Financial Analysis interview preparation

Questions

100

Traced to a firm

42

Firms

28

Updated

September 2026

Asked at

All firmsMoody's7Bain Capital3SSState Street3AMAres Management2BLBlackRock2DED.E. Shaw2MSMorgan Stanley2Oaktree Capital Management2S&P Global2Bridgewater Associates1Citadel1FTFranklin Templeton1Golub Capital1HWHarris Williams1Houlihan Lokey1J.P. Morgan1Jane Street1MWMarshall Wace1Millennium Management1Morningstar1PIMCO1Sycamore Partners1TSTruist Securities1Two Sigma1Vanguard1WMWellington Management1Wells Fargo Securities1Wolverine Trading1

Topic

All topicsThree statements9Accounting policy and standards5Working capital and cash7Ratio analysis8Forecasting and budgeting9Variance and management reporting7Unit economics and costing8Capital budgeting7Cost of capital and valuation7Markets and rates5Modelling, Excel and data8Business partnering6Brainteasers and estimation4Fit and career10

Level

AnyCoreIntermediateHard

Type

AnyTechnicalCaseBrainteaserMarket viewFit

1. 040Revenue beat budget by 6 percent but gross margin came in 200 basis points below. Explain it.Variance and management reportingHardtechnicalCorporate FP&ABusiness finance

   Say this

   Most likely you bought the revenue. Either you discounted, or the growth came from the lower-margin part of the portfolio, or input costs rose and you could not pass them on. A mix and price decomposition tells you which within an hour.

   Then walk it

   1. Run price, volume and mix on gross margin. That immediately separates discounting from mix, which are the two dominant causes and have completely different implications.
   2. Discounting shows up as an adverse price variance concentrated in specific customers or the last few weeks of the quarter. That is a commercial discipline problem and it repeats next quarter.
   3. Mix shows up as volume favourable and realisation down with list prices intact. If the growth came from the entry-level SKU or from a low-margin channel like a marketplace, margin falls by design and the right response may be to celebrate it.
   4. Input cost is the third: raw material, freight, power, or an unfavourable FX rate on imports. Check purchase price variance against standard and check whether a price increase was due and did not happen.
   5. Then the accounting-only explanations, which are worth eliminating early: absorption of fixed overhead over a different volume, an inventory provision taken into cost of goods sold, or a reclassification between cost of sales and operating expenses.
   6. The conclusion I would present: 6 percent more revenue at 200 basis points less margin on a 30 percent gross margin base is roughly flat gross profit in rupees. So the honest headline is that we grew revenue and earned nothing extra for it, and here is which of the four causes did it.

   Where candidates lose it

   Reporting the revenue beat as good news. Convert both movements into rupees of gross profit before you conclude anything. And do not offer a cause without the decomposition, because guessing between discounting and mix is a coin flip.

   Expect next

   • How would you stop end-of-quarter discounting?
   • What if the growth is all in the new low-margin channel?
   • How does fixed overhead absorption distort this?
2. 043What does 'good' look like?Variance and management reportingIntermediatetechnicalGolub CapitalAnalytics · Chicago · 2023

   Say this

   Good is defined against a benchmark and a decision, never in the abstract. So my answer is that I would not accept the question without asking what we are measuring, compared with what, and what we would do differently at each answer.

   Then walk it

   1. The three benchmarks worth naming: our own history, our plan, and someone external, either a competitor or a best-in-class function. A number that beats last year and misses the plan and lags the peer group needs all three to be understood.
   2. Then define it as a level plus a direction plus a consistency. A 14 percent margin that is stable and improving is good; the same 14 percent that swung from 20 to 9 to 14 is not, even though the average is identical.
   3. Then attach it to a decision. For a reporting function, good might be a five-day close with zero post-publication restatements and forecast accuracy inside 5 percent. For a portfolio company it might be EBITDA conversion to cash above 80 percent. If nothing changes at the threshold, the metric is decoration.
   4. In an analytics or credit seat I would answer it about the work itself: good means the number is right, it is reproducible by someone else from the source, it arrives before the decision is made, and it comes with the one sentence that says what to do about it.
   5. And I would be explicit about what good is not: not the most detailed, not the prettiest dashboard, not the most conservative. Those are all ways of avoiding a judgement.
   6. So the short version: good is a defined threshold, against a named comparison, that changes a decision when it is crossed.

   Where candidates lose it

   Answering with adjectives. The question is deliberately open and it is testing whether you instinctively ask 'compared with what, and what would we do differently'. Push back for the benchmark, then give a concrete threshold.

   Expect next

   • Then what does good look like for a reporting analyst?
   • How would you set the threshold if you had no peer data?
   • What does bad look like?

   Reported by candidates at Golub Capital (Analytics, Chicago, 2023). Source: Wall Street Oasis.

3. 045You built a dashboard and nobody uses it. What went wrong?Variance and management reportingIntermediatetechnicalCorporate FP&AGCC finance centres

   Say this

   Usually one of three things: it answers a question nobody asked, it arrives after the decision, or people do not trust the numbers. I would go and watch three users for twenty minutes each before touching the design.

   Then walk it

   1. The relevance failure is the most common. Finance builds what finance finds interesting. If the sales head decides territory allocation weekly and the dashboard shows monthly margin by legal entity, it is irrelevant to them however accurate it is.
   2. The timeliness failure: a perfect pack on day ten when the operating review is on day six. Late and right loses to early and roughly right, every time.
   3. The trust failure: one number that disagreed with the system of record, once, and the whole dashboard is dead. Recovering trust takes a documented definition per metric and a visible reconciliation to the source.
   4. Then the design failures, which are real but secondary: too many metrics, no comparison so the viewer cannot tell good from bad, no drill-down to the transaction, and no commentary telling them what changed.
   5. So my fix sequence is: interview users about the decisions they make and when, cut the metric count hard, reconcile every metric to the ledger and publish the definitions, then land it before the review meeting and include three lines of written commentary.
   6. And I would measure adoption directly, because usage logs are the only honest feedback. If a page has three views a month, delete it rather than defend it.

   Where candidates lose it

   Answering with visual design fixes. The failure is almost never chart choice; it is relevance, timing or trust. Saying you would watch users and check the reporting calendar is what marks out someone who has done this in a real organisation.

   Expect next

   • How would you rebuild trust after one wrong number?
   • What would you cut from a 30-metric dashboard?
   • Actual, budget, forecast or prior year: which comparison leads?
4. 051A customer wants 20,000 units at a price below your full cost. Do you take the order?Unit economics and costingHardtechnicalCost accountingBusiness finance

   Say this

   If the price is above variable cost and you have spare capacity, it adds profit, so on the arithmetic yes. But I would only recommend it if the order does not displace better business and does not reset the price for everyone else.

   Then walk it

   1. The arithmetic first. Full cost 100, of which 70 variable and 30 absorbed fixed. Offer price is 85. Every unit adds 15 of contribution, so 20,000 units adds 3 lakh of profit, because the fixed 30 is being paid anyway.
   2. So the accounting answer is clear, and the reason candidates get this wrong is they compare price with full cost. Fixed cost is irrelevant to an incremental decision unless the order causes it to change.
   3. Then the conditions. Is there genuinely spare capacity, or does this displace full-price volume? If it displaces, the relevant cost includes the contribution you give up, and the answer usually flips.
   4. Does it trigger a step cost? Overtime, a second shift, additional tooling, extra freight or a special packaging run all count as incremental cost even though they look fixed on the standard cost sheet.
   5. Then the commercial risks, which is where finance earns its seat. Price leakage to existing customers, grey-market resale back into your own market, most-favoured-customer clauses, and the precedent that this buyer now expects 85 forever. In an export or institutional channel those risks are managed by segmentation and contract terms.
   6. So my recommendation would be: accept as a contained one-off with a defined volume cap, different packaging or channel, and a written statement that it is not a list price change. And I would name the margin dilution it will show in the monthly pack so nobody is surprised.

   Where candidates lose it

   Rejecting it because the price is below full cost. That is the textbook error the question exists to catch. But saying yes with no conditions is the other half of the trap, because the real answer includes displacement and price-leakage risk.

   Expect next

   • What if you are already at full capacity?
   • How would you stop the price leaking to existing customers?
   • Where does the contribution show up in the monthly variance pack?
5. 052Here are a few figures about an airline. Work out what it should charge for a ticket, and ask me for anything else you need.Unit economics and costingHardcase studyBain CapitalGeneralist · Boston · 2024

   Say this

   I would build cost per available seat kilometre, convert it to cost per seat on the route, divide by the load factor to get cost per sold seat, then add a margin. Before that I need four things: seats per aircraft, sector length, load factor and the split of fixed versus variable cost.

   Then walk it

   1. The structure: total operating cost per flight divided by seats gives cost per seat. Divide by the expected load factor, say 80 percent, and cost per sold seat rises by 25 percent. That step is the one candidates skip and it is the largest single adjustment.
   2. A worked illustration. If a flight costs 15 lakh to operate with 180 seats, that is about 8,300 per seat. At 80 percent load, cost per sold passenger is about 10,400. Add a 10 percent margin and the average fare needs to be around 11,500.
   3. Then I would ask what I am solving for, because the answer differs. The average fare needed to break even on the route is one question; the price of the marginal seat two days before departure is another, and there the only relevant cost is a few hundred rupees of fuel, catering and commission.
   4. That marginal-cost logic is why airlines use dynamic pricing. The same seat is worth 3,000 in a seat-sale ten weeks out and 18,000 to a business traveller on the day, and the fixed cost of the flight is sunk either way.
   5. The inputs I would keep asking for: fuel as a share of cost, aircraft ownership or lease cost per hour, crew and airport charges, ancillary revenue per passenger, and the competitive fare on the route. Ancillary matters enormously for a low-cost carrier; baggage and seat fees can be 15 to 20 percent of revenue.
   6. And the conclusion I would state: cost tells you the floor, competition and willingness to pay tell you the price. In a market with a dominant low-cost competitor, the cost-plus number is often simply unachievable, and then the decision is whether to fly the route at all.

   Where candidates lose it

   Dividing cost by total seats and quoting that as the fare. You must divide by load factor. The second trap is not asking questions: the interviewer deliberately gave you partial data, and the questions you ask are half of what is being marked.

   Expect next

   • What is the marginal cost of the last seat sold?
   • How would ancillary revenue change your answer?
   • A competitor prices 20 percent below your floor. What do you do?

   Reported by candidates at Bain Capital (Generalist, Boston, 2024). Source: Wall Street Oasis.

6. 053How would you build a loyalty programme for a rideshare business, and how would you know if it worked?Unit economics and costingHardcase studyJane StreetProduct and Strategy · New York · 2026

   Say this

   Treat it as an investment with a measurable return, not a marketing scheme. The programme costs you contribution per redeemed reward and buys incremental trips from riders who would otherwise switch. If you cannot measure the incremental trips, do not launch it.

   Then walk it

   1. Start with the economics of one trip: fare, driver payout, payment and support cost, leaving a contribution of maybe 15 to 20 percent of fare. Every rupee of reward comes straight out of that, so the programme has to move behaviour, not just reward it.
   2. Segment before designing. The high-frequency commuter is already loyal and paying them is pure margin leakage. The target is the mid-frequency multi-app user, four to eight trips a month, who is genuinely switchable. That is where incremental trips live.
   3. Design levers: earn rate, tiers with a threshold just above the target segment's current frequency, rewards that cost you less than they are worth to the rider such as priority matching or a waived cancellation fee rather than cash discounts, and expiry to cap the liability.
   4. Then the two supply-side pieces people forget. Loyalty that promises faster pickup requires driver density, so the reward may need a driver-side incentive to be deliverable. And a growing points balance is an accounting liability under Ind AS 115, deferred revenue for unredeemed points.
   5. Measurement is the whole answer: run it as a geo or user-level randomised holdout. Compare trips per user, retention and contribution per user between treated and control. Without a control group you will credit the programme with trips it did not cause, which is how most loyalty programmes are declared successful.
   6. The kill criteria I would write down before launch: incremental contribution per rupee of reward cost above one within two quarters, and no more than a set share of rewards going to users whose frequency did not change. If it fails either, shut it.

   Where candidates lose it

   Designing features without unit economics or a control group. The interviewer wants contribution per trip, a target segment that is actually switchable, and a holdout test. Cash discounts to your existing best customers is the answer that fails.

   Expect next

   • How would you size the incremental trips before launching?
   • What is the accounting liability for unredeemed points?
   • Would you fund it from the driver side or the rider side?

   Reported by candidates at Jane Street (Product and Strategy, New York, 2026). Source: Wall Street Oasis.

7. 058I give you a list of possible projects with their values and their costs, and a fixed budget. How do you choose?Capital budgetingHardcase studyBridgewater AssociatesGeneralist · New York · 2025

   Say this

   Rank by value per rupee of the constrained resource, not by absolute value. Compute the profitability index, NPV divided by the capital required, take them in descending order until the budget runs out, then check the combinations near the cut-off because the greedy answer is not always optimal.

   Then walk it

   1. Profitability index is present value of inflows over the initial investment, or equivalently one plus NPV over investment. Anything above one adds value; ranking by it maximises value per unit of the scarce resource.
   2. A quick illustration. Budget 100. Project A: NPV 30, cost 60, index 0.50. Project B: NPV 18, cost 40, index 0.45. Project C: NPV 16, cost 40, index 0.40. Greedy picks A then B for 48 of NPV on 100 spent. Picking B and C gives 34. So A plus B wins, but you only know that because you checked.
   3. The reason you check is indivisibility. Projects cannot be taken in fractions, so this is a knapsack problem, and greedy ranking can leave budget stranded. With a handful of projects, enumerate the combinations; with many, solve it as an integer programme, which Excel Solver will do.
   4. Then the constraints that make it a real decision rather than an arithmetic one: mutual exclusivity where two projects do the same thing, dependencies where B requires A, and non-capital constraints like scarce engineering time, which may be the binding resource rather than money.
   5. Then the multi-period version. A project may be delayable, so the question becomes which projects this year and which next, and a one-year delay on a positive-NPV project has a real cost you should quantify rather than assume away.
   6. And I would flag the strategic overlay: some low-index projects are mandatory, safety, regulatory or IT security, so they come out of the budget before ranking begins. Pretending everything competes on index is how compliance projects get deferred until they become a crisis.

   Where candidates lose it

   Ranking by NPV alone, which strands capital, or by IRR, which ignores scale. The word the interviewer wants is profitability index, followed immediately by the acknowledgement that indivisibility makes greedy ranking imperfect.

   Expect next

   • What if two of the projects are mutually exclusive?
   • What if the binding constraint is engineers, not money?
   • How would you handle a project you can delay by a year?

   Reported by candidates at Bridgewater Associates (Generalist, New York, 2025). Source: Wall Street Oasis.

8. 074You inherit a model from someone who left last week and the CFO wants to use it on Monday. How do you audit it?Modelling, Excel and dataHardcase studyFinancial modellingBig Four

   Say this

   Work from the outside in. First sanity-check the outputs against reality, then trace the three or four numbers that drive them, then stress-test rather than read every cell. You cannot audit 40,000 formulas by Monday, so you audit the ones the answer depends on.

   Then walk it

   1. Start with the outputs. Does revenue growth, margin and cash flow look plausible against history and against the sector? An implied 45 percent EBITDA margin on a distribution business tells you more in ten seconds than an hour of cell tracing.
   2. Then the structural checks: does the balance sheet balance, does the cash flow tie to the cash movement, does historical data reconcile to the published accounts. Any break here and nothing downstream is trustworthy.
   3. Then mechanical scanning. Use formula view or a spreadsheet audit tool to find hard-coded constants inside formulas, inconsistent formulas within a row, broken links to external files, circular references and unintended ranges. A row that is consistent for eleven periods and different in the twelfth is exactly what you are hunting.
   4. Then stress tests, which are the fastest way to find logic errors. Set revenue growth to zero, set it to 50 percent, set price to nil. If EBITDA does not move sensibly, or the balance sheet breaks under a stress, you have found a hard-code or a broken link.
   5. Then trace the top three value drivers back to their source, and only those. For a DCF that is usually WACC, terminal growth and the revenue build. Verify each against a document, not against another cell in the same model.
   6. And I would tell the CFO on Friday what I had and had not verified, in writing, with a short list of numbers I do not yet trust. Presenting an unaudited model as clean is the career risk here, not the model itself.

   Where candidates lose it

   Saying you would check every formula. There is not time and it is the wrong method. Outputs first, structural checks, mechanical scan for inconsistencies, then stress tests. And say you would disclose what you could not verify.

   Expect next

   • What is the fastest way to find a hard-coded number?
   • You find an error that changes the answer 20 percent. What do you do?
   • How would you hand this model on properly?
9. 075Your three-statement model does not balance. How do you find the break?Modelling, Excel and dataIntermediatetechnicalFinancial modellingCorporate FP&A

   Say this

   Find the first period where the imbalance appears, then look at the size of the difference, because the number usually names the culprit. Most breaks are one of four things: net income not flowing to retained earnings, a balance sheet movement missing from the cash flow, a sign error, or dividends and capex mishandled.

   Then walk it

   1. Step one, locate. Check the balance row across all periods and find the first column that breaks. Everything after it is contamination; the error is in that one period.
   2. Step two, read the difference. If it equals net income, retained earnings is not picking up the P&L. If it equals twice something, you have a sign error. If it equals depreciation, the add-back is missing or double-counted. The magnitude is the diagnosis.
   3. Step three, check the two mandatory links: net income flows to retained earnings less dividends, and closing cash from the cash flow statement equals the balance sheet cash line. Those two account for most breaks.
   4. Step four, confirm every balance sheet line has a corresponding cash flow movement. A new line added to the balance sheet, a lease liability, a deferred tax balance, an FX reserve, and not wired into the cash flow, is the classic mid-project break.
   5. Step five, check the debt schedule and capex. Gross versus net movements in borrowings, and a capex figure taken from the P&L depreciation rather than the fixed asset schedule, both produce clean-looking models that do not balance.
   6. And the prevention, which is what I would say last: build the balance check from the very first day and keep it on screen. Models that balance from row one never accumulate a break you have to hunt for later.

   Where candidates lose it

   Hunting cell by cell from the top. Locate the first broken period, then let the size of the difference identify the cause. Not knowing that the difference often equals net income or depreciation is what makes this take an afternoon instead of five minutes.

   Expect next

   • The difference equals the depreciation charge. What is wrong?
   • Where would you put the balance check?
   • What if it balances but the cash flow does not tie?
10. 080How would you make money with social media data?Modelling, Excel and dataHardcase studyTwo SigmaGeneralist · New York · 2024

    Say this

    Find a place where social data leads a number somebody pays for, and prove the lead is real before you monetise it. The two honest routes are forecasting revenue ahead of official data, and selling the cleaned signal to people who trade or plan on it.

    Then walk it

    1. Start from the decision, not the data. Who pays for being two weeks early? A consumer company deciding on promotional spend, a retailer planning inventory, an investor sizing a quarter. Each of those is a different product.
    2. The forecasting route: mention volume, sentiment and follower growth for a brand, mapped against reported quarterly revenue. If mentions of a product line lead revenue by four weeks with a stable relationship, you have a nowcast that beats consensus.
    3. Then validate it properly, because this is where the question is really marked. Back-test out of sample, check the relationship holds across at least two or three cycles, and control for the obvious confounders: a viral post is not demand, a paid campaign inflates mentions, and bot activity is a large share of raw volume.
    4. Then the monetisation models: sell the cleaned data feed, sell a derived index, or use it internally to trade or to plan. The data business scales better and is easier to defend legally; the internal use captures more of the value but only while the signal stays private.
    5. Then the hard constraints, which a good answer names unprompted. Platform terms of service and API restrictions, privacy law, no personal data, no material non-public information, and signal decay: any edge from public data erodes as more people buy the same feed.
    6. So my proposal would be a narrow pilot. One sector, five brands, six quarters of history, one testable claim: can we predict reported revenue better than consensus. If the back-test fails, stop, because most alternative data claims do not survive it.

    Where candidates lose it

    Jumping to 'sentiment analysis predicts stock prices'. Everyone says it and almost nobody validates it. The marks are in the back-test, the confounders such as bots and paid campaigns, the legal constraints, and naming who pays for the output.

    Expect next

    • How would you back-test that claim?
    • What would you do about bot-generated volume?
    • What happens to the edge once you sell the data?

    Reported by candidates at Two Sigma (Generalist, New York, 2024). Source: Wall Street Oasis.

Firm tags come from public, anonymous candidate reports on Wall Street Oasis: strong signal, not sworn testimony. Firms are named as the places a question was reported, not as partners of Fin Maverick. Answers are written for this page to show how to think out loud; they are not scripts to recite.