The End-of-Round Question Score, Six Dimensions to Ask, Hold, or Cut
Turn a loop schedule and a draft question list into a per-round card: two or three Asks tied to open offer gaps, a routed Hold list, and a Cut list.
Key takeaways
- Published interviewer guides give candidates between 5 minutes (Google re:Work) and 15 minutes (US State Department) for questions, so you only have room for about three.
- Practitioner guides that give both numbers have a median of about 7 questions prepared and 3 asked, so a little over half your list should end up as Hold or Cut in any one round.
- Score a question against an interviewer, not on its own: the same question can be an Ask for the hiring manager, a Hold for a peer, and a Cut by round four.
- In Refolk's index there are 2.04 US Engineering Managers per Technical Recruiter, so spend the recruiter screen on what only the recruiter owns: the loop, the timeline, why the role is open, and how the decision gets made.
- In a sample of 25 US Technical Recruiters from Refolk's index, 5 worked at staffing or consulting firms, so check the recruiter's employer before asking them about the team.
- The 'any hesitations about me' closer only makes sense with someone on the decision path in a late round, and no source recommends it in a recruiter screen.
This guide is a scoring model for the end of each interview round, when the interviewer asks "do you have any questions for me?" It is for candidates who have a loop schedule and a draft list of questions and need to decide which questions to spend on this interviewer, which to save for a named later round, and which to drop. For each round you end up with a card: two or three Ask questions, each tied to an open gap in your own offer criteria, a Hold list routed to specific later rounds, and a Cut list.
What does the end-of-round question score decide?
For one interviewer at a time, the score decides whether each question gets asked in this round, held for a named later round, or cut. Instead of working from a long list, you route each question by who can answer it, what is already public, and what you still need to know before you can decide on an offer.
The pages that rank for "questions to ask at the end of an interview" are lists. Monster offers 30 questions, Indeed offers 35, and techinterview.org groups its questions by stage in a table. None of them scores a question against a specific interviewer. That leaves you guessing, and two things tend to happen. You ask the same question to four people, and you use a popular closer on someone who can't do anything with the answer.
The rest of the guide uses four terms:
- Loop: the full set of interview rounds for one role, from recruiter screen to final round.
- Offer criteria: your written list of what must be true, or must be known, before you would accept.
- Open gap: an offer criterion you cannot yet judge from what you already know.
- Ask, Hold, Cut: Ask means use it in this round. Hold means save it for a named later round and interviewer. Cut means drop it for the whole loop.
You score a question and an interviewer together, never a question on its own. "Why is this role open?" can be an Ask for the hiring manager, a Hold if the recruiter has already promised to cover it, and a Cut by round four because two people have already answered it.
How much time do you get for questions, and how many fit?
Published interviewer guides give candidates 5 to 15 minutes for questions, and practitioner guides settle on preparing about 7 questions and asking about 3. That leaves roughly 2 to 5 minutes per question. You can't weigh six dimensions in that time, so do the scoring before the round.
| Source | Time for questions | Interview length | Share of interview |
|---|---|---|---|
| US State Dept | 15 min | 45 to 50 min | 30 to 33% |
| Google re:Work | at least 5 min | 45 min | about 11% |
| Coursera (recruiter screen) | at least 5 min | 30 min | about 17% |
| OPM | "opportunity", no minutes | n/a | n/a |
The shares are my own calculation from each source's stated minutes. The State Department tells panels to give the candidate 15 minutes at the end. Google re:Work tells interviewers to build in at least five, and its 45-minute template spends 32 minutes on interview questions after a 5-minute introduction. Most of your rounds will look more like Google's template than the State Department's.
| Source | Prepare | Ask |
|---|---|---|
| Monster | about 10 | 2 to 3 |
| LoopCV | 5 to 7 | 2 to 4 |
| CareerTestPrep | 7 to 8 | 3 to 5 |
| CrackInterviewAI | 5 to 8 | 2 to 3 |
| Reed | not stated | 2 to 4 |
Across the sources that give both numbers, the median is to prepare about 7 and ask about 3, a ratio of roughly 2.3 to 1. In practice that means a little over half your list will be Hold or Cut in any one round. That is the system doing its job.
Is the question segment scored?
That is not established publicly. OPM's structured interview guide puts candidate questions after the rated questions and says the selecting official's interview "is unlikely to be rated". OPM's federal-jobs slides mark hiring-manager interview notes "NOT scored". I found no official guide that lists candidate questions as a rubric item. My working assumption is that the segment earns you no formal points. What you say in it can still reach the people who decide, though, because at companies like Google interviewers write feedback that goes to the person making the hiring decision.
Who in the loop can answer what?
An interviewer can only close a gap that falls within their own job. Recruiters own the pipeline and the offer, hiring managers own why the role exists and what the team does, and interviewers brought in from outside the team can't speak for that team's work.
| Loop role | Can answer accurately | Route elsewhere |
|---|---|---|
| In-house recruiter | Loop, timeline, how decisions are made, comp structure | Detailed team and role questions |
| Agency recruiter | Pipeline and timeline | Team, roadmap, why the role is open |
| Hiring manager | Why the role is open, the team, what success looks like | Process logistics already covered |
| Outside-team interviewer (e.g. Amazon Bar Raiser) | How the company hires | The team's roadmap and day-to-day work |
The sources are practitioner sources. Techinterview.org says the recruiter "owns the pipeline and the offer, not the engineering". Another practitioner warns against asking recruiters for detailed role information "because most recruiters simply won't know". Holloway's Ask Me This Instead tells candidates to ask both the recruiter and the hiring manager why the position is open. Amazon's Bar Raiser is outside the team by design: "they aren't the hiring manager and they don't work on the role you're interviewing for." I found no published interviewer-training material that maps every loop role to what it can answer, so use this table as a starting point and correct it with what the recruiter tells you about the loop.
In Refolk's index of professional profiles, there are 43,108 people titled Engineering Manager in the United States against 21,169 titled Technical Recruiter. Recruiters are thin on the ground compared with the managers they hire for, and each candidate gets one short screen with them. Use that slot for the loop, the timeline, why the role is open, and how decisions are made. Don't spend it on the team's tech stack, which a manager can answer better later.
| Segment (Refolk's index) | Count | What it tells you |
|---|---|---|
| Technical Recruiter, US | 21,169 | Baseline |
| Technical Recruiter, UK | 703 | 30.1x fewer than US, likely index coverage, not market size |
| Engineering Manager, US | 43,108 | 2.04 per US Technical Recruiter |
| Title purity, 25-person samples | 84% / 80% / 88% | Titles mostly match the real job, not always |
Read the UK row with care. The gap between US and UK counts most likely reflects how much of each country the index covers, not the real size of either market. The title purity row is the useful one for this guide. Across those three samples, between 12% and 20% of people did not do the job their title suggested. A job title on a calendar invite is a clue, not proof that the person owns the answer you need.
In that sample, three of the agency recruiters were at K2 Partnering Solutions and two were at Experis. If your recruiter works for an agency, they are one step further from the team, and team questions move to the hiring manager.
The six dimensions and how to score each
Give each question a 0, 1 or 2 on six dimensions, once for each interviewer. The first five decide whether the question qualifies. The sixth, follow-up hook, only breaks ties, because the evidence behind it is the weakest.
| Dimension | 2 | 1 | 0 |
|---|---|---|---|
| Decision gap (G) | Could change accept or counter | Shapes prep, not the decision | Curiosity |
| Not public (P) | Not findable before the round | Partly public; ask the rest | In the posting or site |
| Owner fit (O) | This role owns the answer | Knows it secondhand | Cannot know it |
| Routing (R) | Best or only round for it | Another round fits equally well | A later round fits better, or asked twice already |
| Signal (S) | Shows you prepared for this role | Neutral | Shows you didn't read, or raises doubt they can't resolve |
| Follow-up hook (F) | Builds on something they said | Could if the talk goes there | Canned switch |
What each dimension proves, and how it lies
- Decision gap shows that the question is worth the minutes. It misleads when you invent the gap after writing the question. If the criterion was not on your written list before you drafted the question, give it a 0. If you have already tailored your application in Refolk, check the fit score against the posting. The requirements where you match weakly are gaps the hiring manager is likely weighing too, and they make better questions than anything off a generic list.
- Not public shows the answer needs a person. It misleads when you mark something "not public" because you didn't look. Indeed advises against asking things you could find on the company's website. Do the check in step 3 before you score this.
- Owner fit shows this interviewer can give you an accurate answer. It misleads when someone answers confidently anyway. A recruiter can describe a team's roadmap in a way that sounds sure and is secondhand. If the answer matters, ask the owner too.
- Routing shows this is the right round to spend the question in. It misleads when a question is relevant to every role, so you ask it everywhere. The exception is a deliberate cross-check, covered under failure modes.
- Signal shows what the question says about you to this person. It misleads because the same words read differently by stage. Virginia Tech career services says pay, benefits and vacation questions "should be saved for when you are formally offered the job", but in an EU jurisdiction the range may already be in the posting, so asking for it only shows you didn't read it.
- Follow-up hook shows the question responds to the conversation. It misleads when you add a "you mentioned" bridge to a canned question. That doesn't turn it into a follow-up.
Reading the score: Ask, Hold, or Cut
Run the gates in order, and stop at the first one a question fails. Questions that pass every gate are Asks, ranked by their total. Take the top two or three, and use the follow-up hook to break ties.
The gate order for one question and one interviewer
- GapDecision gap is 0, so Cut it for the whole loop
- PublicNot public is 0, so Cut it, or rewrite it to ask only for the part that isn't public
- OwnerOwner fit is 0, so Hold it and route it to the role that owns the answer
- SignalSignal is 0 for this person, so Hold it for someone it is safe with, or Cut it if nobody qualifies
- RouteRouting is 0, so Hold it for the named better round
- RankAdd up G, P, O, R and S, and Ask the top two or three, with F breaking ties
These are the decision rules I use. They are a convention, not a validated instrument, so adjust them to fit your loop:
- Ask if no dimension scores 0, decision gap is 2, and G + P + O + R + S is 8 or more out of 10.
- Hold if the question fails only on owner fit, signal or routing for this interviewer, and some named later round scores it higher. A Hold with no named round is a Cut you haven't admitted to yet.
- Cut if decision gap or not public is 0, or if signal is 0 for every interviewer in the loop.
- Cap each round at three Asks. If more than three qualify, the extras become Holds for the next round that can answer them, or fallbacks if this interviewer covers your top three during the round.
Gap size against answer ownership
Write the result as a card you can glance at in the room. Fill it in the night before, one card per round.
ROUND [n] of [total]: [interviewer name], [role], [in-house or agency / on or off decision path] OPEN GAPS THIS ROUND CAN CLOSE: - [offer criterion] - [offer criterion] ASK (2 to 3, in order) 1. [question] Closes: [criterion] Score: G_ P_ O_ R_ S_ F_ 2. [question] Closes: [criterion] Score: G_ P_ O_ R_ S_ F_ 3. [question] Closes: [criterion] Score: G_ P_ O_ R_ S_ F_ HOLD - [question] -> Round [n], [role]. Reason: [owner / signal / routing] CUT - [question]. Reason: [no gap / in posting / unsafe for everyone] IF ALL ANSWERED: "You covered most of what I planned to ask, including [topic]. Building on what you said about [their point], ..." AFTER THE ROUND (answers, specifics, contradictions): -
Make one card per round. Fill every bracket before the round. If the Hold section has no named round, move the item to Cut.
The procedure, from loop map to logged answers
Seven steps get you from an empty schedule to a scored card for every round. Steps 1 to 4 happen once per loop, step 5 once per round, and steps 6 and 7 in and after every round. The durations below are my estimates, not sourced figures.
Build and run the per-round question cards
- Get the loop map from the recruiterIn the recruiter screen (about 5 minutes), ask who is in each round and how the final decision is made. Done when every round has a role label.
- Write offer criteria and mark open gapsOnce per search (about 30 minutes), list what must be true before you would accept, and mark each item you cannot yet judge. Done when you have a written list of unknowns that affect your decision.
- Pre-check what is already publicCheck the posting's pay range, the careers site and public filings (about 30 minutes). Done when each gap is marked public or needs a person.
- Draft questions and tag the answer ownerWrite one question per open gap and tag it with the loop role that owns the answer. Done when no question is tagged anyone.
- Score each question against each interviewerRun the six dimensions and the gate order for every question and interviewer. Done when every question is Ask, Hold with a named round, or Cut.
- Cross off and convert to follow-ups in the roundCross off anything already answered and tie what is left to something the interviewer said. Done when you have asked two or three questions.
- Log answers and re-route the Hold listStraight after the round, log the specific answers, compare them with earlier rounds, and move open Holds forward. Done when the next round's card is updated.
Step 1 depends on the recruiter, and Google re:Work says recruiters should give candidates "an overview of how hiring decisions are made". Ask for that directly, and ask why the role is open in the same breath. Holloway recommends asking that early, to both the recruiter and the hiring manager.
Before we wrap up, could you walk me through the rest of the process? 1. Who will I meet in each round, and what is each person's role relative to this team? 2. How is the final decision made: does one person decide, or does each interviewer's feedback go to a group or committee? 3. Is this role a backfill or new headcount, and what prompted opening it now?
Use this as one of your recruiter-screen Asks. Adjust the wording, but keep all three parts.
For step 3, the pay check depends on where the job is. Directive (EU) 2023/970 requires employers to give "the pay range or starting salary in the job vacancy notice or before the job interview, without the applicant having to request it." Ireland and the Netherlands require the range to appear in the advert itself. Member states had until 7 June 2026 to transpose the directive, and published counts of how many met that deadline conflict (one source says 4 of 27, another says 5). Don't assume the rule applies. Check the posting and the country's own implementing law, then score not public accordingly. Asking for the range where it has already been disclosed is a Cut. Asking the recruiter about the full compensation structure is still a valid question.
Sources disagree on whether pay belongs in the recruiter screen at all. One practitioner lists "What is the compensation band for this role?" as a recruiter-screen question. Virginia Tech says to wait for the offer. Where the posting is silent and no disclosure law applies, I score a pay question S1 with the recruiter and S0 with everyone else. That makes the recruiter the only place it can be an Ask.
Step 4 goes faster if your application materials are already tied to the posting. Refolk tailors the resume to each posting you apply to, so the posting text and your weak-match requirements are already in one place when you tag questions to owners.
When does "do you have any concerns about my candidacy" help?
This closer only helps with someone on the decision path, late in the loop. With anyone else, expect it to hurt you. Score its signal at 0 for every interviewer who only submits feedback, and that includes every recruiter screen.
The case against it is about how decisions get made. The recruiter Farah Sharghi, who estimates she has run more than 10,000 interviews at companies including Google, Lyft and TikTok, calls it "the worst advice". At Google, she explains, candidates interviewed with members of a hiring committee, who then submitted feedback to the person making the decision. If you prompt a doubt in someone who only writes feedback, it goes into the feedback, and you never get to rebut it. Amazon works differently but carries the same risk. The Bar Raiser decides jointly with the hiring manager, and a former Bar Raiser who ran 813 interviews describes the debrief as a 30-minute meeting that you are not in.
The case for it comes from decision-makers. One commenter on LinkedIn says they would give the candidate "at least one thing to see how they handle the objection." That only helps you if the person raising the objection is the one who gets to drop it. Askcruit says to "skip it in early screening calls with recruiters who aren't making the final decision". No source I found recommends it with recruiters.
If you keep it for a decision-maker, word it as a request for something you can act on in the round, for example: "Is there any part of the role where you'd want more evidence from me before you decide?" Leave yourself time to answer. A closer asked with thirty seconds left gives you no chance to respond to what they say.
How the pick goes wrong
Most failures pass the gates on paper and fail in the room. Each one below includes the check that catches it before you're in the room.
- The hesitations closer asked of someone who doesn't decide. The interviewer may only send feedback to a committee, so you raise a doubt that nobody in the room can resolve. Check: the loop map from step 1. Is this person on the decision path?
- The same question asked in four rounds. Techinterview.org calls asking everyone the same three questions "the single most common mistake". False positive: a question that seems relevant to every role, so you never route it. Check: is the Routing score blank, or is it 2 for every interviewer? If so, you haven't routed it.
- Asking for the pay range where it is already disclosed. In EU jurisdictions that have transposed the directive, the range comes before the first interview, so asking for it only shows you didn't read the posting. Check: step 3. A full comp-structure question for the recruiter is still valid.
- Asking an agency recruiter about the team. In Refolk's US sample, 5 of 25 Technical Recruiters sit at staffing firms. Check: the recruiter's employer, before any team question is routed to them.
- Asking an outside-team interviewer about the roadmap. A Bar Raiser or any interviewer from another team is there by design to be independent of the role. Check: owner fit. Send it back to the hiring manager.
- A "follow-up" that is really a switch. The liking effect in the research comes from real follow-ups to what the person just said. False positive: a canned question with "you mentioned earlier" stuck on the front. Check: could you have written the question before the round? If yes, it is a switch.
- Reading consistent answers as confirmation. Several interviewers giving the same scripted answer can hide real disagreement. Check: ask the same factual gap of two different roles on purpose and compare the details, not the tone. Mark these on the card as cross-checks so they don't count as routing failures.
The research is Huang, Yeomans, Brooks, Minson and Gino, "It Doesn't Hurt to Ask", in JPSP 113, no. 3 (September 2017). The authors trained a language-processing model to detect follow-up questions. They found that people do not expect asking questions to make them more liked. In Study 1, people who asked many questions were rated 6.02 against 5.79 for those who asked few. A correction has been published, and a critique using social-relations-model reanalyses found no follow-up effect. The authors replied in 2019. Take one thing from it: responding to what the person said is more likely to help than reading out another prepared question. Don't count it as proof that follow-ups win offers.
A question earns its minutes only if the answer could change what you sign.
Keeping the card current through the loop
The card changes after every round. Answers close gaps, Holds move forward, and contradictions between interviewers become your most valuable questions for the final round. Spend about ten minutes after each round updating it, before details fade.
Log the specific answer, not your impression of it. "Backfill, previous person moved to the platform team" is useful. "Seemed stable" is not. Then compare answers across rounds. Holloway treats it as a warning sign when different interviewers share contrary information, and a contradiction about why the role is open or how success is measured goes straight onto the hiring manager's card as an Ask.
When an interviewer answers all your Asks before you get to them, don't go to the Cut list. Monster suggests building "on something they mentioned during the conversation". Indeed suggests telling the interviewer you had planned to ask about those topics and have already been given the answers. Both work. Saying you have no questions does not.
Before you walk into each round
- The interviewer's role and position on the decision path are written on the card.
- Every Ask names the offer criterion it closes.
- Each round's card has no more than three Asks.
- No Ask can be answered from the posting, careers site or filings.
- Every Hold names a specific later round and role.
- The hesitations closer only appears on a decision-maker's card, if at all.
- No team question is routed to an agency recruiter or an outside-team interviewer.
- Any deliberate cross-check is marked as one, and only one other round asks it.
- The fallback line is filled in with a real topic from your list.
- Answers from earlier rounds are logged, with any contradictions moved to a later card.
When the loop ends, every criterion on your written list should be closed, or marked as something you will raise at the offer stage. Whatever is still open becomes your agenda for the offer conversation.
Questions job seekers ask
How many questions should I ask at the end of an interview?
Ask two or three per round and prepare more than that. Monster suggests preparing about 10 and asking two or three. Across practitioner guides that give both numbers, the median is about 7 prepared and 3 asked. Preparing extra matters because the interviewer will answer some of your questions during the round, and the rest of your list needs to be routed to other people in the loop, not asked again here.
What should I ask the recruiter vs the hiring manager?
Ask the recruiter about the process and the offer, and ask the hiring manager about the role and the team. The recruiter owns the pipeline: the loop, the timeline, how decisions are made, compensation structure, and often why the role is open. The hiring manager owns why the role exists, what the team is doing, and what success looks like. Practitioners warn that most recruiters simply won't know detailed role information, and agency recruiters are even further from the team.
Should I ask "do you have any concerns about my candidacy"?
Only ask it of someone on the decision path, in a late round. One recruiter with more than 10,000 interviews behind her calls it the worst advice, because an interviewer who only submits feedback writes your prompted doubt into that feedback, and you never get to rebut it. Askcruit says to skip it in early recruiter screens. No source recommends it with recruiters.
What do I do if the interviewer already answered all my questions?
Say that the conversation covered what you planned to ask, then build a question on something they mentioned. Indeed suggests telling the interviewer you have already been given the answers. Monster suggests building on something they mentioned during the conversation. A real follow-up on what they said is better than a backup question pulled from a list, and it beats ending with nothing to ask.
Is it okay to ask about salary in the first interview?
It depends on where the job is and what the posting already says. Under the EU Pay Transparency Directive, employers must give the pay range before the interview, so asking for the range only shows you didn't read the posting. Asking a recruiter about the full compensation structure is still fair. Sources disagree on timing elsewhere. Some practitioners ask for the band in the recruiter screen, while Virginia Tech career services says to wait for the offer.
Do interviewers score the questions I ask at the end?
That is not established publicly. I found no official interviewer guide that lists candidate questions as a rubric item. OPM places them after the rated questions and says the selecting official's interview is unlikely to be rated. Formal points may not be at stake, but what you say can still end up in written feedback that reaches the person making the decision.
Put this to work
Paste your career in once. Every application after that is written for you.
Drop a resume or a LinkedIn URL. I rank the live openings against it, rewrite the resume and write a cover letter for the best of them, and fill in the employer's form when you press the button. You read, you decide what goes out.
01Drop your resume
A PDF or a LinkedIn URL. About a minute, once.
02I rank the openings
Every weekday morning, the live catalog scored against your history. Up to 20 worth your time, not two hundred links.
03Each one is written up
Resume rewritten for the posting, a cover letter, a fit score. Press send, or let me fill in the form.
- New matches ranked and written before you are up.
- Every bullet stays inside what your history supports. Nothing invented.
- Queued, submitted, interviewing, offer: one screen, not a spreadsheet.
Free, no card. Up to 25 applications sent a week, and nothing is sent until you say so.