Your First 30/60/90 Days as a New FP&A Analyst

The first time I opened the operating model my predecessor left behind, the revenue sheet had eleven hardcoded numbers labeled "Per Mike, Q3 2024." There were three circular reference warnings stacked in the corner of Excel. The headcount tab had comments from three previous owners, two of whom no longer worked at the company, arguing about whether contractor cost should be loaded into OpEx or COGS. One comment thread ended with "we'll fix this next quarter," dated eighteen months prior.

Welcome to FP&A.

Most new FP&A analysts at B2B SaaS companies inherit a model nobody fully understands. The person who built the original tabs left two roles ago. The person who patched it last left six months ago. You're now the third or fourth owner, and the CFO expects you to defend every number in next week's BvA. The instinct is to rebuild the whole thing in week two. Every analyst who did that lived to regret it.

The job for your first 90 days isn't to rewrite the model. It's to rebuild trust in the numbers and ship one visible win — without breaking the close.

Why the 30/60/90 Frame Matters Here

FP&A is one of the few jobs where you can lose credibility in a single bad meeting. If you walk into your first board prep and say "the forecast is wrong but I don't know why yet," you're stuck doing variance commentary for the rest of the year. If you walk in saying "I traced this number to its source, here's the assumption, here's where I'd push back," you get pulled into the strategic conversations.

The 30/60/90 isn't an HR ritual. It's the time you have to earn the right to touch the model, the right to challenge a forecast, and the right to sit at the table when next year's plan gets shaped.

Days 1-30: Audit, Don't Touch

The single best advice I got my first week: don't change any formulas for thirty days. You will be tempted. You will see hardcodes that look obviously wrong. You will see broken named ranges and formatting nobody fixed. Leave them alone.

Here's what to do instead.

Audit every assumption

Open the model. Open a blank Google Doc. Walk through every tab and write down:

Every hardcoded number in a formula cell (cells that should reference but don't)
Every circular reference warning
Every #REF! or #N/A that's been "temporarily" suppressed
Every named range that's been overwritten or shadowed
Every comment from a previous owner (quote them verbatim with the date)

Don't fix anything. Just inventory. By day 14 you should have a 30-50 line audit doc. This becomes your map.

The first hardcode I documented was "ARR Q1 2026: 4,200,000 (per Sarah, board package)." I traced it. There was no Sarah. There was a Sara who left in 2024 and a Sarah in CS who'd never seen the model. The number was wrong by $180K. I didn't touch it for another month, but I knew exactly which formula to fix when I got the green light.

Sit in five business partner meetings — listen, don't pitch

Before day 21, get yourself invited to one weekly meeting for each of these:

Sales (pipeline review or forecast call)
Customer Success (renewal forecast or churn review)
Marketing (pipeline contribution or budget pacing)
Product (roadmap and resource asks)
RevOps (data quality, attribution, lead routing)

Bring a notebook. Don't pitch your model. Don't volunteer ideas. Listen for these specific things:

What number does the team argue about every week?
What number does the team trust without question?
What number does the team quietly distrust but never flag?

Those three questions tell you where the real work is. The numbers people argue about are usually overdriven by noise. The numbers they trust without question are usually wrong but invisible. The numbers they quietly distrust are usually the ones the CFO is asking about right now.

Map every data source

By end of week three, you should be able to draw a flow diagram from raw source to model output:

Salesforce → ARR by stage (which report? owned by whom? refresh cadence?)
NetSuite → revenue actuals (which segments? which currency conversion logic?)
Stripe → MRR roll-forward (does it match NetSuite? if not, by how much, and why?)
HubSpot → marketing-sourced pipeline (deduped against Salesforce? attribution model?)
HRIS → headcount feed (does it include open reqs? contractors? loaded cost or base?)

If two systems give you different numbers for the same metric, find out which one finance treats as truth. There's always a tiebreaker. Often nobody has written it down.

Learn the close calendar cold

Get the close calendar from the controller. Memorize:

Day of soft close
Day of hard close
BvA delivery deadline
Flux commentary owner per cost center
Board package deadline (and who reviews before CFO sees it)
Reforecast cadence (monthly? quarterly? rolling 4Q?)

Miss one of these and you'll burn credibility you can't get back. The close runs on a clock that doesn't care that you're new.

By day 30, your deliverable is a five-page audit doc shared with your manager. Not pretty. Not polished. Just: here's what the model does, here's where the data comes from, here's what's broken and what's just confusing.

Days 31-60: One Fix, One Driver, One Forecast

Now you can touch the model. But only three things. Not four. Not the whole thing.

Kill one hardcode

Pick one hardcoded number from your audit doc, ideally one in a high-visibility cell (top-line revenue, headcount, top-3 OpEx line), and replace it with a documented driver.

A documented driver is:

A formula that references upstream inputs, not a typed value
A named range with a description in column A or in a "drivers" tab
Tied to a source system or assumption owner (e.g., "headcount Q2: HRIS feed, refreshed monthly, owner = People Ops")

When you ship this, write a one-paragraph note to your manager: "I replaced the Q2 ARR hardcode with a driver fed from Salesforce closed-won. Old number was 4.2M, driver-derived number is 4.18M. Variance of 20K traced to two deals reclassified post-close. Documented in the drivers tab."

That paragraph is the entire deliverable. Don't dress it up.

Fix one broken driver

Pick one driver that's already in the model but produces a number nobody trusts. Common candidates:

Net Revenue Retention (NRR): usually broken because it includes or excludes the wrong customer cohorts
CAC payback: usually broken because the marketing spend denominator doesn't match the customer acquisition window
Magic number: usually broken because the new ARR numerator includes expansion when it shouldn't
Gross margin by segment: usually broken because COGS allocation hasn't been refreshed since the last product launch

Fix exactly one. Document the change. Get the affected business partner (Sales for NRR, Marketing for CAC, etc.) to validate the new logic before it goes into the next forecast.

Build one mini-forecast for a business partner

Pick one of the five business partners you sat with in days 1-30. Build them something useful in 5-10 hours of work. Examples:

A pipeline coverage view for the VP of Sales (current pipeline by stage vs. target, with conversion benchmarks pulled from your own historical data)
A hiring plan model for the VP of Engineering (req-to-start lag, ramp curve, fully loaded cost by level and geo)
A marketing spend pacing tool for the CMO (budget by channel, spend-to-date, projected EOQ vs. plan, with one efficiency metric per channel)

Two rules. First: build it for the partner, not for finance. The output should answer their question, not yours. Second: make it small. A 200-row model that answers one question well beats a 2000-row model that answers four questions confusingly.

Your day-60 deliverable is one fixed hardcode, one fixed driver, and one shipped mini-forecast. Three artifacts. That's enough.

Days 61-90: Own the Number

Now you own something. The job in days 61-90 is to take responsibility publicly, and to set up the next 90 days.

Own a forecast accuracy metric

Pick one number you'll be measured on. Common starting points:

Total revenue forecast accuracy (forecast vs. actual, monthly, last 6 months trailing)
New ARR forecast accuracy (Sales-led, useful if you're embedded with Sales finance)
Headcount forecast accuracy (useful if you're embedded with the People team)
One segment's gross margin (useful if the company has divisional reporting)

Compute the trailing 6-month accuracy as a baseline. Set a target for the next two quarters. Tell your manager what the target is and how you'll defend it.

Forecast accuracy is the metric that converts a junior FP&A analyst into a respected one. You can't fake it. You can't politic it. The number either tightens over time or it doesn't. Pick the one you can actually move and put your name on it.

Present the 90-day report

Around day 85, schedule 30 minutes with your manager. Bring a one-page report:

What I found (5-7 bullets covering biggest data quality issues, biggest assumption risks, biggest business partner pain points)

What I fixed (3 bullets: the hardcode, the driver, the mini-forecast)

What's still broken (5-7 bullets, ranked by business impact, not by how easy they are to fix)

What I want to own next (3 priorities for the next quarter, with the case for each)

Keep it to one page. The CFO and your manager don't have time for a deck. The one-page report is also your audit trail. When someone asks six months from now whether you raised the NRR calc issue early, you have a dated document that says yes.

Propose an H2 plan

Your last deliverable is a forward-looking H2 plan. Not a wish list. Three priorities, each with:

The problem (in business terms, e.g., "Sales doesn't trust pipeline coverage" not "the model has a CAC issue")
The work (what you'll do in finance terms)
The success metric (how you and your manager will know it worked)
The timeline (which month, which quarter)

If you can't get to three priorities, get to two. If you can get to four, cut one. Three is the right number for someone who's been in the seat for 90 days.

The Panic Moments

Two will happen. Probably more. Here's the triage.

"The formula broke and I don't know why." It's the morning of the board meeting. The model is throwing #REF!. Your hands are shaking. Stop. Don't fix. Save a copy of the broken file with timestamp. Open the last clean version (you do have version history, right? if not, that's the lesson). Use the last clean version for the board package. After the meeting, diff the two files cell by cell to find the break. Write up what happened, when, and what you'll do to prevent it.

Don't tell the CFO the model broke unless they ask. They don't care. They care that the package was on time and accurate. If the last clean version was off by less than 1%, ship it and reconcile later. If it was off by more than 1%, escalate to your manager, never directly to the CFO without your manager in the loop.

"My number doesn't match the controller's number." Don't argue in Slack. Walk to their desk (or jump on a call). Bring both numbers, both sources, and both formulas. The answer is almost always one of three things: timing (you pulled different snapshots), scope (one of you is including or excluding a segment), or definition (you're computing a slightly different metric). Resolve it before either of you puts the number in front of leadership. Controllers and FP&A analysts who don't reconcile in private end up reconciling in board meetings, which is bad for everyone.

Measuring Success

By day 90, the answer to "did this go well?" comes down to four things:

One hardcode killed and documented
One driver fixed with the affected business partner's signoff
One forecast shipped that someone outside finance actually uses
One business partner who will take your call without scheduling a meeting first

That fourth one is the most important. The technical work compounds. A hardcode killed in month two saves debate hours in month four. But the relationship compounds faster. The business partner who trusts you in month three will bring you the ambiguous problems in month six. Those ambiguous problems are where the strategic FP&A work lives.

If you've shipped all four by day 90, you're past the survival phase. The next 90 days are about scope expansion: owning a second metric, partnering with a second function, killing a second hardcode. The pattern repeats. The model gets cleaner. The forecast gets tighter. Your scope grows.

The analyst who tries to do this in 30 days fails. The analyst who patiently does the audit in days 1-30, ships three artifacts in days 31-60, and takes public ownership in days 61-90 — that's the analyst who gets pulled into the H2 strategy review instead of just writing the variance commentary for it.

Financial Analyst Playbooks