Choosing a fractional CFO or CTO depends on the bottleneck

A fractional CFO and a fractional CTO fix different kinds of uncertainty. The CFO turns revenue, spending, cash, and financing into decisions. The CTO turns product commitments, engineering capacity, architecture, and technical risk into decisions. When founders confuse the two, they hire an impressive adviser who cannot remove the constraint that is hurting the company now.

The choice is not really finance versus technology. It is whether your next irreversible decision depends more on trustworthy numbers or trustworthy delivery. Close the gap that blocks that decision first, give the first executive a narrow mandate, then bring in the second when the shared roadmap exposes work that crosses the boundary.

Start with the decision that keeps slipping

The first fractional executive should own the decision your current team repeatedly postpones, revises, or makes by instinct. Look at the last six weeks of management meetings. If the same question returns because nobody trusts the cash model, pricing logic, gross margin, or board forecast, the finance gap is already costing you. If it returns because nobody can estimate delivery, explain outages, challenge a vendor architecture, or say what the product can safely promise, the technology gap is already costing you.

Founders often diagnose by discomfort. A messy monthly close feels embarrassing, so they assume they need a CFO. A slow release feels technical, so they assume they need a CTO. That shortcut misses the cause. A bookkeeper or controller may repair the close. An engineering manager may repair sprint discipline. Fractional executives earn their fee when they connect specialist work to company choices, challenge the founder, and accept ownership for a companywide outcome.

Use a decision test before discussing candidates. Write down the three commitments you expect to make during the next two quarters. They may include raising capital, changing prices, signing an enterprise contract, replacing a platform, entering a regulated market, or cutting burn. For each commitment, ask what evidence would let you say yes, what could make the decision expensive to reverse, and who can produce that evidence today. The unanswered evidence tells you which gap comes first.

Timing matters because a fractional role has limited hours. Do not hand that person every neglected task in a department. A useful initial mandate has one company outcome, a small set of decision rights, access to the people and systems that contain the facts, and a date when the founder will judge progress. "Fix finance" is not a mandate. "Produce a suitable for the board cash forecast, reconcile it to the close, and define the hiring envelope before the next planning cycle" is.

CFO signals appear in the quality of commitments

Hire the fractional CFO first when financial uncertainty prevents the company from making or keeping commitments. The strongest signal is not that the books are untidy. It is that the founder cannot explain how a change in bookings, collections, hiring, or infrastructure spending changes cash and runway.

Several symptoms usually travel together. Revenue reports disagree because sales, billing, and accounting use different definitions. The board receives a forecast that management does not use. Hiring approval depends on the current bank balance rather than a model of collections and obligations. Gross margin moves, but nobody can separate hosting, support, implementation, discounts, and payment fees. A fundraising plan names an amount without tying it to milestones or the month cash gets tight. These are management finance failures, even when tax filings and bookkeeping are current.

The distinction between accounting and finance is easy to blur. Accounting records what happened under a consistent method and supports reporting and compliance. Finance builds a view of what may happen and recommends what the company can afford. The IRS recordkeeping guidance says business records should clearly show income and expenses and support financial statements and tax returns. That is necessary evidence, but clean records alone do not decide whether another engineer pays back before runway ends.

A competent fractional CFO should be able to make four views agree: the accounting close, a cash forecast, the operating plan, and the metrics used with investors or lenders. "Agree" does not mean every number is identical. Accrual revenue, bookings, and cash collections answer different questions. It means the bridges between them are explicit enough that a founder can explain a variance without opening five spreadsheets.

Expect a concrete first deliverable, not a redesigned chart deck. One useful artifact is a rolling cash forecast covering 13 weeks with columns such as week, opening cash, customer receipts, payroll, taxes, vendors, debt, financing, and closing cash. Each material receipt needs an owner and an expected date. Each material payment needs a source, not a rounded guess. The model should show a base case and a downside case, reconcile opening cash to bank balances, and record forecast versus actual every week. That pattern exposes collection risk much sooner than a yearly budget does.

A CFO should not become a premium bookkeeper. If the close is late because transactions are uncategorized, hire or fix accounting operations and let the CFO define the reporting standard. Pay executive rates for judgment about capital, pricing, margin, hiring, and risk, not for chasing receipts.

CTO signals show up in unreliable delivery

Hire the fractional CTO first when the company cannot turn product promises into safe, predictable delivery. The signal is not merely a long backlog. It is the absence of a credible technical owner who can connect architecture, people, operations, security, and product scope to a business deadline.

Watch what happens when a large customer asks for a feature, security review, migration, or service commitment. If sales gets a different answer from every engineer, nobody can estimate the work across systems, and the founder approves the deal without knowing its operating cost, the technology gap is executive sized. The same is true when releases depend on one person, production incidents have no learning loop, cloud bills rise without ownership, or contractors make architecture choices that no employee can defend.

A senior developer is not automatically a CTO. The developer may make excellent local choices while avoiding hiring design, vendor negotiation, portfolio tradeoffs, security accountability, and communication with the board. A fractional CTO should give the company a technical operating model: who decides, what evidence decisions require, how work enters the system, how releases are observed, and where the company refuses risk.

Ask for an engineering health sheet that can be checked weekly. It should name the current product commitments, an owner for each, a confidence level, blocked decisions, recent production failures, recovery status, material security work, vendor dependencies, and engineering spend against plan. Avoid a giant dashboard. A small sheet that forces an honest conversation beats dozens of metrics nobody uses.

Two public frameworks help set the floor. NIST SP 800-218 says secure development practices have to be integrated into each software development life cycle rather than treated as a separate cleanup exercise. I agree with that direction, but an early company should translate it into named owners and observable routines instead of copying every example into a policy. The AWS Well-Architected Framework similarly treats operational excellence, security, reliability, performance, cost, and sustainability as connected concerns. A CTO has to make tradeoffs among them; assigning cost to finance and reliability to engineers does not remove the coupling.

Do not hire a fractional CTO to bless a rewrite. Rewrites attract founders because they turn years of mixed decisions into one clean proposal. They often postpone customer work, hide uncertain scope, and preserve the same weak decision process in new code. Require the CTO to show which constraint cannot be corrected incrementally, what evidence supports that claim, and how the company will operate during the transition.

Score the bottleneck instead of debating titles

A scored diagnosis makes founder disagreements visible. It is not mathematics pretending to be certainty. It forces the team to attach evidence to the choice and reveals when both gaps are severe. Score each statement from 0 to 2: 0 means the evidence is current and owned, 1 means it exists but is disputed or stale, and 2 means nobody can produce it reliably.

Score the first five statements as follows:

• Weekly cash forecast and explained variances: CFO 2, CTO 0.
• Margin by product or customer type: CFO 2, CTO 0.
• Hiring envelope for two quarters: CFO 2, CTO 1.
• Forecast of major delivery commitments: CFO 0, CTO 2.
• Release and recovery without one hero: CFO 0, CTO 2.

The remaining statements capture decisions that already cross the boundary:

• Technical cost of a large deal: CFO 1, CTO 2.
• Financial and technical risks in board reporting: CFO 2, CTO 2.
• One owner for finance and engineering tradeoffs: CFO 1, CTO 1.

Add each column, then inspect the high scores rather than blindly following the total. A CFO score driven by missing unit economics and a financing deadline deserves urgent attention. A CTO score driven by an undocumented system may be tolerable for a month if delivery is stable and a capable lead owns it. Severity, decision timing, and reversibility matter more than a difference of one point.

Use three tie breakers when the scores are close. First, which unresolved issue can make the company unable to pay, deliver, or meet a signed obligation soonest? Second, which decision becomes most expensive to reverse? Third, can a manager or specialist contain one side while an executive fixes the other? If collections are uncertain and cash gets tight soon, finance goes first even when engineering is noisy. If an enterprise launch can expose customer data or fail a contractual requirement, technology goes first even when the forecast needs work.

A founder sometimes wants to hire both immediately to end the debate. That can work, but only if the company can absorb two new planning systems and the roles have a shared sponsor. Otherwise each executive builds a correct model with incompatible assumptions. The CFO freezes hiring while the CTO plans a migration that requires more people. The CTO commits to reliability work while the CFO assumes all engineers remain on revenue features. The conflict reaches the founder later, with sunk time attached.

Sequence the roles around irreversible risk

Sequencing should reduce the most dangerous uncertainty first and preserve options for the second executive. The first role does not need to finish its entire agenda. It needs to create enough reliable evidence that the company can scope the second mandate without guessing.

When the CFO starts first, the initial sequence usually looks like this: reconcile actual cash and obligations, produce a weekly forecast, define revenue and margin terms, model the operating plan, and set spending envelopes. The CTO can then evaluate hiring, vendors, infrastructure, and product scope against real constraints. Finance should not dictate architecture, but it should make the cost and timing boundaries visible.

When the CTO starts first, the sequence begins with delivery commitments, system and team ownership, production risk, vendor dependencies, and a realistic capacity view. The CFO can then model scenarios that reflect actual technical choices. A forecast that assumes every engineer creates new features will be wrong if reliability debt consumes a quarter. Technology should not set the financing strategy, but it must disclose the capacity and risk assumptions behind it.

The popular rule "hire a CFO when you raise and a CTO when you build" is too shallow. Every software company is always building, and sound finance matters between fundraising events. Trigger the role with a decision, not a calendar event. A financing process may expose the CFO gap, while customer diligence may expose the CTO gap, but the work remains after the event ends.

Interim containment is legitimate. A controller can stabilize the close while a CTO handles an urgent delivery failure. An engineering manager can establish release ownership while a CFO prepares financing scenarios. Containment fails when the substitute receives executive accountability without authority, time, or experience. State which decisions remain with the founder and set an expiration date for the temporary arrangement.

If both sides score high and cash permits only one role, choose the role that protects survival first. Cash insolvency has a hard boundary. So do certain security, safety, and contractual failures. Compare the earliest credible failure on each side, not the loudest stakeholder. Then ask the first executive to specify the evidence and conditions that should trigger the second hire.

The two executives need one operating roadmap

The CFO and CTO should share one roadmap for company commitments, with different ownership inside it. Separate finance and engineering roadmaps invite contradictions because most important initiatives consume cash, engineering capacity, and management attention at the same time.

Use one row per company outcome, not one row per department task. A practical set of columns is: outcome, commercial deadline, decision owner, finance assumption, technical assumption, cash effect, capacity required, leading evidence, risk threshold, next decision date, and status. Keep detailed sprint work and accounting schedules in their own systems. The shared roadmap contains only the facts each executive needs from the other.

Consider an enterprise launch. The CFO owns the collection assumptions, payment terms, margin treatment, and effect on runway. The CTO owns delivery confidence, hosting and support demand, security work, and recovery requirements. Product or the founder owns the customer outcome. The row cannot turn green until the financial and technical assumptions are compatible. A signed contract with negative contribution and an impossible service commitment is not a win.

Decision rights need the same clarity. The CFO can reject a plan whose funding assumption is unsupported. The CTO can reject a commitment whose security or reliability risk exceeds the agreed threshold. Neither should veto the other's domain by preference. When a tradeoff crosses both domains, they prepare options with consequences and the CEO decides. Record the decision and the assumption that would reopen it.

AWS advises establishing a partnership between finance and technology for cloud cost management. That is more useful than telling engineers to "spend less." The CFO can show the economic constraint and the CTO can connect usage to architecture and customer behavior. Together they can distinguish waste from capacity that protects revenue. A blunt percentage cut cannot do that.

Run a short weekly review of changed assumptions and upcoming decisions, plus a deeper monthly review tied to the close and delivery plan. Do not let the meeting become status theater. If an assumption changed, name the affected outcome, the owner, and the decision date. If no decision or threshold changed, the item does not need airtime.

Measure the mandate before extending it

A fractional executive should make the company less dependent on hidden knowledge, including their own. Judge the engagement by the decisions that became faster, more explicit, and more reliable, not by the volume of documents or meetings.

For a CFO mandate, useful evidence includes a forecast reconciled to actual cash, agreed definitions for revenue and margin, an operating plan with visible assumptions, a financing or profitability path, and managers who understand their spending envelopes. For a CTO mandate, look for owned delivery commitments, visible system risks, repeatable release and recovery routines, a capacity model, and technical options expressed in business terms.

Set baseline and target conditions at the start. "Improve predictability" cannot be evaluated. "Every material product commitment has one owner, a confidence assessment, and a documented scope decision before sales signs it" can. Some targets will be qualitative, but they should still leave inspectable evidence.

Beware of borrowed certainty. A polished forecast can hide weak collection assumptions. A clean architecture diagram can hide a system nobody has tested under failure. Ask the executive to show where the model has been wrong, what changed because of the variance, and which assumption remains dangerous. Good fractional leaders expose uncertainty in a form the company can act on.

The exit condition matters from the first month. The company may need a permanent hire, a stronger manager, a controller, an architect, or only a lighter advisory cadence after the acute gap closes. Extend the fractional role when the work remains executive in scope and the fractional rhythm fits the decision load. Do not extend it because the person has become the only one who understands the spreadsheet or deployment process.

Cost belongs in the same comparison as delay

Compare the engagement cost with the cost of the blocked decision, the likely delay, and the exposure it contains. Founders often compare a fractional fee with the salary of a permanent executive while ignoring recruiting time, equity, onboarding, and the fact that the current gap is already producing losses. The reverse mistake also happens: a fractional title looks economical even though the company needs daily leadership.

A fractional CFO fits when the work is periodic and decision heavy: board preparation, planning, cash management, pricing analysis, financing support, or building a finance function. A permanent CFO becomes more sensible when transactions, stakeholders, controls, fundraising, and team management demand continuous executive attention. Bookkeeping and controllership still need enough operating capacity beneath either model.

A fractional CTO fits when senior judgment can be concentrated around architecture, team design, delivery governance, vendor choices, risk, and coaching. It fits poorly when the company has no capable technical operator between sessions, production requires daily command, or a major transformation consumes daily leadership. Do not buy eight hours of advice and pretend you bought forty hours of ownership.

The cost of delay differs by gap. A weak cash model can cause a late fundraise, a missed tax or debt obligation, or hiring that runway cannot carry. Weak technology leadership can cause lost deals, repeated incidents, security exposure, or months invested in the wrong architecture. Estimate exposure as a range and write the assumptions. False precision adds nothing.

For companies whose largest expense and constraint is engineering, oleg.is offers a Team & AI Audit completed in five business days at a fixed $5,000, with at least $50,000 per year in identified savings or the audit is free. That is relevant when the diagnosis points to engineering cost and operating design; it does not replace CFO work on cash, reporting, financing, or controls.

Interview for evidence, not executive vocabulary

A strong candidate should improve the quality of the diagnosis before you hire them. Give each candidate a real decision with sanitized company data and ask how they would frame it, what they would verify, what they would refuse to assume, and what the founder would still need to decide. You are testing judgment and method, not unpaid consulting or the ability to produce a complete plan in an interview.

For a CFO candidate, share a simplified profit and loss statement, bank balance, receivables aging, headcount plan, and one disputed growth assumption. Ask what prevents a responsible cash view. A serious candidate will distinguish missing records from uncertain forecasts, question collection timing, and ask how revenue definitions connect to invoices and cash. Be cautious if the answer jumps straight to fundraising, cost cuts, or a new finance system before establishing which numbers can be trusted.

For a CTO candidate, share the next major customer commitment, a simple system map, current team responsibilities, recent incident notes, and the engineering cost envelope. Ask what prevents a responsible delivery commitment. A serious candidate will examine ownership, dependencies, recovery, security, capacity, and the commercial promise. Be cautious if the answer begins with a preferred cloud, programming language, rewrite, or hiring plan before the candidate understands the failure mode.

References matter, but ask references about behavior under disagreement. Did the executive make assumptions visible? Could managers act between sessions? Did the person change a recommendation when evidence contradicted it? Did the founder know which decisions remained theirs? Generic praise about intelligence and communication tells you little about how the candidate carries part time accountability.

The engagement design can ruin a good hire. Give the executive direct access to the necessary records, managers, and operating systems. Set a regular decision meeting with the CEO. State where the executive can decide, where they recommend, and where they must escalate. If every request passes through the founder, the executive will see a curated company and produce a curated answer.

Watch for conflicts of interest. A candidate who also sells implementation, recruiting, financing introductions, or a preferred vendor may still be useful, but the incentive must be disclosed before recommendations begin. Separate diagnosis from a large subsequent purchase when possible. Ask how the person gets paid if the recommended path requires more of their own services.

Do not select on executive vocabulary. Some candidates can discuss strategy for an hour without naming a decision, owner, source, threshold, or date. Ask for one anonymized example of a forecast variance, delivery miss, financing choice, incident, or architecture tradeoff they handled. They should explain what they believed, which evidence changed the view, what action followed, and what they would now do differently. Confidentiality may limit detail, but it does not prevent a candidate from explaining their reasoning process.

Finally, test whether the person can leave. Ask what capability should exist inside the company when the mandate ends, who will own it, and how they will verify the transfer. The best answer may recommend a controller, engineering manager, full time executive, or no executive hire at all. A fractional leader who designs permanent dependence has solved the wrong problem.

Make the first mandate force the second decision

The best first hire leaves you with an explicit trigger for the other role. A CFO may conclude that the plan depends on delivery capacity and infrastructure economics nobody can verify. A CTO may conclude that the safe product plan depends on a hiring envelope, margin target, or financing date nobody owns. Those findings are progress, not scope failure.

Write the trigger into the engagement. Examples include "bring in a CTO before signing a service commitment that requires a new reliability model" or "bring in a CFO before approved hiring reduces the downside cash case below the board's threshold." The exact threshold belongs to the company. The discipline is to decide it before urgency distorts the discussion.

Do not ask either executive to impersonate the other. A CTO should understand unit economics and cash constraints, and a CFO should understand delivery capacity and technical risk. Shared literacy improves decisions. Professional accountability still matters, especially for tax, accounting, security, and contractual judgments that require specific expertise.

If you still cannot choose, gather the missing evidence for one week rather than interviewing both roles against a vague brief. Reconcile bank cash, list binding obligations, map the next two quarters of product commitments, identify dependencies on one person, and name the decisions with fixed dates. Then rescore the bottleneck. The argument usually ends when the assumptions become visible.

Hire the role that can turn the nearest irreversible decision from an opinion into a controlled commitment. Give that person authority over the evidence, keep the CEO accountable for companywide tradeoffs, and require the shared roadmap to reveal when the second gap becomes more expensive than waiting.