Costs
Part of Five kinds of analytics careers, and what actually separates them
Analytics careers questions: a grounded overview, updated for 2027
Analytics careers assessed from the other side: fifteen interview questions sorted by who can actually answer them, and how to read a vague reply.
An interview is a two-way assessment that only one side usually prepares for. The questions below are about the data function you would be joining, and they are sorted by who can actually answer them.
Asking the right person matters more than asking the right question. A hiring manager cannot tell you what the job feels like on a Wednesday, and a peer cannot tell you what the mandate is.
What to take away
- Three questions predict most of the experiencewho sponsors the function, what happens when a request is refused, and who reviews the work.
- Ask the same question of two people. The gap between the answers is the finding.
- A vague answer to a specific question is itself an answer, and it is usually the honest one.
Ask the hiring manager
1. Who asked for this role to exist, and what would change if it were not filled?
Hiring manager questions
- Who asked for this role?
- What happens when team says no?
- Which decisions wait on this team?
- What was retired in the last year?
- How is success described to your manager?
You are looking for a specific person and a specific consequence. A role created because headcount was available has no sponsor, and sponsorship is what determines whether your work gets used.
2. What happens when the team says no to a request?
Ask for the last time it happened. If the answer is that it does not, the team has no prioritization and you will be working a queue you cannot control.
3. Which decisions currently wait on this team, and how long do they wait?
Turns an abstract mandate into something checkable. It also reveals whether the function is on the path of any decision at all, or sits beside it producing material.
4. What was retired here in the last year?
A function that has never removed anything is accumulating maintenance, and the accumulation lands on the newest person.
5. How is the team's success described to your own manager?
You want to hear about decisions or about problems that stopped, not about volume delivered. If it is volume, expect to be measured that way.
Ask a peer on the team
6. Where does a number come from, end to end?
Peer questions to ask
- Trace a metric end to end
- Shared definition layer or per-report?
- Who reviews your work before it goes out?
- What broke recently, and what changed?
- What is the most annoying part of the week?
Pick a metric they mention and ask them to trace it. A fluent answer means the lineage is understood; a hesitant one means the data lineage exists only in individual heads, and you will be the next head.
7. Is there a shared definition layer, or does each report define its own?
Whether there is a semantic layer or its equivalent tells you how much of your time will go to reconciling versions of the same number.
8. Who reviews your work before it goes out?
A named person and a real process, or nobody. This single answer determines how fast you will improve, and it is more important early in a career than almost anything else on offer.
9. What broke recently, and what happened afterwards?
You are listening for whether a failure produced a change or a blame conversation. Teams that write up incidents in the blameless postmortem style improve; teams that do not repeat the same failures.
10. What is the most annoying part of the week?
The genuinely useful question. Everyone answers it honestly, and the answer is usually the thing you will spend a third of your time on.
Ask a stakeholder, if you can get one
11. When did the data team last change your mind?
Stakeholder questions
- When did the data team last change your mind?
- What do you ask for that you wish you did not?
- What do you not trust?
If they cannot think of an instance, the function is producing material rather than influencing decisions, whatever the job description says.
12. What do you ask them for that you wish you did not have to?
Reveals the standing manual work that nobody has automated and that will land on you.
13. What do you not trust?
The specific numbers a stakeholder distrusts are a map of the function's unfinished work, and the person telling you will be unusually candid because they are not being assessed.
Ask yourself afterwards
14. Which axis of my own development does this role advance?
Technical depth, domain understanding, communication, or ownership. A role that advances none of them is a lateral move regardless of the title.
15. Whose judgment will I be exposed to?
Over two years, this matters more than the tooling, the sector, or the title. Name the person and ask yourself whether you want to think the way they think.
Reading the answers
Vagueness on question one is disqualifying more often than any technical weakness in the team. Vagueness on question eight is a warning if you are early in your career and much less so if you are senior and can supply your own standards.
If questions six and seven both go badly, the function has no governed foundation, and expect the first year to be definitional work rather than analysis. That can be excellent work, but it is not what most job descriptions promise, so know which one you accepted.
The shape of that work is described under data governance.
Related reading on this site
The kinds of work these questions are probing are described in analytics careers, with a way of placing yourself in four axes and a planning pass. For the mis-scoped role patterns these questions detect, see nine career mistakes and five career composites. The function you are assessing is described from the inside in business intelligence.
Common questions
Fifteen questions is more than an interview allows.
Pick three for the hiring manager and three for the peer. Questions one, eight and ten alone will tell you most of what you need.
Is it acceptable to ask to speak to a stakeholder?
Yes, and the response is informative either way. A function confident in its relationships will arrange it; one that hesitates is telling you something without meaning to.
What if the honest answers are all bad but I need the job?
Take it with your eyes open and a plan. Knowing that the first year is definitional work, with no review and an unclear sponsor, lets you set up your own review and look for a sponsor deliberately.
Should I ask about tooling at all?
Briefly, and last. Tooling is the easiest thing to learn and the easiest thing to change, and time spent on it in an interview is time not spent on the things that will not change.
