Interview questions that reveal the job

Use the last ten minutes to investigate the work

The interviewer says, “What would you like to know?” You have a little time, and a fairly important decision to make. Spend it on questions whose answers could change that decision. A recent example usually tells you more than an invitation to describe the company at its best.

Jayden Reed, our fictional data-integrity specialist, wants more autonomy and creativity. He is comparing two invented employers: Lantern Relay, which favors frequent, bounded product experiments, and Cairn Ledger, which emphasizes traceable evaluations and peer review. Both advertise assistant-evaluation work. He needs to discover what a person in that role would actually get to decide.

Harvard’s interview guidance recommends researching the employer and role before the conversation. Jayden uses that preparation to form a few hypotheses, not to memorize compliments. “They mention ownership” becomes “Which decisions would I own?” “They mention rigorous evaluation” becomes “Who can change an evaluation that is measuring the wrong thing?”

Your questions may concern predictable hours, supervision, travel or pay instead. Choose what you need to know before accepting a job. You are allowed to be interested in how an ordinary Tuesday would feel.

Ask for an episode the other person can remember

For Lantern Relay, Jayden asks: “Could you walk me through a recent experiment that changed direction after you saw the results? Who could make that call?” He is asking about the relationship between evidence and authority. The answer may reveal a weekly review, a product lead’s decision or a team member stopping an experiment. Each would give him something to explore.

For Cairn Ledger, he asks: “When an evaluation case is ambiguous, how do you decide whether to revise the case or change the assistant?” A team that talks about careful evaluation should have interesting disagreements about what counts as a good answer. Jayden wants to understand where his judgment would fit.

Compare either question with “Do you value innovation?” A pleasant yes has almost no information in it. A remembered decision has people, constraints and consequences. If the interviewer cannot discuss a customer example, Jayden can ask for a hypothetical or a description of the process. Respecting confidentiality need not end the conversation.

Follow the interesting answer instead of your next bullet

Imagine a Lantern Relay manager says, “We rolled an experiment back because our test set missed a failure.” Jayden could rush to his next prepared question. I would rather he ask, “What happened next?” That gives the manager room to describe who was consulted, how the team repaired the test and whether the same failure appeared again.

An imagined Cairn Ledger colleague says, “Every evaluation change gets reviewed.” A rubric is the written set of criteria used to judge an answer. Jayden follows with: “Suppose I think the rubric is rewarding the wrong behavior. How would I propose a change, and what would help the team decide?” He is looking for a path from noticing a problem to improving the system. A review requirement alone does not tell him whether that path is open.

Listen for an answer that actually connects the steps. “We care a lot about quality” leaves his question unanswered. “Last month someone brought three counterexamples, we changed the rubric and reran the previous results” supplies a process he can ask about. A polished answer can be thin; an untidy, specific answer can be useful.

You do not need to cross-examine anyone. Choose two or three questions, follow the answer that matters and leave room for a conversation. Ask a future peer about day-to-day work and a manager about authority and expectations. A recruiter may be the right person for compensation or the hiring process.

Give just enough of your own context

Jayden has useful experience to bring into the conversation. He helped classify recurring catalog errors and turn the findings into a checklist used by three support teams. His personal learning project uses synthetic examples to test ambiguous data and structured outputs. Neither requires him to recite his entire résumé before asking a question.

“I’ve worked on a checklist for recurring data errors. How do your product and support colleagues get evaluation failures into the next version?” That opening explains why he cares, then gives the floor back. The interviewer can respond to a practical question instead of waiting through a second sales pitch.

A leading question is less helpful: “You let engineers own everything, right?” Jayden would be advertising the answer he wants. “Which decisions require review?” may produce an answer he likes less, but it is much better information before he accepts a job.

Use one sentence of relevant context. Then stop talking long enough to hear the reply. This is particularly useful when you are changing fields: a thoughtful question can show how you approach a problem without pretending you already know the team’s work.

Write down the answer before you turn it into a verdict

Afterwards, Jayden separates what he heard from what he inferred. “The manager described a weekly evaluation review” belongs in his notes. “Nobody has autonomy” is an interpretation. He might ask a future peer who decides what enters that review, or ask the manager what a new hire could change in their first few weeks.

He can make a small decision record: what matters to him, the example each person gave, and the question still open. Lantern Relay may offer faster ownership but less time to investigate a problem deeply. Cairn Ledger may offer room to design difficult tests while requiring more review before a release. He needs to find the particular trade-off he wants, rather than award a company a culture score.

Before your next conversation, choose one condition you need to do good work and prepare a question about a real situation. Add a follow-up you could use if the answer stays vague. Bring the note, then let the conversation surprise you. An answer that makes you reconsider the job has done useful work.