You Hired a CFO to Fix Your Books. That's Not What CFOs Do.
Here's a conversation I've had more times than I can count.
A founder calls. Revenue is growing, the board wants a real forecast, and they've just brought on a fractional CFO to "get finance sorted." Great instinct. Then the CFO opens the books and finds revenue booked in the wrong period, expenses scattered across categories that don't mean anything, and a cash balance that doesn't match any bank statement in the building.
Now your expensive strategic hire is doing data entry. And you're paying CFO rates for controller work.
If that story sounds familiar, you're not alone, and you didn't do anything wrong by hiring too early. You just hired the wrong layer first. Most startups do. Nobody hands you an org chart for finance when you raise your round.
The four mistakes I see on repeat
Hiring a CFO to clean up the books. This is the one above, and it's the most common by a wide margin. A CFO's job is to interpret clean numbers and help you make decisions with them. If the numbers underneath aren't clean, you're not buying strategy. You're buying an very well-paid archaeologist.
Expecting a bookkeeper to think like a strategist. A good bookkeeper will hand you a monthly P&L you can trust. That is not nothing. But somewhere along the way, founders start expecting that same person to weigh in on the hiring plan or run point on a fundraise. Clean reporting is an input to strategy. It is not strategy.
Skipping the controller layer entirely. This is the quiet killer. Startups go straight from "someone does the books" to "we have a CFO," and skip the layer that actually makes the numbers trustworthy. Then diligence hits, and the gaps show up all at once: cash receipts booked as revenue, Stripe fees tangled up with deposits, contractor costs filed under the wrong bucket, accruals nobody made, balance sheet accounts nobody reconciled, three people with three different definitions of ARR. A CFO can build a beautiful model on top of that. It just won't be a model you can trust.
Hiring three vendors who've never spoken to each other. One firm does the books. Another handles taxes. A CFO shows up once a month for a board deck. Nobody owns the handoffs between them, which means everyone holds a piece of the truth and nobody holds the whole picture. You find out how expensive that gap is exactly when you can least afford to: mid-diligence, mid-audit, mid-crisis.
So when do you actually need the next layer?
Honestly, there's no revenue number that tells you. I wish there were, it would make this a much shorter article. What actually predicts the need is complexity, not ARR.
You need bookkeeping help when:
Transactions are piling up faster than anyone is entering them
A founder or ops person is still doing the books on the side
Reconciliations are inconsistent from month to month
You can't easily tell what's been paid, billed, or reimbursed
You want financials you can actually rely on every month
You need controller support when:
Close takes forever and the timeline moves every month
You've got recurring revenue, deferred revenue, inventory, or multiple entities
Your statements wouldn't survive a GAAP review
A fundraise, audit, or real diligence is on the horizon
Nobody can explain a variance without a shrug
Your entire finance process lives in one person's head
You need fractional CFO support when:
Runway and burn have become real strategic conversations, not spreadsheet trivia
You're heading into a raise
The board wants reporting that reads like it came from a real finance function
Hiring and expansion decisions need actual scenario planning
You genuinely don't know your unit economics
Leadership keeps debating big calls without a model to test them against
Here's the part that surprises people: you may need more than one of these at once. That's not a sign something's broken. That's just what growth looks like.
The three roles are a chain, not a menu
Think of it less like picking a title off a menu and more like a supply chain for decision-making.
The bookkeeper captures the transactions. The controller verifies them, closes the books, and standardizes the numbers so they mean the same thing every month. The CFO takes that foundation and uses it to guide decisions.
In practice: the bookkeeper logs payroll, vendor bills, and customer payments. The controller reconciles the accounts, books the accruals, and closes the month on schedule. The CFO uses that closed, reliable month to update the runway model, stress-test the hiring plan, and write the board narrative.
The work is connected. So the accountability has to be too. The second you let three unconnected vendors own three links in that chain, you've built a system where nobody's actually responsible for the output.
Where AI actually fits into all of this
AI can strip a lot of manual work out of every layer of that chain. It cannot replace the judgment that makes the chain useful in the first place.
Where it genuinely helps: matching transactions and supporting reconciliations, pulling data out of QuickBooks, Stripe, Brex, Ramp, and Rippling without a human copy-pasting between tabs, drafting journal entries, flagging transactions that look off, running variance analysis, keeping recurring reports current, and producing a first-draft management report that a human then sharpens.
But here's the part every vendor pitching you "AI-powered finance" conveniently leaves out: automation sitting on top of a broken process doesn't make the process faster. It makes the confusion faster. You just get bad numbers with more confidence attached to them.
Our approach at SISU is to map the actual workflow before we automate anything, then decide what should run on its own, what needs an AI agent with a human checking its work, and where a person needs to be in the loop no matter what. AI handles the routine. Humans still drive the judgment. That's the whole idea behind our human-led AI finance operations approach: experienced operators, working alongside purpose-built agents, not replaced by them.
What most scaling startups actually need
You don't necessarily need three separate hires to get this right.
Early on, one strong finance partner can reasonably cover bookkeeping, controller work, and a lighter version of CFO support. As transaction volume and decision complexity climb, those responsibilities need cleaner separation, whether that's three specialized people or one partner with the systems and staffing to run all three layers properly.
That's the model we built SISU around: one finance operations partner covering day-to-day accounting and bookkeeping, month-end close and controller oversight, runway planning and fundraising models and board reporting, systems integration and workflow redesign, and AI-powered automation with an actual human governing it.
We're not trying to sell you a bigger finance team than you need. We're trying to match the right layer to the actual problem in front of you, which is a very different goal than most finance vendors are optimizing for.
To be straight with you: most clients see measurable improvement within 30 days, but that depends entirely on your systems, your transaction volume, and how messy things were when we started. If someone promises you a flawless close or instant automation without ever looking at your actual workflows first, that's a pitch, not a plan.
The bottom line
A bookkeeper records the past. A controller makes the past reliable. A fractional CFO uses that foundation to plan the future.
If your books are a mess, fix the data layer first. If your close is unpredictable, shore up your accounting operations. If the numbers are clean and it's the next decision that's fuzzy, that's when fractional CFO services actually earn their cost.
None of this is about collecting the right titles on an org chart. It's about finding and removing whatever's actually standing between you and your next decision.
Not sure which layer you're missing? Book a 20-minute conversation with SISU. No pitch deck. Just a straight answer about what's broken, what can wait, and what to fix first.
