When Should a Startup Hire a Fractional CFO

All right, here is the direct answer. Hire a fractional CFO when a decision arrives that your books cannot answer. In practice that means six triggers: a raise six to twelve months out, runway under nine months, a pricing model that has outgrown the spreadsheet, enterprise buyers, a second entity, and unit economics nobody can defend.

Notice that not one of those is a revenue number. That is deliberate. Almost every version of this question arrives shaped as a threshold. At what ARR do I need a CFO. At what headcount. A threshold is easy to plan around, and a founder already carrying nine jobs would like to know which month the tenth one starts. But the framing is wrong in a way that costs money.

The real trigger is not a size, it is a decision. Somewhere in the life of a company the decisions start outrunning the books. You are still closing the month, the balances still tie, the reports go out. What changed is that the questions in front of you are no longer about what happened. They are about what happens next, and your accounting system was not built for those.

The threshold question is the wrong question

One thing to clear first. I have written separately about whether a fractional CFO is the right kind of help at all, and how that job differs from a bookkeeper, a controller and a CPA. If you are not sure which of the four you are missing, start with that piece, because hiring the wrong one is the commoner and costlier mistake. This piece assumes you cleared that gate and what you want to know is timing.

Timing matters because the cost of being wrong is asymmetric. Being early costs you a retainer. Being late costs you a term sheet, a quarter of runway, or a number you cannot defend in the one room where defending it was the whole job. That is why I would rather hand you six situations to watch for than a figure to cross.

The six triggers, in one place

Here they are as a single list, written so you can lift it into a board deck without me. If two or more are true today, the timing question has answered itself. If one is true, you are in the window where it is worth a conversation. If none are, you need a good controller and a quarterly check in, and I would tell you that rather than sell you something.

1. A raise six to twelve months out. Not a raise you are in. A raise you can see. A data room's financial section is built over quarters, and the modelling behind a valuation cannot be reverse engineered from a cash basis ledger in the weeks after a term sheet lands.
2. Runway under nine months. Below nine months every operating decision becomes a runway decision, and what answers that is a weekly cash model, not a monthly close. Nine is roughly where a fundraise can no longer be started calmly.
3. A pricing model that has outgrown the spreadsheet. Usage tiers, hybrid contracts, overages, credits, custom terms per deal. Once price varies by customer in ways your billing system does not model, your revenue number is an estimate and nobody knows by how much.
4. Enterprise buyers. Moving upmarket changes the shape of your cash, not just the size of your deals. Longer cycles, procurement, security review and annual billing all land on the forecast, and they land before the revenue does.
5. More than one entity. A second legal entity, a foreign subsidiary, a holding company, or enough states to trip economic nexus. Each adds a consolidation, an allocation and intercompany decisions that must be defensible after the fact.
6. Unit economics nobody can defend. Not unknown. Undefendable. You have a customer acquisition cost, you just cannot say what is in it, and the deck version and the ledger version are different numbers.

The rest of this is the reasoning under each, because a list like that only helps if you can tell which items you are failing. Founders are consistently wrong about which of the six is biting them, and a wrong self diagnosis is how you buy the wrong help.

The raise you can see, and the runway you cannot ignore

Founders get the fundraising trigger wrong in the same direction every time. They wait for the process to start, and by then the useful window has closed. Look at what it consumes. In the Gompers, Gornall, Kaplan and Strebulaev survey of 885 institutional venture capitalists at 681 firms, "The average deal takes 83 days to close; the average firm spends 118 hours on due diligence over that period."¹ That is three months of someone reading your numbers closely, and 118 hours finds whatever is not tied out.

The founder side is worse than most expect. DocSend's analysis of 170 seed pitch decks found "the average number of investors contacted is 66 in 2023, up from 48 in 2022," while meetings set fell from 56 to 38, and half of successful raises took thirteen to twenty four weeks.² The Angel Investment Network's 2025 survey of 610 US founders prices the distraction: "Nearly half (46%) of respondents are spending 30% or more of their week actively engaged in fundraising."³ Cashflow was the biggest risk those founders named, cited by 84% of them.

The gap between rounds has stretched too, so the raise you are planning is further off than the one your investors remember. Carta found that "the median wait time between a seed round and a Series A is 84% longer today than it was three years ago," moving from 420 days in late 2021 to 774 days three years later.⁴ Bridge rounds absorbed the difference: "During the second quarter of 2025, about 16.6% of all cash raised by startups on Carta came via bridge rounds," up from 11.8% a year earlier and 22.5% at Series A.⁵

Which brings me to runway, because the two triggers are one trigger seen from opposite ends. Kruze Consulting, working across a large book of venture backed companies, sets the planning point plainly: "When you've got 12 months left in your runway, you need to start planning your next fundraise."⁶ Their published distribution has consistently shown a third to two fifths of startups under six months.⁷ Twelve months to start, against a process that eats three months of calendar and a third of your week. Nine is where those facts collide.

Under nine months the monthly close stops being the right instrument. What answers a runway question is a rolling weekly view of cash, and I have written the mechanics of the thirteen week version and of runway itself elsewhere. The point here is sequencing. You cannot build that model and earn trust in its output in the same quarter you are pitching.

None of which argues that finance saves a company. CB Insights analysed 431 venture backed companies that shut down since 2023 and found that "Ran out of capital" tops the list at 70%, then said the honest thing: that this is "almost always the final cause of death, not the root problem."⁸ Poor product market fit accounted for 43% and unsustainable unit economics 19%. A CFO does not fix product market fit. A CFO makes sure you find out early enough to act.

When your own price list becomes an estimate

The pricing trigger is the least discussed of the six and the fastest to do damage. It fires when your pricing stops being a list and becomes a set of arrangements. Usage based pricing is now ordinary rather than exotic: a July 2022 survey of 490 SaaS professionals by Maxio with The SaaS CFO and RevOps Squared found 52% of SaaS companies already using a usage based strategy, at nearly identical rates across sales led, product led and hybrid motions.⁹ The same report is candid about the cost, with Maxio's own product lead granting that "Usage billing is challenging to do accurately."¹⁰

The accounting is where it bites. Under the revenue standard you identify the performance obligations in a contract, then deal with consideration that moves, and the standard lists why it moves: "An amount of consideration can vary because of discounts, rebates, refunds, credits, price concessions, incentives, performance bonuses, penalties, or other similar items."¹¹ Every one of those is something a founder gives away in a negotiation without thinking of it as an accounting event, and each has to be estimated, constrained, then trued up.

Bessemer Venture Partners, writing on AI pricing from work with dozens of teams, names the failure mode almost exactly as I see it: "What works at the seed stage can become a liability at Series B."¹² Their fuller description is of pricing complexity proliferating across contracts, with custom deals running unchecked. That is a forecasting problem wearing a pricing costume, and no billing tool cleans it up on its own. Snowflake, whose whole model is consumption based, tells its own investors that its ability to forecast revenue and remaining performance obligations is limited.¹³ If a company that size says so in a 10-K, the expectation for a Series A running three pricing experiments is not better.

The enterprise deal that arrives before the cash does

Moving upmarket is the trigger founders are proudest of and least prepared for. The good news and the problem are one fact: the deals are bigger and they behave differently. Okta puts it in its own risk factors better than I could: "We are increasingly focused on sales to larger organizations, which often involve lengthy purchasing approval processes and less predictable sales cycles."¹⁴ The sentence before it matters more, because Okta plans expenses on assumptions about sales cycle length. Move upmarket and you invalidate the assumptions your spending plan rests on.

Then there is the diligence your buyer runs on you, a real operating cost nobody budgets. Whistic's survey of 525 risk and security leaders found "the typical vendor spends 179 hours every month on completing assessments," at 4.8 hours per request.¹⁵ Most buyers there use a SOC 2 report in that process. Worth being precise, since it sits next to my own boundary: a SOC 2 is an attestation examination performed by a licensed CPA firm under the AICPA's standards, and the Journal of Accountancy notes that "Because most of the new tool providers are not CPA firms, they cannot attest that the controls in place are effective and appropriate."¹⁶ That work is not mine and never will be. Knowing when you need it is.

The cash effect is what I would flag first on a call. Enterprise contracts change when money arrives, not only how much. SaaS Capital's fourteenth annual survey, with more than 1,000 private B2B SaaS respondents, found billing frequency had little effect on growth rates for a fourth straight year, then added the caveat carrying the whole point: "annual billing companies enjoy a meaningful cash flow advantage over monthly billers."¹⁷ A large logo on annual upfront terms is a different company from the same logo billed monthly, and the deferred revenue it creates is not profit even though it is very much cash.

Two entities, and a number nobody can defend

The multi entity trigger sounds like a tax question and is really a control question. The second entity is where informal record keeping stops being survivable, because now there are intercompany balances that must eliminate, expenses allocated on a basis you can explain, and a consolidation that has to exist before anyone sees the group whole. US rules require every legal entity to be evaluated for consolidation, and the standard setters rewrote that analysis because companies were reaching wrong conclusions about which entities belonged inside their statements.¹⁸

Geography compounds it. The Supreme Court removed the physical presence requirement for state sales tax collection, concluding in Wayfair that "the physical presence rule of Quill is unsound and incorrect," in a case where South Dakota's threshold was $100,000 of goods or services or 200 separate transactions.¹⁹ The Government Accountability Office, surveying revenue agencies in every state with a statewide sales tax, reported that "As of June 2021, all 45 states with a statewide sales tax and the District of Columbia had adopted requirements governing sales tax collection" on an economic rather than physical basis.²⁰ Thresholds vary from $100,000 in most states to $500,000 in California and Texas.²¹

Cross a national border and you add transfer pricing, where US regulations require related party transactions to reach a result consistent with what unrelated parties would have reached, in every case,²² plus an information return for each foreign corporation with real penalties attached.²³ Research spending has its own version: expenditures in years beginning after 2021 had to be capitalised over five years domestic and fifteen foreign, and the 2025 law restored an immediate deduction for domestic research.²⁴ The filing itself is handled by a tax partner we trust under a separate engagement, not by me.

Which leaves unit economics, the trigger founders are most confident they have covered. Here is the finding that changed how I ask about it. Benchmarkit's 2025 survey of 583 B2B SaaS companies reported that "Less than 50% of companies measuring CAC Ratio are calculating the Expansion CAC Ratio," so fewer than half of those already tracking acquisition cost know what it costs to grow an existing account.²⁵ Their 2026 edition, across 342 companies, put median CAC payback at 16 months, first quartile 10 and fourth quartile 24.²⁶ ICONIQ's study of 127 software companies found top quartile payback near 16 months, lengthening as companies land larger deals.²⁷

Read those together and the picture is uncomfortable. A spread of 10 to 24 months on the single number governing whether growth spending is an investment or a leak, in a population that mostly does not measure the expansion half of it, while the metric worsens in the exact direction your enterprise motion is pushing you. David Sacks defined the cleanest diagnostic as "Burn Multiple = Net Burn / Net New ARR," and argued that needing three times burn or more to buy growth signals that product market fit is not what it appears.²⁸

Why fractional, and the test I would actually run

Suppose two or three triggers are firing. The question becomes what kind of senior finance help, and the honest comparison is the full time hire, because that is the alternative already in the back of your mind. Robert Half's 2026 guide puts national starting salary for a chief financial officer at $195,500 low, $269,750 mid and $321,750 for extensive experience.²⁹ That is base pay before bonus, equity, payroll taxes, benefits or recruiting, which is how a full time CFO commonly reaches $250,000 to $450,000 or more a year all in.

Federal wage data puts the median for financial managers at $161,700 in May 2024, and reports that "The lowest 10 percent earned less than $86,490, and the highest 10 percent earned more than $239,200."³⁰ Someone who has carried a company through a raise and a diligence process sits at the top of that distribution, not the middle. Against that, senior judgement on a monthly retainer priced to the scope of the work is the right structure for a company whose finance needs are real but not yet full time, and I have laid out the full comparison separately.

You also get timing a search cannot give you, because a CFO search takes months the triggers will not wait for. Carta's read on the market was that "Venture is back, but it is not back for everyone," with more than 60% of capital going to AI companies and a Series A valuation gap that is not subtle.³¹ If you are in the everything else column, the standard of proof on your numbers is higher than it was two years ago.

So forget the six triggers for a second and answer one question. What is the most important decision you will make in the next ninety days, and can your current financial reporting answer it. Not inform it. Answer it. If the decision is whether to hire two more engineers, and your books can tell you what you spent last month but not what that hire does to your runway or your burn multiple, you have found your gap, and your ARR does not matter. The six triggers are simply the six situations where the answer is reliably no. They let you see the gap a quarter before it turns urgent, and urgent is the only time senior finance help is expensive.

One last thing I want to be upfront about. Bringing in a fractional CFO because two triggers are firing does not mean you have lost control of your company, and it does not mean you should have done it sooner. It means the business got complicated enough to need a different instrument, which is the thing you were trying to make happen. Ultimately you have to do what is best for the business. But if you read that list and found yourself nodding at three of them, the timing question is settled, and what is left is a conversation.