CFO Interview Questions: 36 to Ask in the AI Era

What a CFO Interview Is Really Testing

Not whether they can do the job. By the time somebody reaches a CFO shortlist, the technical work is assumed: they can close the books, build the model, read a covenant and survive an audit. The interview exists to establish something narrower and much harder to fake, which is how they behave when the numbers are ugly.

Three things sit underneath every question worth asking. The first is judgement with incomplete information, because a CFO is paid to commit to a number before the data is finished. The second is what they will refuse, since an executive who has never said no to a colleague with a budget request has not been tested. The third is ownership, meaning whether they treat the forecast as something they produce or something they report on.

There is also a fact about your own search that most interview guides ignore. Almost every CFO hire is a replacement, and the vacancy has a shape. Somebody left, or was asked to leave, and the reason is usually one of four: the numbers were late, the numbers were wrong, the relationship with the chief executive broke down, or the company outgrew the person. Decide which one you are living with before you write the question list, then weight the sections below accordingly. A company recovering from late reporting should spend its time on the close and controls. A company that outgrew its last finance lead should spend it on capital and the board.

The AI Questions Most CFO Interviews Still Skip

This is the section that has changed, and the one almost nobody asks. Gartner forecast in September 2024 that by 2026 ninety percent of finance functions would deploy at least one AI-enabled solution, while fewer than ten percent would see any reduction in headcount. Both halves matter. The tools have arrived nearly everywhere, and the savings mostly have not, which means the interesting question is no longer whether a candidate uses them but what they have done with the capacity.

The gap between adoption and readiness is wide. EY surveyed 1,610 chief financial officers and heads of finance between February and March 2026 and published the results that June: only twenty one percent described their function’s preparedness for AI as leading or advanced, and sixty one percent named data quality and bias as the top barrier to investment. A separate benchmarking study of 100 chief financial officers at companies between $50 million and $500 million in revenue, published in February 2026 by a firm that sells finance automation and worth reading with that in mind, found seventy nine percent saying at least a quarter of the finance workload already runs through AI agents, and eighty six percent saying they had hit inaccurate or fabricated output.

So the ground truth is this: it is already inside the close, it is already wrong sometimes, and most finance functions are not ready. Eight questions.

1. Which parts of your close are automated today, and what would you automate next?

Tests whether they have looked at their own process recently.

Look for: Specific steps rather than product names: reconciliations, accruals, variance commentary, and a reason why the next one is next. The best answers include something they chose not to automate.

Red flag: A list of product categories instead of a list of steps.

2. An agent drafts a journal entry that ends up in the quarterly numbers. What is your control?

The single best question in this section, because it forces the candidate to treat the tooling as part of the financial reporting process rather than as productivity software.

Look for: Review thresholds, who signs, what is logged, and how the entry is traceable back to its inputs.

Red flag: They say a human reviews everything, and cannot say what the reviewer is actually checking.

3. How would you know the forecast model was wrong before the board did?

Tests instrumentation and humility at once. Which numbers should actually decide is the same argument in a different setting.

Look for: Named leading indicators, a habit of checking the model against something outside it, and a real example of catching a break early. The best answers describe a number they watch precisely because it disagrees with the model.

Red flag: Confidence in the model itself, with no mention of how it is checked.

4. Where have you refused to deploy AI in finance, and why?

The refusal question. Anybody can list wins.

Look for: A line drawn somewhere and a reason for it: revenue recognition judgement, anything touching a disclosure, anything a regulator would ask to see the reasoning for.

Red flag: No line anywhere, or a line drawn purely because leadership said so.

5. What did your team stop doing after the tooling landed, and what did they start doing instead?

This is where the headcount finding bites. If nothing stopped, nothing was saved and the tools are sitting on top of the old process.

Look for: Specifics about the work that disappeared, and about what the same people do now, usually more analysis, more business partnering, more scenario work.

Red flag: “It made us faster”, with nothing behind it.

6. Somebody in another function brings you a large AI capital request. How do you evaluate it?

The CFO is now the person who decides which of these get funded, and most of the pilots in the market have not paid for themselves.

Look for: A structure: what it replaces, what it costs to run rather than to buy, what evidence would close the case, and what the exit looks like if it fails.

Red flag: They evaluate it on the vendor’s business case.

7. What is the state of your data, and what would you fix in the first ninety days?

Two thirds of finance leaders name data quality as the thing holding investment back, so a candidate who talks about AI without talking about data has not done it.

Look for: Definitions, ownership, and the unglamorous work of reconciling two systems that disagree.

Red flag: The answer is a platform.

8. When a model output turns out to be wrong, who is accountable?

Watch how they talk about their team here. Accountability that lands on an analyst is a culture problem you will inherit.

Look for: The candidate naming themselves without hesitation, then how the error gets caught and what changes afterwards.

Red flag: Any version of the tool being at fault.

Financial Strategy and Capital

These are the questions that reveal whether the candidate can think about the business rather than about finance. Ask them early, while the candidate is still reasoning out loud rather than delivering.

9. Walk me through how you would take a company like ours from where it is now to the number we want in three years.

Make them use your real figures. The point is not the plan.

Look for: Which constraint they reach for first: pricing, mix, cost of acquisition, capacity, working capital.

Red flag: Growth described entirely as more of the same, faster.

10. If you had to take ten percent out of the cost base without slowing growth, what would you stop, pause or redesign?

Forces a trade-off with no comfortable answer.

Look for: The three categories separated deliberately, and a clear statement of what they would protect at any cost.

Red flag: An across-the-board cut, which is what people propose when they do not know where the value is.

11. How would you think about our capital structure over the next two years?

Tests whether they can hold growth and risk in the same sentence.

Look for: A view on the cost of capital, on what covenants would constrain, and on what they would want in place before they needed it.

Red flag: A preference stated as a principle, with no reference to your situation.

12. What level of cash reserve is right for a company at this stage, and how did you get there?

The working is the answer, not the number.

Look for: Reasoning from burn, from the volatility of collections, and from how long it would take to raise if they had to.

Red flag: A rule of thumb delivered as a fact.

13. Tell me about a financial decision you got wrong. What did it cost and what changed afterwards?

Everybody has one. The candidates worth hiring answer in about four seconds.

Look for: A real number, a named consequence, and a change to how they work that survived the incident.

Red flag: A failure that was somebody else’s, or one that turned out to be a strength.

14. Given what you know about us, where would you add value that we have not asked about?

The closest thing to a free look at their judgement.

Look for: Evidence they have read what is public, noticed something, and are willing to say it to your face in an interview.

Red flag: Flattery, or a generic answer that would fit any company of your size.

Forecasting and the Close

Ask about accuracy and you will get a rehearsed answer. Ask about a specific miss and you will get the truth, because the miss is the part nobody prepares.

15. Over the last three years, where did you land against forecast on revenue and on expense?

Separating the two is the whole question. Plenty of finance leaders hold costs beautifully and have no grip on the top line.

Look for: Both numbers, the direction of the error admitted, and an explanation of what drove it.

Red flag: Only one of the two numbers arrives.

16. Tell me about re-baselining a forecast mid-year after something external hit.

Tests composure and sequencing.

Look for: How quickly they moved, who they told in what order, and what they changed about the planning process afterwards.

Red flag: The re-baseline happened at quarter end because that was the calendar.

17. When finance and revenue leadership disagree about the number, who owns it?

The question no published list asks, and the one that causes the most damage when nobody has answered it. The honest answer is that finance owns the number that goes to the board and the revenue side owns the number the team runs at, and the two have to be reconciled deliberately rather than argued about at month end. It is worth understanding what a chief revenue officer is accountable for before this interview, and where revenue operations sits, because the boundary is where most of these arguments actually live.

Look for: The mechanism, described without prompting: a single set of definitions, one source for the pipeline, an agreed conversion assumption.

Red flag: The answer is that finance owns it, full stop.

18. How long does your close take, and what is the constraint?

Anyone can quote days. The constraint is the real answer.

Look for: The constraint named precisely, a system, a person, a dependency on another function, an approval, and a plan for it.

Red flag: The number of days presented as an achievement with nothing behind it.

19. Which number told you something was wrong before the profit and loss did?

Every good finance leader has one or two of these, and they are always specific to the business.

Look for: Something concrete: a collections ageing, a cohort, a renewal rate, a single cost line that moves early.

Red flag: They do not have one.

Controls, Audit and Risk

This is the section that protects you from the expensive kind of surprise. It matters more than usual if the last CFO left because something was wrong rather than because something was late.

20. Have you taken a team through an audit, and what came back?

The finding is the question. Everybody has been through an audit, and not everybody will tell you what was raised.

Look for: The finding named, the remediation, and the date it closed.

Red flag: A clean history with no detail.

21. Tell me about a discrepancy you found. What control exists now that did not exist before?

Tests whether they treat an incident as an event or as information.

Look for: A permanent change to how the work is done, rather than a one-off correction.

Red flag: The fix was more diligence from the team.

22. What controls would you put in place here in the first ninety days?

Say as little as possible before they answer, because the quality lies in what they ask you first.

Look for: Questions back before any commitment: your approval limits, who can move money, how many systems hold the master data.

Red flag: A standard list delivered without a single question.

23. Describe a risk you accepted deliberately.

A CFO who has never accepted a risk on purpose is a brake, not an executive.

Look for: The reasoning, the ceiling they put on it, and who else knew.

Red flag: They cannot think of one.

24. What would you refuse to sign?

Short question, long silence, and the silence is informative.

Look for: An immediate and specific answer.

Red flag: A general commitment to integrity with no example attached.

The Board, Investors and Hard Conversations

Most of the CFO job that goes wrong in public goes wrong here, in the space between knowing something and saying it.

25. Walk me through a time you delivered bad news to the board.

The best single question in any CFO interview. It tests whether they manage up without making the chief executive look blindsided, which is the balance the role lives on.

Look for: When they told the CEO relative to when they told the board, what they brought with the news, and whether a recommendation arrived alongside the problem.

Red flag: The board heard it first, or the news arrived without options.

26. How do you brief a chief executive who does not want to hear it?

You are hiring the person who has to tell you things.

Look for: A pattern rather than a confrontation: early signals, written down where it matters, and a willingness to be on the record.

Red flag: They have never had to.

27. Describe explaining a complex financial position to people who do not read financial statements.

Ask for the actual explanation, not a description of it. The ability to compress without distorting is most of what a board wants from a CFO.

Look for: Plain language, the right level of detail, and no accuracy lost in the compression.

Red flag: The explanation gets more technical under pressure, not less.

28. What goes into your board pack that nobody asked for?

Reveals what they think their job is.

Look for: Leading indicators, a risk that is not yet a problem, or a decision they want made before it becomes urgent.

Red flag: The pack is exactly what was requested, every time.

The Team and the Shape of the Function

You are not only hiring a CFO. You are hiring whatever the finance function becomes over the next three years, and the two are not the same decision.

29. What does the right finance team look like here, and what are the lanes?

Tests whether they design a function or inherit one.

Look for: Roles described by the work rather than by title, and clarity about what they would keep outside.

Red flag: A headcount number arrives before any description of the work.

30. How do you measure the impact of a finance team?

Harder than it sounds, and the answer separates the strategic from the administrative.

Look for: Measures pointed at decisions made rather than at reports issued.

Red flag: Close speed and error rate, and nothing else.

31. Tell me about someone you hired who did not work out.

Every hiring executive has one.

Look for: How long it took them to know, what they did about it, and what they changed in how they hire.

Red flag: A long tenure of the wrong person, explained as loyalty.

32. What would your last team say you were hardest about?

People answer this one honestly more often than they expect to, because it sounds like a strength question and is not.

Look for: A real edge named plainly, and some awareness of what it costs the people around them.

Red flag: A strength delivered in the shape of a fault.

Exit and Diligence Readiness

Ask these even if you have no transaction planned. A data room is the most rigorous examination a set of numbers ever gets, and a candidate who has survived one thinks about evidence differently from a candidate who has not. Mark has been through three exits, one of them a nine-figure private equity rollup, and the pattern is consistent: what gets discounted in diligence was visible eighteen months earlier and nobody had been asked to look.

33. Have you been on the sell side of a data room? What broke?

Specifics or nothing.

Look for: What was actually queried: revenue recognition, contract terms that did not match the model, a customer concentration nobody had stated plainly, working capital that behaved differently under scrutiny.

Red flag: They were involved but cannot name what was queried.

34. What would you fix eighteen months before a process, rather than three months before?

The whole difference between a defensible number and an expensive one sits in this answer.

Look for: Contract hygiene, consistent definitions held over time, and a clean audit trail built while it is cheap.

Red flag: Everything they name is something you would do in the final quarter.

35. How do you keep a revenue number defensible to somebody else’s accountants?

Tests whether they build for scrutiny by default.

Look for: Policies written down before they were needed, and applied consistently through a change of system or of team.

Red flag: Defensibility described as a reporting exercise.

36. Looking at what you know of our numbers, what would a buyer discount?

The most uncomfortable question here and the most useful. A candidate willing to answer it in an interview will be willing to answer it in a board meeting.

Look for: Something specific they have actually noticed about your business.

Red flag: Nothing, or a compliment.

If any of this landed closer to home than expected, the work of getting a business ready to be examined is its own project and it runs on a longer clock than a hire does. That is what exit planning and M&A work is pointed at.

Red Flags Across Every Answer

Some patterns show up regardless of which question triggered them, and they are worth more than any single response.

1. Nothing ever went wrong. A CFO with no scar tissue has either been lucky or is editing.
2. The last chief executive is the explanation. Once is context. Twice is a pattern you are about to inherit.
3. Tools instead of decisions. Asked what they did, they describe what they installed.
4. No number they got wrong. Forecasting is a discipline of being wrong in known directions. Somebody who cannot name a miss is not measuring.
5. They have never said no. Not to a colleague, not to a vendor, not to a board member.
6. Precision where they should be uncertain. Confident answers about your business from someone who has spent an hour with it.
7. Vagueness where they should be precise. Their own results, their own team, their own mistakes.

What a Strong Candidate Will Ask You

The questions coming the other way are data too, and a shortlist sorts itself surprisingly well on this alone. Expect a serious candidate to ask most of these.

• Why did the last CFO leave, and what would they say if I called them?
• What decision are you hoping this hire will make for you?
• Where do you and the board currently disagree?
• What in the numbers do you not trust?
• Who owns the forecast today, and who will own it after I start?
• What does this company need to be true in three years for this to have been the right hire?

A candidate who asks none of these is interviewing for a job rather than for your job. A candidate who asks all of them has done this before.

Frequently Asked Questions

How many rounds should a CFO interview take?

Three or four, and no more. A screen, a working session on your actual numbers, a board or investor conversation, and a reference-led final. Searches that run to six rounds are usually compensating for a decision nobody wants to own, and the strongest candidates withdraw first.

Who should be in the room?

The chief executive in every round, one board or investor representative for the strategic conversation, and at least one operating peer the CFO will have to negotiate with. Finance candidates are good at meeting chief executives. The peer conversation is where the friction shows up.

How do you test whether a CFO is genuinely fluent in AI without setting a technical test?

Ask what their team stopped doing, and what control sits around anything a model produces that reaches the numbers. Both questions are impossible to answer well from reading about the subject, and neither requires you to assess anything technical yourself.

When is a part-time finance lead the better hire?

When the work is real but not yet continuous: a clean close, a board pack that stands up, a model somebody trusts. Below roughly fifteen million in revenue, the full seat is often bought to signal seriousness rather than to do a job, and the money is better spent on a strong controller plus senior time by the day.

What should you ask a CFO candidate’s references?

One question does most of the work: what did they push back on, and what happened next. Ask it of the chief executive they reported to and of a peer they had to say no to. The second call is the one that tells you something.

Final Thought

The list above is longer than any interview, and it is supposed to be. Pick the sections that match the vacancy you actually have, keep the AI block whatever the shape of the search, and hold the diligence questions for the final round when the candidate is comfortable enough to be honest.

One last thing worth saying plainly. Hiring a CFO to fix a revenue problem is one of the more expensive mistakes a founder can make, and it is common, because finance is where the problem becomes visible. If the numbers are late, this is the hire. If the numbers are fine and the growth is not, the constraint sits somewhere else, which is the question a growth advisor is brought in to answer.