The first 90 days as a CFO at a startup break into three phases: stabilize the numbers in days 1 to 30, build the model in days 31 to 60, and own the narrative in days 61 to 90. One decision comes first, in week one: how the execution work will get done, because every deadline after it depends on that answer.
Most 90-day CFO guides skip that decision entirely. They are written for executives inheriting a finance organization, with a controller, an FP&A team, and an ERP already in place. A startup CFO inherits a founder's spreadsheet, a part-time bookkeeper, and a board that expects investor-grade reporting within a quarter. This guide covers what the standard playbooks leave out, and it sits inside our complete guide to the fractional CFO for startups decision, which covers when to bring in finance leadership and in what form.
Why the standard 90-day CFO playbook fails at a startup
Search for a new CFO 90-day plan and the top results tell you to assess your finance team, audit existing processes, and evaluate the current tech stack. BCG, Workday, and Sage all frame the first 90 days as a listening tour through an organization that already works. That advice assumes there is something to assess.
At a pre-seed through Series A company, there usually is not. The "finance team" is a founder who closes the books at midnight and a bookkeeper who categorizes transactions once a month. The "tech stack" is QuickBooks, a banking portal, and eleven spreadsheets with names like model_v7_FINAL_use_this_one. The "processes" live in one person's head.
That changes the job. An enterprise CFO manages a function; a startup CFO builds one. The enterprise playbook says spend 90 days learning before you change anything. The startup version says you have 90 days to produce a trustworthy close, a working forecast, and a board-ready narrative, starting from raw material. First-time CFOs who apply the enterprise sequence run out of calendar before they produce anything the board can use.
The 30/60/90 day plan at a glance
The plan below compresses the build into three phases. Each phase has one job, a set of concrete outputs, and a definition of done that a founder or board member could verify.
| Phase | Focus | Key outputs | Done when |
|---|---|---|---|
| Days 1-30 | Stabilize the numbers | Clean chart of accounts, reconciled books, first monthly close, cash runway math, execution-layer decision | The founder trusts the P&L more than their spreadsheet |
| Days 31-60 | Build the model | Driver-based financial model, revenue recognition policy, scenario plans, budget vs. actuals cadence | You can answer "what happens if we hire two engineers" in an hour |
| Days 61-90 | Own the narrative | Board deck, investor update rhythm, metrics dictionary, fundraising-readiness file | The board hears the numbers from you, with no surprises |
The phases stack. A model built on unreconciled books is fiction, and a board narrative built on a broken model collapses under one diligence question. Resist the pull to skip ahead, even when the CEO asks for the fundraising deck in week three.
Days 1 to 30: Stabilize the numbers
The first month is forensic. Before you forecast anything, you need to know what is actually true, and at most startups the honest answer is that nobody fully knows. Start with the bank accounts and work outward, because cash is the one number that cannot lie to you.
Your first-month checklist:
- Reconcile every bank and credit card account to the general ledger, going back at least six months.
- Rebuild the chart of accounts so it maps to how the business actually runs: hosting under cost of revenue, contractors split by function, founder Amazon purchases out of "Miscellaneous."
- Run one full monthly close yourself, end to end, and write down every step. That document becomes your close process.
- Calculate real cash runway from reconciled actuals, not the founder's estimate. The two numbers rarely match, and the gap is usually unpleasant.
- Meet every money-touching vendor relationship: payroll, banking, tax, and any outsourced bookkeeping.
Runway deserves special attention because it resets every other priority. If the company has 20 months of cash, you can sequence the build calmly. If it has seven, the model phase becomes a fundraising-prep phase and the whole plan compresses. Run the math early with a startup runway calculator and pressure-test the founder's assumptions about collections and payment timing.
The week-one decision: who does the execution work
Here is the decision the enterprise playbooks defer to year two and a startup CFO must make in week one: who, or what, does the transactional work. Someone has to categorize transactions, chase invoices, process bills, and maintain the close. A first-time CFO who absorbs that work personally becomes a $200,000-per-year bookkeeper, and the strategic work the board hired them for never happens.
There are three options. Hire a staff accountant or controller, which costs $90,000 to $180,000 per year plus a two-to-three month search at exactly the moment you have no time. Outsource to a bookkeeping firm, which works for the ledger but leaves forecasting, AR follow-up, and board prep on your desk. Or run the execution layer on AI agents. Futureproof's six agents handle bookkeeping, receivables, payables, forecasting, revenue operations, and investor reporting for $1,000 per month flat, which means the function exists on day one instead of after a hiring cycle.
Whichever route you choose, decide it in week one. The 60-day and 90-day outputs assume the transactional layer is handled, and a decision deferred to day 45 pushes everything behind it. We wrote about the trap of becoming the connective tissue yourself in stop being the human API for your finance stack.
Days 31 to 60: Build the model
With reconciled actuals underneath you, the second month is about the forward view. A startup financial model is not a valuation exercise; it is an operating tool the CEO uses to make hiring and spending decisions. Build it driver-based: pipeline converts to bookings, bookings convert to revenue on a schedule, headcount drives payroll, usage drives infrastructure cost.
Two accounting policies need to be settled in this phase because they distort everything downstream if left loose. First, write a revenue recognition policy, even a simple one. SaaS companies that book annual contracts as revenue on the day cash lands overstate growth and misstate every ratio a Series A investor checks. Second, fix the treatment of burn rate: gross versus net, cash versus accrual, and which one-time items get excluded. Founders and boards argue about burn constantly, and the argument is almost always a definitions problem.
Then make the model earn its keep with scenario planning. Build a base case, a downside where the next raise slips six months, and an upside where the current growth rate holds. The point is not prediction. The point is that when the CEO asks what two more engineers do to runway, you answer in an hour with a model everyone already trusts.
Close the month by standing up a budget versus actuals cadence. Every month, compare the plan to reality, explain the top five variances in plain language, and adjust. This single ritual, run consistently, does more for your credibility than any dashboard. The tooling to support it does not need to be expensive; we covered lean options in the seed-stage finance stack and in the finance stack for founders who want to stay lean.
Days 61 to 90: Own the narrative
The third phase converts infrastructure into trust. By day 60 the numbers are clean and the model works; by day 90 the company's financial story should come from you, in a form investors and the board can consume without translation.
Start with the board package. Pick the eight to twelve metrics that actually describe the business, define each one in a written metrics dictionary, and commit to reporting the same definitions every quarter. Changing metric definitions mid-year is the fastest way to lose a board's confidence, and it is the most common first-time CFO mistake we see. Pair the deck with a monthly investor update that ships on the same day each month, good news or bad.
Then build the fundraising-readiness file, even if the next raise is a year out. A data room assembled in a weekend panic looks like one; a data room maintained monthly looks like operational discipline. Historical financials, the model, the cap table, key contracts, and the metrics dictionary all live there. When the raise starts, diligence becomes a link you send instead of a fire drill.
Finally, be honest about where your own hours go. If most of your week still goes to transaction-level work at day 90, the execution-layer decision from week one needs revisiting. A full fractional CFO runs $3,000 to $12,000 per month and still needs someone doing the underlying work. The order matters: automate the execution layer first, then spend leadership dollars on judgment.
What done looks like at day 90
A startup CFO who runs this plan ends the quarter with a repeatable close, a driver-based model the CEO actually uses, a board that gets consistent numbers on a consistent schedule, and a data room that is always one month from raise-ready. Just as important, they end it with their own time pointed at strategy rather than data entry.
The startups that get there fastest make the execution decision early and let software carry the transactional load. Futureproof gives a first-time CFO a working finance team from day one: Vic on bookkeeping, Remi on receivables, Theo on payables, Margo on forecasting, Hugo on revenue operations, and Nia on board and investor reporting, all for $1,000 per month flat. Start your first close with Futureproof and spend your 90 days on the work only a CFO can do.
FAQ
What should a CFO do in the first 90 days at a startup?
Stabilize the numbers first: reconcile accounts, rebuild the chart of accounts, run one full close, and verify real cash runway. Then build a driver-based financial model with scenario plans in days 31 to 60. Spend days 61 to 90 on the board deck, investor updates, and a fundraising-readiness file.
How is the first 90 days different for a startup CFO versus an enterprise CFO?
An enterprise CFO inherits a finance organization and spends 90 days assessing it. A startup CFO inherits a founder's spreadsheet and a part-time bookkeeper, so the first 90 days is construction, not assessment. The deliverables are the same in name only; at a startup, the close, the model, and the reporting must be built before they can be improved.
What belongs on a new CFO checklist for week one?
Week one covers bank and credit card reconciliation status, a runway calculation from actuals, a review of the chart of accounts, introductions to payroll and tax vendors, and the execution-layer decision: whether transactional finance work will be done by a hire, an outsourced firm, or AI agents. That last decision gates everything else in the plan.
Should a first-time CFO hire a finance team right away?
Usually not in the first quarter. A controller search takes two to three months and $90,000 to $180,000 per year, and the work cannot wait for it. Most first-time startup CFOs get further by putting the execution layer on software or AI agents immediately, then hiring for judgment-heavy roles once the function is stable and the company's scale justifies it.
When does a startup need a full-time CFO at all?
Most companies before Series B do not need one full time. A founder with a strong execution layer, or a fractional CFO layered on top of it, covers the ground at a fraction of the cost. The trigger points are a complex fundraise, heavy M&A activity, or revenue operations that outgrow part-time leadership.
