Accenture Data Analyst Interview Questions
A data analyst at a global consulting and technology services company often arrives at the start of an engagement, when a client has a broad question and data scattered across departments. Prepare to show how you turn a vague question into something you can measure, set a baseline before a change begins, check whether a new system is truly being used and present findings to senior people briefly. What to show here: clear findings a client's leaders can act on. Each question below has what the interviewer is checking, an answer frame and a sample spoken answer to adapt.
How Accenture hires, in its own words, and its questions for every job →
1. Tell me about a time you presented an analysis to senior people who had only a few minutes. How did you decide what to leave out?
- Describe the analysis and the audience.
- Explain how you found the one message.
- Describe what you cut and where you kept it.
- Share how the meeting went.
“I had spent weeks on an analysis of delivery delays and was given ten minutes with the client's operations head. My first draft had twenty slides. I asked myself what he needed to decide, which was where to put extra staff. I cut the talk to one page: the two depots that caused most delays, the likely reason and what I recommended. Everything else went into an appendix in case he asked. He asked two questions, and both were covered there. He approved a trial at one depot. The result was your result.”
Weak answers walk through the method step by step and reach the finding when the time is over.
2. A client says, tell us why our customers are leaving. Describe a time you received a question that broad, and how you turned it into something you could answer.
- Describe the broad question.
- Explain how you defined the terms and the decision it would inform.
- Describe the smaller questions you agreed.
- Share what you delivered.
“In a past project a manager asked me why customers were unhappy. Before opening any data, I asked what he would do with the answer. He wanted to decide where to spend a service budget. I then asked how he defined unhappy, and we agreed on customers who complained or cancelled. I broke the question into three parts: who they are, when in their journey it happens, and what they complain about. I wrote those down and he confirmed them. That kept the work to three weeks. The main finding was your result, and it fed straight into his plan.”
Weak answers start pulling data at once and return weeks later with charts that answer no decision.
3. A client is starting a large change programme that promises faster order processing. What would you measure before it begins, so that the benefit can be proved afterward?
- Define the promised benefit as a measurable quantity.
- Measure the current state over a fair period.
- Note other things that could move the number.
- Agree how and when the after measurement will be taken.
“I would first turn the promise into a clear measure, such as the time from order received to order confirmed, with agreed start and end points. Then I measure it now, over enough weeks to cover quiet and busy periods, and split by order type, since the mix may change later. I would record other things that affect the number, like staffing or seasonal peaks. All of this is written down and agreed with the client before the change starts. If we wait until the old system is switched off, the baseline can no longer be measured and the benefit becomes opinion.”
Weak answers plan to compare after go-live with a number somebody remembers from before.
4. A client rolled out a new system region by region over several months. Describe a query that compares processing time before and after the rollout in each region.
- Join each order to its region's rollout date.
- Label the order as before or after that date.
- Group by region and label, and compare typical times.
- Leave out the changeover weeks and check the order mix.
“I would keep a small table with one row per region and its rollout date. I join orders to it on region, and a CASE expression labels each order as before or after, based on its own region's date. Then I GROUP BY region and label, and compare the typical processing time. I prefer the median, since a few stuck orders can distort an average. I would leave out the weeks right around each rollout, when staff were still learning. I would also compare the same order types on both sides, so a change in mix is not mistaken for improvement.”
Weak answers use one calendar date for all regions and compare averages without looking at the order mix.
5. At the start of an engagement, the client sends you dozens of spreadsheets from different departments, each with its own layout and names for the same things. How do you get to data you can analyse?
- List the files with their owner and meaning.
- Profile each one: columns, types, gaps, duplicates.
- Map them to common definitions agreed with the client.
- Load with a script and keep the originals untouched.
“I would make an inventory first: each file, who owns it, what period it covers and what each column means. Then I profile them, looking at value ranges, empty cells and duplicates. The hard part is meaning. One department's customer may be another's account, so I sit with the owners and agree common definitions in a short data dictionary. I load everything with a script, never by hand, so each step can be repeated and explained. The original files stay unchanged. Every gap I find goes into a list that I share with the client early.”
Weak answers clean files by hand in place, or assume that columns with the same name mean the same thing.
6. A client's staff have been given a new tool to replace an old process. How would you measure whether they have truly adopted it?
- Define adoption as completing real work in the tool.
- Measure the share of work done there against the old route.
- Split by team, site and role.
- Look for signs of workarounds and ask people why.
“Logging in is not adoption. I would define it as completing the real task in the new tool from start to finish. So the main measure is the share of all such tasks done in the tool, compared with those still done the old way, by email or spreadsheet. I would split that by team and site, because adoption is usually uneven. I would also look for signs of workarounds, such as data entered in bulk at the end of the week. Numbers tell me where. To learn why, I would speak to a few people in the low-use teams.”
Weak answers report the number of accounts created or logins as proof that the tool is used.
7. Your analysis shows that a saving your own team promised the client in its proposal is unlikely to be reached. What do you do?
- Recheck the analysis and the assumptions in the proposal.
- Show your lead the gap and its causes.
- Work out what saving is realistic and what could close the gap.
- Support an honest talk with the client.
“I would check my work first, and then read the assumptions behind the promised saving, since proposals often use early estimates. If the gap is real, I take it to my lead quickly and in private, with the evidence: this is what was assumed, this is what the data shows. I would bring what is realistic and any ideas that could close part of the gap. The client needs to hear it from us, early, and not find out at the end. I would not soften the numbers. Trust lost over a hidden shortfall costs far more than an early correction.”
Weak answers adjust the analysis until it fits the promise, or stay silent because it is not their decision.
8. Why do you want to be a data analyst at Accenture, working on client questions that change with each engagement?
- Say what you enjoy about new questions and domains.
- Link it to advising clients.
- Give an example of analysis that led to action.
- Say what you want to learn.
“I enjoy the first weeks with a new problem, when I have to learn how a business works before the data makes sense. Accenture provides consulting and technology services to client companies in many industries, so each engagement would bring a new domain and a new question. I also like that the work does not end with a chart. It ends when someone decides something. In my last project, one page of findings led to a trial at a depot, and that mattered more to me than the model behind it. I want to become better at presenting to senior people.”
Weak answers want to work on one dataset for years and are uneasy about presenting to clients.
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.