What a Fractional AI Finance Function Owns in Its First 90 Days

It is Tuesday, 6:14 PM. Your CFO opens the close checklist for the third time today. Twelve items still open on day 12. The July financials were due to the board on the 5th. The new Senior FP&A hire started six weeks ago at $185K base, and the close slipped a day longer than last month, not shorter. The CFO drops a line into the exec channel that reads "July numbers by Friday, running the last two accruals now."

The founders in this seat keep hiring another FP&A body and wondering why the close does not get faster. The answer is that FP&A is five queues stitched together, and hiring a manager gives you one person holding all five at 40 percent throughput. A fractional AI finance function is scoped differently, and the first 90 days follow a shape the CFO can price against the JD sitting in her drafts folder.

Day 1 through 30: the close and the cash queues

The first sprint lands two functions on the CFO's desk: month-end close orchestration and daily cash reporting. Both are queue-work, both have a fixed cadence, and both live in five source systems that nobody reconciles end to end.

The close agent reads NetSuite journal entries, Ramp expense feeds, Stripe payout ledgers, the AR aging from Salesforce, and the CFO's manual accrual sheet on a nightly cadence from day 20 of the month. Flags every account with an unusual variance against the trailing three-month average. Drafts the accrual entries for legal, cloud spend, and revenue-share splits with the source-doc links attached. Posts the close-status dashboard to #finance at 7:30 AM every day of the close window with the six open items, the owner, and the blocking dependency.

The 12-day close collapses to a 5-day close by the second month-end, and the CFO stops writing the "numbers by Friday" Slack. The second queue is daily cash. The agent reads the bank feeds across three accounts, the AR aging, the AP queue in Bill.com, and the payroll calendar. Drafts the 13-week cash forecast every Monday at 6 AM against the prior week's actuals, flags any week that drops below the 90-day runway threshold, and routes the exception to the CFO's DM before standup. Two functions live on the finance surface by day 30, and the CFO stops running the close from her inbox and starts reading the summary in a channel that updates on cadence.

Day 31 through 60: the vendor and the revenue queues

The second sprint adds SaaS vendor management and revenue reconciliation. The 214-tool SaaS bill is the queue every CFO points at and nobody works on Tuesday. The agent reads Ramp, Brex, the SSO login logs from Okta or JumpCloud, the Slack integration list, and the last 90 days of invoice line items. Every tool with zero logins in the trailing 30 days lands on the CFO's Monday review with the contract renewal date and the cancellation SLA. Every auto-renewing SaaS inside 60 days of renewal lands on the same list with the option to cancel flagged.

The revenue reconciliation agent reads the Stripe payout ledger, the Salesforce closed-won invoice line, the NetSuite deferred revenue schedule, and the customer usage feed from the product database. Reconciles the four sources against each other every night. Flags every account where invoiced revenue and recognized revenue drift more than the threshold the CFO signed off on. The controller who spent 14 hours a month tracing revenue leakage now spends 90 minutes on the exception rows. Four functions live by day 60, the finance team stops staffing a queue, and the calls that used to eat a controller's Thursday collapse into a fifteen-minute exception review.

Day 61 through 90: the board pack and the audit trail

The third sprint lands the board pack and the audit-ready trail. Both are exec-visible. Both are what the CFO stops sleeping over on the Friday before the board meeting. The board pack agent reads NetSuite, Salesforce, the CFO's KPI sheet, Amplitude for product metrics, and the prior quarter board deck on the last business day of the month.

Drafts the finance section against the CFO's template with the numbers reconciled across the five source systems and the narrative bullets already written from the variance analysis. Lands the draft in the CFO's shared drive by 7 AM the first business day. The CFO edits two slides and adds the strategic framing. The 14-hour board prep week collapses to a 90-minute edit.

The audit trail agent runs in parallel from day 60 and matures by day 90. Every journal entry, every accrual, every reversal, every reconciliation touch gets logged with the source document, the timestamp, the approving user, and the underlying data snapshot. The audit-prep sprint that ate three weeks of controller time last year becomes a Slack search. The Big Four field auditors ask for the accrual support on the July close and the controller sends a link that resolves in 4 seconds. Five functions live by day 90, and the finance function stops looking like a queue nobody staffed and starts looking like a system the CFO reads once a day.

The math against the FP&A JD

Run the two paths over 12 months. Path A hires the Senior FP&A manager at $185K base, 15 percent bonus at $28K, 22 percent taxes and benefits at $47K, equity refresh at $28K, recruiter at 22 percent of first-year cash amortized at $47K, tools and seat at $12K. Loaded year-one cost lands $347K. She owns close orchestration at 60 percent throughput by month 4, holds the cash forecast on the side, and the SaaS ledger, revenue reconciliation, board pack, and audit trail remain queues without owners.

Path B runs three 14-day sprints across the first 90 days at low to mid five figures each, lands five functions live by day 90, and runs at $8K to $14K a month on API spend, monitoring, and the exception loop. Add a fractional finance operator at $6K a month who owns the exception queue, the retro, and the quarterly close review. Total year one: $180K to $220K in build and run against $72K in fractional judgment equals $252K to $292K. Five owned functions against one 40 percent hire. Same P&L line, different close.

The four numbers a CFO runs before the next FP&A req

The founders reading this are ten days from signing a JD for the third FP&A body in eighteen months. Before the signature, run four numbers against the finance function, not the org chart. Score the queue, not the headcount.

Days to close. Measure the median business days from month-end to the board-ready P&L. A healthy function lands under 5 days. A stuck function lands 10 to 14 days, and every board update ships late.

Reconciliation exceptions per month. Count the number of times invoiced revenue and recognized revenue had to be corrected in the trailing 90 days. A healthy function runs under 4. A stuck function runs 12 to 20, and the CFO stops trusting the deferred revenue line on the board pack.

SaaS shelfware ratio. Divide the number of contracted SaaS tools with zero logins in the trailing 30 days by total contracted tools. A healthy function runs under 10 percent. A stuck function runs 30 to 45 percent, and the CFO signs a renewal in Q3 for a tool that stopped being used in Q1.

Board-pack lead time. Measure the median hours the CFO spends building the finance section of the board deck the week before the meeting. A healthy function runs under 3 hours. A stuck function runs 12 to 20, and the CFO's calendar loses two exec-level days a month to slide assembly. Any two numbers in the stuck zone means the queue is the problem, not the headcount, and you can scope the finance function in a 30-minute call this week.