What you ask her
Every answer you give is one she has heard a version of from four other candidates. The questions are where you are unlike anyone else in the process — and they are the only way you find out whether this job is the one you think it is.
Four mechanics, and they matter more than which question you pick.
One sentence, then stop. The commonest failure is burying a good question inside forty seconds of set-up — “so, having worked in a few places where the hand-off was difficult, and obviously every org is different, I was wondering…”. By the time the question arrives she is answering the preamble. Say the question. Say nothing else.
Never two at once. A double question gets one answer, and it will be the easier half. If you have a follow-up, hold it until she has finished.
Let the silence sit. The three seconds after a good question feel much longer to the person who asked it. Filling them with “or, you know, whatever comes to mind” converts a sharp question into a soft one. This is the single hardest thing on this page and it is worth practising deliberately.
Write the answer down. Visibly. Two lines, in a notebook. It signals the answer mattered, it slows the conversation to your pace, and you will genuinely want it later when you are deciding whether to accept.
Ask three to five, not fifteen
The set below is a menu, not a script. One early, two in the middle, one or two at the close. A candidate who works through a list is doing an exercise; a candidate who asks four questions and listens hard is having a conversation, and only one of those is memorable.
Questions are not interchangeable. Early ones change what you emphasise for the rest of the hour, so they are close to free. Late ones are for the decision you have to make afterwards, and asking them early reads as presumptuous.
| When | What it is for | How many |
|---|---|---|
| First ten minutes | Aiming everything you say afterwards | One |
| Middle | Getting her to describe the actual job in her words | Two |
| Last ten minutes | Finding out whether the seat is real, and shaping it | One or two |
These cost you almost nothing and change every subsequent answer. Ask one, in the first ten minutes, ideally before you have answered anything substantial.
Why it works. It changes the register from interrogation to conversation, and it is a question with no wrong answer, so it costs her nothing to answer honestly.
What her answer tells you. Mostly greenfield means they want someone who can define from nothing, and your discovery and prototyping stories should lead. Mostly fixing means the constraint is legacy and politics, and your stories about inherited process and stakeholders who already have opinions matter more.
Why it works. It is neutral and factual, so it does not read as a critique. And it is the highest-yield question on this page, because her answer describes the exact gap you are proposing to fill, in her own words, before you have claimed anything.
What her answer tells you. If she describes a rich artefact — acceptance criteria, field definitions, agreed edge cases — then the gap you are claiming is already closed, and you should soften the claim and ask what the real gap is instead. If she describes a process map and a deck, and the build team then spends two weeks asking questions, your entire pitch is now something she has just described herself. All you have to do is name it.
Follow-up if she is vague: “What's the first thing they usually come back and ask about?”
Why it works. People defend their failures and explain their successes. Asking what went right gets a franker, more detailed answer than asking what went wrong — and the ingredients she names are, implicitly, her theory of how this should work.
What her answer tells you. Whatever she names is what she will look for in you. If it is “we had the definitions nailed before we started”, you are in luck. If it is “the stakeholder was engaged the whole way”, she is telling you the relationship is the scarce resource, which sharpens why you should concede it.
Why it works. Still the best single question available. It gets her describing the job rather than evaluating you, and the second half quietly asks her to define success in her own terms.
What her answer tells you. This is your brief. Whatever she says here should be echoed, deliberately, in at least one answer afterwards — not parroted, but reflected. It is also the answer to put in your follow-up email.
Why it works. Surgical, and almost nobody asks it. It elicits exactly the load you are offering to take — but she names it, which means when you describe your scope five minutes later you are offering to solve a problem she just raised, rather than proposing a role she has to evaluate.
What her answer tells you. If she names something in your wheelhouse — writing specs, chasing definitions, building mock-ups to settle arguments — your position answer can be restated as literally that, and the round is effectively over. If she names something you cannot help with, that is worth knowing too.
Why it works. It demonstrates domain fluency without any claim attached, and it sets up your data-definitions ground without you having to argue for it. It also happens to be the question underneath all three named projects.
What her answer tells you. A clear answer means governance exists and your job is to work within it. A pause, or “it depends who shouts loudest”, means definitional authority is unowned — which is the single most available piece of ground in the whole org, and the strongest possible argument for the seat you are describing.
Why it works. It sizes the pain in her currency — her team's time — and it invites her to complain slightly, which builds more rapport than any amount of agreement.
What her answer tells you. A big number means the role's value can be argued in capacity terms, which is the easiest business case to make. A small number means the pain is elsewhere and you should stop pushing on this.
These are more pointed. They are fair questions and they are also slightly uncomfortable, so spend them only once the conversation has warmed up.
Why it works. The best test of whether the seat is real or aspirational, and asking it reads as exactly the judgment they are hiring for.
What her answer tells you. A name and a specific thing they would stop doing means the role is real and funded against actual pain. “Nobody really — it's all a bit spread out” means you would be defining the job as well as doing it. That is survivable, and occasionally excellent, but you should know it before you accept rather than in month three.
Why it works. If there was someone, you learn the failure mode from the inside and can address it directly. If there wasn't, her answer to that is informative on its own.
What her answer tells you. Listen for whether she describes the difficulty as personal — “they weren't quite senior enough” — or structural — “we never agreed what they owned”. The second is honest and workable. The first, especially if it is slightly dismissive, tells you how the debrief on you might sound if this goes wrong.
Careful: keep the tone genuinely curious. Delivered flatly it can sound like you are hunting for a reason to be worried.
Why it works. Very few candidates ask anything like this, and it invites her into the internal debate rather than the job description. It also assumes competence on her part — that a shape was chosen deliberately — which makes it flattering to answer.
What her answer tells you. The rejected version is where the real disagreement lives. If she says “we thought about making it more technical and decided against”, that confirms the re-scope came from a considered place. If she says “we've been going back and forth on whether it sits with us or with them”, you have just been handed the state of the negotiation you are in the middle of.
Why it works. This is the highest-value question on the page for your specific situation, because placement is undecided and the boundary is the thing that decides whether this job works. It is polite, it is answerable, and it surfaces a disagreement that would otherwise reach you in month four.
What her answer tells you. “Yes, we've talked about it” — ask what the sentence is, and listen for whether it matches yours. A laugh, or “probably not, no” — that is not a bad sign, it is honesty, and it makes your request that the scope be written down at offer stage entirely reasonable rather than defensive.
Delivery matters. Curious, faintly amused, not lawyerly. It is a question about the org, not about your contract.
Why it works. Better than “what does success look like”, which invites a recital of objectives. This asks for her judgment, and judgment is more revealing and more memorable than a metric list.
What her answer tells you. Whether she thinks in outcomes or in activity. “Fewer things get rebuilt” is an outcome. “There's good documentation” is activity, and it tells you the role may be measured on artefacts rather than on what they cause.
These prove fluency in a way nothing else on this page does, because only someone inside the problem asks them. Use one, and only if the conversation has already gone deep on that project. Two reads as a demonstration.
Each is on the learn page with the reasoning behind it, so you can hold the follow-up conversation rather than just asking the question.
“Is there anything about my background that gives you pause? I'd rather answer it now than have it be the thing nobody says out loud.”
The direct version is in your pocket below, and it is high-risk. These get most of the same information at a fraction of the cost, and none of them require her to say something unkind to your face.
One thing worth being honest about
A question engineered to make you sound good usually does not. Interviewers at this level can hear it, and it lands as performance. The questions that work do it as a side effect — they are genuinely useful to her, and sounding good is what happens on the way. Anything that would only flatter you has been left off this page deliberately.
Why it works. The best of the set. It asks about the ideal candidate rather than about your flaws, so she can answer completely honestly without either of you being rude. She describes the target; the gap becomes self-evident and you address it as a comparison rather than a defence.
What her answer tells you. If the profile she describes is more technical than the re-scope suggested, the role may still be contested internally. If it is more senior, expect a levelling conversation. If it is essentially you, stop talking and let it sit.
Why it works. Her answer is her doubt, restated as a ramp-up timeline rather than as a disqualifier. It also quietly presupposes you are in the seat, which does a second job. It is close to impossible for this one to go badly.
What her answer tells you. Whatever she names is the thing to address in your follow-up email, with a specific plan rather than reassurance.
Why it works. It offers to do more, and whatever she asks for reveals the residual doubt. Rarely used outside sales, and it belongs here.
What her answer tells you. “Could you speak to so-and-so” is a buying signal, not a hurdle. “No, I think we've got what we need” is neutral — do not read it as either good or bad.
Why it works. Fully indirect — it is about past hires, not about you. Her answer is the failure pattern she is watching for, which tells you what to reassure without her having to say it is about you.
What her answer tells you. Listen for whether the difficulty she names is structural or personal. Structural is honest and workable; personal, especially if slightly dismissive, tells you how a debrief on you might sound.
If she names a doubt — three sentences, then stop
Acknowledge it as fair. Give one concrete counter-example with something specific in it. Say what you would do about whatever remains. Then stop, and do not return to it later in the conversation. The length of your answer is what signals whether the concern was warranted — a short answer says minor, a long one confirms it.
Not by being clever. By asking things only someone who has actually lived this problem would think of.
Why it works. Estimate variance is the best available proxy for requirement quality, and nobody else in the process will ask it. It sets up your entire pitch without you claiming anything.
What her answer tells you. “A range, usually a wide one” is the gap you are describing, confirmed by her, unprompted.
Why it works. Most organisations have no feedback loop at all. It also signals that you consider adoption your problem, which is rarer than it should be.
What her answer tells you. “Someone mentions it eventually” means there is no measurement culture — which is a risk to you, because your work will be judged on impressions.
Why it works. Name nothing; let her pick. It is domain-fluent, and it invites her to describe the exact ground you are claiming.
What her answer tells you. If she has one ready, definitional authority is unowned and your seat has an obvious first job.
Why it works. Shows you think about the team's standing rather than only your own scope. Managers notice that, and it is the kind of thing that gets repeated in a debrief.
The mechanism is presupposition — a question asked from inside the role makes her simulate you in it. That is not a trick, as long as the question is one you actually want answered.
Why it works. The second half is the important part. It invites her to define your boundary favourably, in her own words, which means the scoping becomes her framing rather than your claim.
What her answer tells you. Whatever she says to leave alone is the thing she is protecting. Take it seriously and say so.
Why it works. Concrete enough that a vague answer is itself informative.
What her answer tells you. A ready answer means the seat is real and she has just imagined you doing it. No answer is the most useful negative signal available on this page.
Why it works. Frames you as relief rather than as headcount. Her answer is effectively the job description you want, in her words.
Why it works. Small, practical, forward-leaning. It turns an abstract role into people and a calendar, which is what makes a hire feel real.
Don't
“What's the culture like?”
Unanswerable, and every candidate asks it. You get a description of the culture the company would like to have.
Instead
“What's been tried here already that didn't work?”
Same information, arrived at through a specific. It also protects you from proposing something that failed two years ago.
Don't
Anything answerable from the job posting, the website, or the last earnings release.
It spends a question and signals you did not prepare. There are only three or four slots.
Instead
Use the public fact as the set-up, and ask about the consequence — “with the acquisition closing, does the integration land on the same three projects, or push them?”
Shows the homework and asks something only she can answer.
Don't
Comp, band, level, notice period, holiday.
All of it goes to the recruiter, and raising it here changes what the conversation is about.
Instead
Ask the recruiter separately whether the req and levelling were updated when the scope was re-cut.
It is a real question with real consequences for your review criteria, and it belongs in that channel.
Don't
The question that is really a statement — “how do you think about making sure requirements are testable, because in my experience…”
She will hear the pitch and answer the pitch. You have spent a question and learned nothing.
Instead
Ask the plain version and let your follow-up carry the point.
The information you get back is worth more than the impression you were trying to make.
Commit before the day. Five written down, in order, so that under pressure you are choosing from a short list rather than improvising. Saved in this browser only.
One early, two middle, one or two at the close. Write them in the words you would actually say — the phrasing above is a starting point, not a script.
Rule of thumb — if you only get to ask two, they should be the hand-off question and the one-sentence question. Everything else is upside.
The test of a good question here
Not whether it made you sound thoughtful. Whether her answer changed what you said next, or changed whether you want the job. Anything that fails both is decoration.