Financial analyst interview questions — and the CV claims behind them
The modelling, forecasting and business-partnering questions financial analysts actually get, what each is really probing, and which line on your CV an interviewer will press hardest.
What this interview is actually testing
Whether you understand the business behind the numbers — and whether the models on your CV are ones you built or ones you maintained.
The questions, and what each one is really asking
1.Walk me through a model you built from scratch.
What it probes: Built versus inherited, which is the central ambiguity on finance CVs. Interviewers ask how the structure was decided, because that is what a builder can answer and a maintainer usually cannot.
2.What were your key assumptions and how did you set them?
What it probes: Whether assumptions were reasoned, sourced from the business, or copied from last year. The last is extremely common and worth being honest about.
3.Tell me about a forecast you got badly wrong.
What it probes: Whether you review your own accuracy. Analysts who never revisit a forecast cannot improve one.
4.How do you explain a variance to a non-finance manager?
What it probes: Business partnering. The answer that starts with an account code has already lost the room.
5.What question did the business ask that the numbers couldn't answer?
What it probes: Whether you know the limits of your data — the same discipline as knowing what a claim can support.
6.How did you check your own work?
What it probes: Controls. Reconciliation, a second pair of eyes, sanity checks against an independent source. Everyone has broken a model; the signal is whether you catch it.
7.What would you have done differently in that budget cycle?
What it probes: Whether you experienced the cycle as a process to run or a system to improve.
The CV claim they'll press hardest
Ownership of a model or a saving — "built the group forecasting model", "identified £1.5M in cost savings".
Both are ambiguous in the same direction. Finance models are almost always inherited and extended, so "built" invites a question about who designed the structure and what you added — and "extended the group model with a new driver-based revenue build" is a strong, specific answer that "built" only muddies. Savings figures have the analyst's version of the attribution problem: you identified it, a budget holder decided, and the realised number may never have been checked against the forecast. Interviewers in finance are professionally trained to ask what was actually booked, and "identified in the analysis, £900k of it landed in the following year's budget" is a far better answer than a round headline figure you cannot reconcile.
Finance interviews have a reputation for being technical, and the technical part is real. But the technical part rarely decides them, because most shortlisted candidates can do the technical part.
What decides them is whether you understand the business the numbers describe.
The variance question is the whole interview
“How would you explain this variance to a non-finance manager?” looks like a communication question. It is a comprehension question.
An analyst who explains a variance by walking through the accounts is describing what happened. An analyst who says “we’re £180k over on contractor spend because two permanent hires slipped a quarter and we backfilled with agency at roughly twice the rate” is describing why — and they can only do that if they talked to the hiring manager rather than reading the ledger.
Prepare a variance you can explain that way. It is the single most useful thing you can bring into this room.
Built or inherited
Finance CVs share one ambiguity above all others, and interviewers open on it as a matter of routine: who built the model?
Almost every model in a going concern is inherited. Someone made it years ago, it accreted tabs, and successive analysts extended and repaired it. That is the normal condition, and maintaining a large model well is genuinely difficult work that nobody in finance disrespects.
Which is why “built the group forecasting model” is a poor trade. It invites the question of who designed the structure, and the honest answer weakens a claim that did not need making.
The specific version is stronger in every way: “the three-statement model predated me. I rebuilt the revenue build to be driver-based, added scenario toggles, and cut the close cycle by two days because the old version needed manual reconciliation every month.” Nobody hears that as a smaller achievement. They hear someone who knows exactly what they did.
Savings figures and what actually got booked
The other probed number is cost savings, and finance interviewers are trained to ask a specific follow-up: what was actually booked?
The chain is familiar — you identified an opportunity, built the case, a budget holder decided, some version was implemented, and the realised figure may never have been reconciled against the forecast. Every step attenuates the claim.
“Identified £1.5M in savings” therefore usually means the business case total. The better answer names the split: “the analysis identified about £1.5M; roughly £900k went into the following year’s budget, and I don’t have visibility on what was realised after I moved teams.”
In any other function that would be caveating. In finance it is the answer of someone who understands the difference between a forecast and an actual — which is, more or less, the job description.
Preparing
For every model on your CV, write what you inherited and what you changed. For every savings or improvement figure, write the business case number, the booked number, and the realised number, marking the ones you do not know.
Then pick a forecast you got wrong and prepare it. Almost nobody does, and it is asked constantly.
Our free claim checker flags the lines likeliest to invite a follow-up; the STAR guide covers the answer shape.
Questions about the interview itself
- What questions are asked in a financial analyst interview?
- A technical portion (Excel or modelling, sometimes accounting fundamentals or a three-statement question), a case or variance-analysis discussion, and behavioural questions about business partnering. Expect at least one question that requires explaining a financial concept to a non-finance stakeholder, because that is most of the day job at any level above junior.
- How do I talk about a model I didn't build myself?
- Say what you inherited and what you changed. 'The three-statement model was already there; I rebuilt the revenue build to be driver-based and added the scenario toggles' is more impressive than 'built the model', because it is specific and it survives every follow-up. Maintaining a complex model well is a real skill and nobody in finance thinks otherwise.
- Do financial analyst interviews test Excel live?
- Frequently, and increasingly with a shared screen. If your CV claims advanced Excel or financial modelling, assume it is testable. Being precise about what you actually use — index/match and pivot tables daily, VBA rarely — is safer than a blanket 'advanced', which sets an expectation the test will calibrate against.