Goldman Sachs Data Analyst Interview Questions
Data analysts at an investment bank support people who trade in markets and advise companies on deals, so the work covers client orders, how well they were executed, and figures about companies gathered from many sources. Prepare to describe queries in words, measure the quality of execution, match one company across different data sets, and explain what made a desk's result change from one day to the next. What to show here: figures you can defend line by line to a demanding reader. Each question below has what the interviewer is checking, an answer frame and a sample spoken answer to adapt.
How Goldman Sachs hires, in its own words, and its questions for every job →
1. Tell me about a time you found an error in an analysis after it had already gone to senior people. What did you do?
- Describe the analysis and who had received it.
- Explain how you found the error and how large it was.
- Show how you corrected it and told people.
- Share what you changed in your process.
“I once sent a summary to two senior managers and the next morning noticed I had counted some transactions twice because of a join on a field that was not unique. The main conclusion still held, but one figure was too high. I did not wait to be asked. I sent a corrected version within the hour, with a short note that said what was wrong, what had changed and what had not. One manager thanked me for being direct. I now check the number of rows before and after every join. Since then, your result.”
Weak answers hoped that nobody would notice, or corrected the file quietly without telling the readers.
2. Describe a time a team preparing for a client meeting needed an analysis from you within hours. How did you decide what to include and what to leave out?
- Describe the request and the deadline.
- Explain how you narrowed it to the real question.
- Show what you delivered and the limits you stated.
- Share the outcome.
“A team asked me one afternoon for a full review of a sector for a client meeting the next morning. A full review was not possible, so I asked what decision the client faced. It came down to how three companies compared on growth and debt. I built that comparison carefully from sources I trusted, and listed at the top what I had not covered and one figure I could not confirm. They used it as it was. I learned that a small analysis with its limits stated is better than a large one nobody can rely on. The result was your result.”
Weak answers tried to deliver everything and checked nothing, or filled gaps with estimates presented as facts.
3. You have a table of client orders and a table of fills, where one order can be filled in several parts. Describe in words the query that lists orders still only partly filled at the end of the day, and how much of each remains.
- Sum the filled quantity for each order in the fills table.
- Join that total to orders so orders with no fills are kept.
- Work out what remains, treating no fills as zero.
- Keep orders where the remainder is above zero and leave out cancelled ones.
“I would first sum the fills for each order in a subquery, using GROUP BY on the order id. Then I would LEFT JOIN that result to the orders table, so that an order with no fills still appears. The filled amount uses COALESCE to turn a missing total into zero. The remainder is order quantity minus filled quantity. The WHERE clause keeps rows where the remainder is above zero and the order was not cancelled. Summing before the join matters, because joining first and then adding other tables can count fills more than once.”
Weak answers use an inner join and lose orders with no fills, or join first and count fills twice.
4. A client asks how well the bank executed its share orders last quarter. Which measures would you use, and what would you compare them with?
- Compare the price achieved with the market price when the order arrived.
- Compare it with the average market price over the period of the order.
- Add how much was filled and how quickly.
- Group by order size and market conditions.
“I would begin with the difference between the average price achieved and the market price at the moment the order arrived, since that shows what waiting and trading cost. I would also compare with the average market price over the time the order was worked, weighted by volume. Then I would add how much of each order was completed and how long it took. One average hides a lot, so I would group the results by order size against normal trading volume and by how much the market was moving. I would show the difficult orders as well as the easy ones.”
Weak answers give one average for all orders, or choose the comparison that makes the result look best.
5. The same company appears under different names and identifiers in three data sources you need to combine. How do you match the records, and how do you know the matching is right?
- Match on standard identifiers where they exist.
- Clean names and match the rest with rules and similarity.
- Review uncertain matches by hand.
- Keep a mapping table and measure errors on a sample.
“I would match first on standard identifiers, such as those used for listed securities or legal entities, because they are the most reliable. For the rest, I would clean the names by removing endings like limited or incorporated and fixing case and spacing, then compare using similarity along with country and sector. Matches that score in the middle go to a person for review. Parent companies and their subsidiaries need special care, as the names are close but they are different entities. I would keep the result as a mapping table with the source of each match, and check a random sample by hand to estimate the error rate.”
Weak answers join on the company name as typed, or accept fuzzy matches without any review.
6. A trading desk's profit for the day is very different from the day before. How would you break the change down to explain where it came from?
- Separate positions held from the day before from trades done today.
- For held positions, estimate the effect of each market move.
- Add the result of new trades, fees and funding.
- Compare the sum with the real figure and look into the remainder.
“I would split the day's result into parts. First, the positions held at the start of the day: how much did their value change because prices, rates or currencies moved? I would estimate each of those effects separately. Second, the trades done today: the difference between the price traded and the closing value. Third, fees, commissions and the cost of funding. I would add the parts and compare the total with the reported figure. The remainder should be small. If it is large, something is missing, such as a late booking or a changed input, and that is where I would look.”
Weak answers describe the market in general, or never check that the parts add up to the total.
7. What interests you about analysing data for people who trade in markets and advise on deals, and why Goldman Sachs?
- Say what you enjoy about this kind of analysis.
- Tie it to the users and their decisions.
- Link it to your experience.
- Say what you want to learn.
“I like analysis that is read closely by people who will challenge it. Traders and bankers make decisions quickly and ask where every figure came from, which makes me more careful. Goldman Sachs works in markets and on deals for companies and investors, so the data covers both fast trading records and slower company figures. I enjoy both kinds. In my past work on your project, the hardest and most useful part was making data from different sources agree. I want to learn more about how the products work, so that my analysis answers the question behind the request.”
Weak answers speak about liking numbers in general and cannot say who would read their work.
8. A deal team needs a comparison of similar companies by morning. One company reports in a different currency and its financial year ends in a different month from the rest. What do you do?
- Put the periods on the same basis as far as the data allows.
- Convert currency using suitable rates.
- Show the original figures beside the adjusted ones.
- State every adjustment and its limits.
“I would not place the raw figures side by side, because the comparison would mislead. For the year end, I would build a matching period from the company's interim reports, adding the latest months and removing the same months from the year before. For currency, I would convert results earned over a period at the average rate for that period, and balance sheet items at the rate on the closing date. I would show the original figures beside the adjusted ones and list each adjustment in a note. If the interim data is not detailed enough, I would say so plainly.”
Weak answers compare the raw numbers, or make adjustments without telling the deal team.
No figures on this page: pay depends on the level, the city and the offer in front of you. When the offer comes, the free salary negotiation script helps with that part.
The questions above are the prep. On the real call, ClapAssist helps with the ones you did not practice.
ClapAssist is an AI interview assistant for Mac and Windows. It listens to the interview on your computer and shows you what to say, in short lines you can read while you talk. Your live interview audio and screen are never stored. Your resume and notes are saved to your account so the app fills them in on any computer. It is hidden from screen share on every plan, including Free; only you can see it.