By: Natalie Johnson
A founder can lose a valuation point for every question they fumble in due diligence, and most of those questions are about numbers the company already has, just not in a form anyone outside the building can trust. That is the gap Mark P. Beltran, a fractional chief financial officer (CFO) and founder of Silicon Valley Consulting who has spent over 25 years in the venture capital and private equity ecosystem and has helped founders raise capital, spends his time closing. Most founders assume investor-grade finance only arrives with a full-time CFO.
It does not. It arrives when a company’s books, model, and story say the same thing. For early-stage AI companies, whose inference bills scale differently from traditional software costs, that alignment is one of the clearest signals of whether a round closes in six weeks or drags on for four months.
What Investor-Grade Finance Means for an AI Startup
Beltran’s definition is unsentimental. “Investor-grade finance means your numbers can survive a third party’s diligence without you in the room,” he says. The mechanics are specific: accrual books closed on time every month, revenue recognized the way an auditor would recognize it, and inference and hosting costs sitting in cost of revenue, where investors expect to find them. Above that sits a driver-based model tying hiring, compute, and pricing to cash, with one annual recurring revenue (ARR) definition, one gross margin, and one burn number, not several versions depending on who is asking.
The failure modes repeat often enough to work as a checklist: inference costs buried in research and development (R&D), an ARR figure padded with pilots that may never convert, and a financial model nobody can trace back to the general ledger. Each one looks harmless inside the company, which is why it survives until diligence, where a third party unwinds it. “Investors are pricing risk, and messy numbers are risk they can see,” Beltran says. Closing those gaps does not require a full-time CFO. It requires someone who has sat on the investor’s side of the table and knows which questions are coming.
The Margin Curve That Reprices the Company
Ask Beltran what he checks first, and the list is short enough to run in ten minutes: Does gross margin include the true cost of serving customers? How much revenue would survive if every pilot ended tomorrow? How much cash does it take to add a dollar of ARR? Each question tests whether a headline number can bear weight. Stage matters too. At seed, investors want proof of demand: users a company can acquire who activate and get real value. By Series A, attention shifts to retention, gross margin, and burn efficiency, the measures that decide what growth is worth.
Gross margin is where AI founders most often trip, and it is a pricing problem disguised as a bookkeeping one. Teams park inference and hosting costs in R&D or operating expenses, then set pricing and hiring off a margin that was never real, without modeling how cost of revenue moves as they scale. “When a diligence team rebuilds that curve, a margin that looked like software can start to look like services, and the valuation conversation changes with it,” Beltran says. His remedy: build the curve proactively and honestly. A margin that improves with scale beats a flattering number today, because investors underwrite trajectory. It is also why he built the Fundable Finance Score, a free three-minute assessment that shows founders how their finances look through an investor’s eyes and which gaps to close first.
Credibility Is Built Before the Raise
“You can’t build credibility during a raise; you can only spend what you built before it,” Beltran says. His prescription is a full quarter before the first pitch: a month to clean the books, a month to rebuild the model on top of them, and a month to pressure-test the story against the questions investors will ask. Skip it, and the same work happens live, in diligence, on the investor’s clock. That is what turns a six-week round into a four-month one, and the cost compounds. Investors price risk into valuation, so every question a founder cannot answer cleanly reads as risk, and risk shows up as a lower price or tougher terms. Prepared companies set the pace of their raise. Unprepared ones have it set for them.
What AI Can Close and What It Cannot Decide
AI now handles a large share of the volume work in accounting: categorizing transactions, reconciling accounts, drafting the close, flagging anomalies, and producing a first-pass report. It is also getting good at finding problems. It can flag that a company’s fastest-growing customer is losing money on every API call. What it cannot do is decide what to do about it: whether to keep that customer because they are the logo that closes your Series A, how to reprice them without losing them, and how to walk into that conversation and keep the relationship. “AI made the answer cheap. It didn’t make the question any easier,” Beltran says.
The judgment sits in four places: knowing which question to ask, weighing trade-offs the data cannot settle, timing calls like when to raise or cut, and leading the hard conversations with customers, boards, and investors. Beltran reaches for Captain Sullenberger, who had 208 seconds between the bird strike and the Hudson and no system on the aircraft telling him to land on the river. That call came from decades of situations already seen, not from a better instrument panel. For an early-stage company, the highest-ROI model is AI and a lean accounting team handling the volume, with senior judgment applied where the important calls get made. Knowing which calls those are is a judgment call in itself.
When to Bring In a Fractional CFO Instead of a Full-Time Hire
The right time, Beltran says, is usually earlier than founders think. A full-time CFO is among the most expensive hires an early-stage company makes once salary, equity, and the cost of getting it wrong are counted. And most Seed and Series A companies do not yet have a full-time CFO’s worth of work. What they have is a handful of decisions each quarter that swing the outcome, and a fractional CFO lets them buy the judgment for those decisions without buying the whole seat.
The signs of readiness are concrete: a raise within 12 months, a pricing or revenue model that has outgrown a spreadsheet, or a board asking questions that are being answered with estimates. “If two are true, you’re already paying for the lack of senior financial judgment, just not on an invoice,” Beltran says. Even before those signs appear, he argues, a company’s first financial model is reason enough to bring in a fractional CFO, because building it well teaches the leadership team the drivers of its own business.
He likens the role to the Bay pilot, who boards a large ship before it passes under the Golden Gate. The captain still runs the ship, but the pilot’s known unknowns are the captain’s unknown unknowns. The risk he flags is what a capable accountant with a CFO title misses while learning the strategic part of the job on the founder’s dollar, timeline, and reputation. Beltran has sat on all three sides of the table, as CFO, founder, and investor, and in his experience the companies that benefit most bring that judgment in a round before they think they need it. “When every startup has the same tools,” he says, “judgment becomes the edge.”
Follow Mark P. Beltran on LinkedIn for more insights on fractional CFO leadership, startup unit economics, and fundraising readiness. Founders can take the free Fundable Finance Score at siliconvalley.consulting/score.
Disclaimer: This article is for informational and educational purposes only and does not constitute financial, investment, accounting, or legal advice. The strategies and opinions discussed are general in nature and may not apply to every business. Readers should consult qualified financial, accounting, or legal professionals before making business or investment decisions.
