half is mostly questions shaped like "what happens if we do this?" How many months of runway two more senior engineers cost. How much cash arrives earlier if the annual-payment discount is raised to twenty percent. The method is always the same: change that one input in the table you already maintain and show how the number at the far end moves. Keeping the model tidy on a quiet week is what makes a fast answer possible on a loud one, and it's why this half of the job rarely appears on anyone's calendar even though it consumes real hours.

This looks like a solitary spreadsheet job, but the numbers get gathered rather than invented, which means sitting down with most of the company once a month.

Reporting goes to the chief financial officer (CFO), or straight to the founder at a smaller company. The tables produced here become the basis for how large hiring gets and when the company goes out to raise, so the person building the numbers stays close to the person deciding with them.

Last month's confirmed figures come from the accountant, and that flow only runs one direction. Closed numbers don't get adjusted. What does get raised is the chart of accounts: if advertising and event spend sit in the same account, a gap can be calculated but never explained, so the request goes back to split them.

Team leads are effectively co-authors. A budget belongs to the team that spends it, and that team is the only place an explanation for a gap can come from. Sales comes up most often of all, because the list of open deals is the raw material for next quarter's revenue line — if that list is written optimistically, the whole table tilts with it, and the tilt only surfaces two quarters later when the cash fails to arrive.

People operations is the other standing meeting. Salaries are more than half of costs at most companies, so the headcount plan — who gets hired and when — is effectively the cost plan. Board packets pull in SaaS metrics alongside all of this.

"Let me check with FP&A and get back to you. I need to know whether this quarter's budget has room."

Whether money is still available is something only the person holding the plan knows.

"Marketing came in 40 percent over plan this month. What drove that?"

The monthly routine of calculating a gap and asking the owning team about it.

"Two more hires costs you two months of runway. Still want to do it?"

Not making the decision — just pricing it.

At Pagedeck, Grace opens each quarter with a meeting built around three versions of the plan: hold the current pace, add five engineers, add three salespeople. Each one carries a single line showing the cash balance and runway twelve months out, and the meeting starts by comparing those three lines.

Sofia takes the same note every time. The open-deal list comes in with close probabilities set high. If only three in ten actually closed over the last four quarters, that's the rate the table applies, and arguing for a better one means bringing evidence for it. This push and pull — history holding down the forecast — is the friction FP&A runs into most often.

The most common misreading is that the job is getting the forecast right. Building future numbers sounds like a task you either succeed or fail at. But nobody predicts a startup's revenue twelve months out without error, and accuracy was never the goal. The work is a loop: spot the size of the miss quickly, find the reason, correct the next version. Score the role on accuracy and what comes back is a plan sandbagged low enough to be safe, and a plan like that can't support a single decision about how many people to hire.

Reading it as an accountant who is good at spreadsheets is nearly as common. Half the day genuinely is spent in the model. But strip out step 3, the part where someone sits with each team and digs for the reason, and what's left is a page of numbers that add up correctly. The value of the seat is knowing why the numbers came out that way, not the numbers themselves.