Bahasa Indonesia

A Day in the Life of a Financial Analyst (What the JD Doesn't Tell You)

The job description said "drive insights for the business." That's a fine sentence. It's also the most useless sentence in finance recruiting, because it tells you nothing about what Tuesday actually feels like.

Tuesday feels like this. You're half an Excel jockey and half a translator between three people who don't speak each other's language: a Sales VP who wants to know why his commission accrual moved, a controller who needs your variance comments by 11am, and a CFO who slacks you at 2:14pm asking for ARR by segment by close date "by EOD if possible." Your model is open. Your inbox is open. Slack is open. So is the companion JD you read before applying. Except now you're the one living the bullet points, and the bullets left a few things out.

This is the day the JD doesn't describe. If you're an FA, Senior FA, or FP&A IC at a $20M-$300M ARR B2B SaaS, you're going to nod a lot. That's the point. Someone should name the shape of this.

8:00am — The Variance Scan

Coffee. Laptop. Adaptive Planning is already loaded because you didn't actually close it last night. You pull the daily flash and walk the P&L top to bottom. Revenue, COGS, S&M, R&D, G&A. You're looking for anything that moved more than 5% versus forecast since yesterday's snapshot.

Most days, three lines move. Two of them have an obvious reason: a deal slipped from Friday into Monday, a vendor invoice posted late. The third one is the line that's going to eat 90 minutes of your afternoon. You don't know it yet. You just know it's the one nobody can explain in a Slack thread inside ten minutes.

This morning it's S&M, up $47K versus the rolling forecast. Could be a contractor invoice. Could be a re-class. Could be the new SDR cohort that started last week and someone forgot to add the loaded cost to the headcount sheet. You drop a note in your scratch tab: S&M +$47K, investigate before exec readout. Suspect headcount load.

The Adaptive screen is the source of truth, but you're already exporting to Sheets because that's where you actually think. This is normal. Adaptive (or Mosaic, or Cube, or Pigment, depending on which planning tool the company bought during the last finance leadership change) is where the numbers live and where the controller will check them. Sheets is where you figure out what they mean.

A note for the analysts who got hired in the last six months: nobody told you in the interview that you'd spend roughly 40% of your time moving data between Adaptive and Sheets. They told you about "strategic finance" and "business partnering." Those things exist. They live in the gaps between the data movement.

9:30am — The Business Partner Standup

Standup with the GTM finance business partner, or the Eng FBP, or whoever owns the function whose numbers moved.

Today it's GTM. The VP of Sales is on the call because his AE comp accrual looks "completely different" from last month's. He's already typed this can't be right into Slack twice this morning. Your job in the next fifteen minutes is to:

  1. Find out what changed.
  2. Decide if it's a data problem or a reality problem.
  3. Tell him without making him feel small, because you'll need this guy to give you accurate forecast inputs in three weeks.

You open NetSuite. Sometimes it's Sage, depending on the company. You pull the GL detail for the comp accrual account and there it is: a re-class the controller posted on the 28th, moving $63K from "Sales Commissions: Direct" into "Sales Commissions: Channel." Nobody told you. Nobody told the VP. The total is the same; the shape is different; he's looking at the line he always looks at, and it's down.

The mini-script for this moment matters more than the model:

"Hey, quick one. The number isn't wrong, the line moved. Controller re-classed $63K from direct comp to channel comp on the 28th, so your direct line shows lower but total is flat. I'll add a footnote to the flash starting tomorrow. Want me to walk you through the channel breakout while we're here?"

You acknowledged the panic. You named the cause. You gave him a souvenir (the channel breakout). You did not say "the number is correct" because the number being correct is not the point. The point is that he can't explain it to his SVP, and now he can.

The first time I missed a re-class on something like this, I spent two hours rebuilding the comp model from scratch trying to find a "bug" that didn't exist. The bug was a journal entry I didn't read. Read the journal entries.

11:00am — Mid-day Model Build

This is the part of the day the JD was actually describing when it said "drive insights." It's also the shortest part of the day, on average.

You get about two real hours of building. Today it's a headcount roll-forward for the FY27 plan refresh. The CHRO sent over an updated hiring plan last Friday, and the CFO wants to see the loaded cost impact by department before Friday's exec sync. Sheets is your scratchpad. Adaptive is where it has to land. The math has to tie out at three levels: total comp, total opex, and ending headcount by department.

You build the roll-forward as a tab. Column A is current period headcount. Column B is starting salary (loaded with benefits, ER taxes, equity expense). Column C is the hire date. Column D is the prorated cost for the period. You write a SUMIF that pulls the department roll-up. You color the assumption cells yellow. You put a "last updated" timestamp at the top of the tab.

You also save the file as headcount-rollforward-FY27-v3-WIP.xlsx. Then v4. Then v4-clean. Then v4-clean-FINAL. Then v4-clean-FINAL-shareable.

Version control on .xlsx files is, genuinely, the single biggest unsolved problem in FP&A. Adaptive has version history. Sheets has version history. The hand-off between the two does not. Every FP&A team I've worked on has lost at least one afternoon to "wait, which version did the CFO see?" There's no good answer here. The best teams I know have a naming convention they enforce religiously and a single shared drive folder per planning cycle, and they accept that on month four someone will still email a v2 by accident.

1:30pm — Lunch (Optional)

Twenty minutes at your desk, eating something that came out of a container. You scroll Slack. The CRO posts a screenshot of a Looker dashboard with a question mark emoji. You don't respond yet. The question mark emoji is bait.

2:14pm — The Ad-Hoc That Eats the Afternoon

Slack from the CFO:

"Hey, can you get me ARR by segment by close date by EOD? Need it for the prep call tomorrow morning."

The Looker dashboard for this exists. You know it exists because you helped build it nine months ago. You also know it's "kinda wrong." The segment cuts use a customer-tier mapping that hasn't been updated since the new pricing tiers launched in February, and the close date logic uses contract effective date instead of booking date, which is fine for some questions and very wrong for others.

The temptation is to say "yes" and start pulling. Don't. The 15-minute scoping question saves your afternoon.

You reply:

"On it. Two quick scoping qs so I get this right the first time:

  1. Segment = current tier mapping (Feb '26+) or original SMB/MM/Ent buckets?
  2. Close date = booking date or contract effective date? Default I'd use: current tier + booking date, last 6 quarters. Sound right? ETA 4pm."

Three things just happened. You bought yourself permission to use your defaults. You signaled that you understand there's more than one valid answer. You set an ETA before they could set one for you.

The CFO replies "current tier + booking, sounds right, thx." You now have a real estimate, a real scope, and a real deadline. You also know the deliverable is Sheets, not a deck, because the CFO said "for the prep call," which means she's going to put it in front of someone else and walk them through it live. A deck would be friction. A clean tab with a chart up top is the right shape.

You pull the raw export from NetSuite (or Sage). You cross-reference the segment mapping in Salesforce. You build the cube in Sheets. You add a chart. You sense-check the totals against the Looker dashboard you don't trust, and the gap is 2.1%, which is the size of the segment-mapping drift you predicted. You note it: delta vs Looker dashboard = 2.1%, driven by Feb '26 tier remap. Looker not yet updated.

Sent at 3:51pm. CFO replies "perfect" with a thumbs up. The 2pm Slack is the most common "where did my afternoon go" pattern in FP&A. The 15-minute scoping question is the single highest-leverage habit you can build.

4:00pm — The Long Tail

You context-switch back to the headcount roll-forward, except now there's a Slack from the GTM business partner asking about the channel comp split from this morning, and the controller wants to know if you've reconciled the S&M variance, and you haven't.

This is the hour where the day's real cost shows up. Every interruption costs you ten minutes of re-loading the model in your head. You have to pick. Today you pick the S&M variance, because the controller's deadline is harder than the GTM partner's curiosity. You'll send the GTM partner a note at 5:30 that says "circling back tomorrow morning, before standup."

The S&M variance turns out to be a contractor invoice that hit two weeks late and got coded to the wrong cost center. Twelve minutes of digging in NetSuite. You write the variance comment:

S&M +$47K vs forecast: $52K contractor invoice (Q1 SDR ramp coaching, vendor: BlueArrow) coded to S&M-Direct instead of S&M-Programs; ($5K) timing on travel reimbursements. No underlying spend issue.

Two sentences. The second sentence is the one the CRO will actually read.

5:00pm — End-of-Day Model QA

This is the boring part. It's also the part that decides whether you keep your job.

You go back to the headcount roll-forward and run the QA pass:

  • Tie-out check: total opex in the model = total opex in the Adaptive snapshot you used as a starting point, plus the deltas you intentionally introduced. If they don't match, you have an unintentional change. Find it.
  • Sum-check totals: the department roll-up totals = the sum of individual rows. Sounds dumb. People miss it constantly.
  • Assumption cell audit: every yellow cell you created is still yellow, has a value, and has a comment if the value is non-obvious.
  • Stale-tab check: the SUMIF that pulls from the headcount tab. Is it pulling from the current tab, or from a tab you renamed an hour ago and forgot about?

Tonight you find one. The SUMIF for R&D loaded cost is pulling from a tab called Headcount-archive instead of Headcount-current. The total R&D loaded cost in the summary is off by $211K. If you'd sent the model to the CFO without catching it, you'd have spent tomorrow morning rebuilding her trust in your numbers, which takes weeks. Tonight you spent four minutes fixing a cell reference.

The boring 5pm QA pass is the highest-ROI hour of the day. Nobody will ever congratulate you for it. They'll only ever notice if you skip it.

A Short Word on Month-End

Everything above is a normal day. Month-end is not a normal day.

Days 1 through 7 of the month, the day above does not exist. You're on a different clock:

  • Day 2: controller closes the books. You're on standby for re-classes you'll need to explain.
  • Day 3: variance pull, by department, by GL account, against forecast and against prior month.
  • Day 4 by 11am: variance comments due to controller for the management report.
  • Day 5-6: business partner reviews. Every BP wants to know why their line moved. Every answer takes longer than the question deserves.
  • Day 7: board deck draft. You will rewrite the same opex narrative four times. The fourth version will be worse than the second version. You will keep the second version.

Month-end is the week your relationship with sleep gets evaluated. Plan around it. Don't schedule dentist appointments in the first week of the month. Don't promise a one-pager to the CHRO on day 4. The veterans on your team have learned to say no during this week, and you should learn it from them.

What Separates the FA Who Gets Promoted

Here's the thing the JD really, really doesn't tell you.

It's not modeling speed. Almost everyone in FP&A is fast in Excel within twelve months. It's not Adaptive proficiency. The training is decent. It's not how late you stay during board prep — that's a tax, not a skill.

The thing that separates the FA who gets promoted to Senior FA, then to FP&A Manager, is the translator skill. Specifically: turning a variance into one sentence the CRO will actually read. Turning a $211K headcount load into a question the CHRO can answer in a meeting. Turning a 2pm CFO Slack into a Sheets tab that lets her run her 9am prep call.

The model is the table stakes. The translation is the job.

Two FAs at the same company, same tools, same hours. One sends a 14-line variance comment with the underlying journal entries cited. The other sends two sentences and a verb. The CRO reads the second one and forwards it. Six months later, that FA is presenting in the board deck dry-run. Twelve months later, she's running a function.

The translator skill compounds. The modeling skill plateaus.

If you finish reading this and you only take one thing forward, take this: every time you write something for someone above you in the org chart, ask the brutal question — would the CRO read past sentence one? If the answer is no, the model isn't done.

That's the day. That's the part the JD doesn't tell you. Go close out your tabs.

Learn More