Investment Banking interview preparation
Every question below is either traced to a named firm from a public candidate report, or tagged at desk level when we could not trace it. Answers are written the way you would actually say them out loud — answer first, then the mechanism, then the limitation.
100 questions, mapped to the firms that asked them
- Questions
- 100
- Traced to a firm
- 100
- Firms
- 46
- Updated
- September 2026
010I am showing you a set of financials where one number is wrong. Find the error.JefferiesInvestment Banking · New York · 2025
Say this
I would check whether the balance sheet balances first, then whether the cash flow statement ties to the change in cash, then whether the income statement subtotals add. One of those three checks will catch almost any planted error.
Then walk it
- First test: do assets equal liabilities plus equity? If not, the error is a balance sheet line or a missing retained earnings flow.
- Second test: does ending cash on the cash flow statement match the balance sheet cash? A break here points to a missing working capital or financing line.
- Third test: does net income on the cash flow statement match net income on the income statement? People plant errors right there.
- Then sanity-check the ratios. A gross margin that moved 800 basis points with no explanation, or D&A larger than gross PP&E, is usually the plant.
- I would say my checks out loud as I run them, so you can see the process even if I have not found it yet.
Where candidates lose it
Going silent and hunting line by line. This question tests whether you have a systematic tie-out routine, not whether you have sharp eyes. Narrate the three checks; the process is the answer.
Expect next
- You found it. What would you do next if this were a live client model?
- Which single ratio tells you most about earnings quality?
- How would you audit a model you inherited from a departing analyst?
Reported by candidates at Jefferies (Investment Banking, New York, 2025). Source: Wall Street Oasis.
012Do a DuPont analysis for a hospital business.Credit SuisseInvestment Banking · Mumbai · 2020
Say this
DuPont splits return on equity into net margin, asset turnover and leverage. For a hospital the story is almost always thin margins, heavy assets and therefore low turnover, with leverage doing a lot of the work on ROE.
Then walk it
- ROE equals net margin times asset turnover times the equity multiplier. Three levers, and each one tells a different operating story.
- Net margin for a hospital is driven by payer mix and case mix. Private-pay and high-acuity surgical work carry far better margin than government-scheme volume.
- Asset turnover is structurally low, because you have bought land, a building and imaging equipment. The operating metric behind it is occupancy and average revenue per occupied bed.
- That heavy asset base is why leverage matters so much. Hospitals fund expansion with debt, so the equity multiplier is doing real work in the ROE.
- The banker's conclusion: a hospital chain improves ROE mainly by filling existing beds and shifting case mix, not by cutting costs. Incremental occupancy has almost no marginal cost.
Where candidates lose it
Reciting the DuPont formula and stopping. The question names a hospital on purpose. If you cannot say what drives each of the three terms for that specific business, you have shown formula recall and nothing else.
Expect next
- Which of the three levers would you push first?
- What metrics would you ask the CFO for?
- How would this look different for a diagnostics chain?
Reported by candidates at Credit Suisse (Investment Banking, Mumbai, 2020). Source: Wall Street Oasis.
014Walk me through what OpenAI's income statement probably looks like.LazardInvestment Banking · San Francisco · 2026
Say this
Large and fast-growing revenue from subscriptions and API usage, a gross margin far below normal software because inference costs real compute, then enormous R&D and compute spend that puts operating income deeply negative.
Then walk it
- Revenue splits into consumer subscriptions, enterprise seats, and API consumption. The API line is usage-based, so it behaves more like a utility than like seat-based SaaS.
- Cost of revenue is the interesting part: every query costs GPU time. That is why gross margin sits well below the 75 to 85 percent you would expect from software.
- Below that, R&D dominates, and most of it is training compute plus a small number of very expensive people.
- Sales and marketing is unusually light for the growth rate, because distribution has been largely organic.
- So the shape is high growth, compressed gross margin, and a big operating loss funded by capital rather than cash flow. If I were valuing it I would care most about whether inference cost per query is falling faster than usage is rising.
Where candidates lose it
Treating it as generic SaaS with 80% gross margins. The entire point of the question is whether you understand that inference is a variable cost of goods sold. Name that and you have answered it, even if every number you guess is wrong.
Expect next
- How would you value it then?
- What would you need to believe for this to be worth its last round?
- Compare the business model to Microsoft's.
Reported by candidates at Lazard (Investment Banking, San Francisco, 2026). 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.