How to Train Hiring Managers to Interview

The worst interview I have ever run was one of mine, about four hires into the company. I talked for thirty-one minutes of a forty-five minute slot. I know the number because I counted afterward, on a recording I had taken for a different reason, and because the candidate told a mutual contact that she had not been asked a single question about her actual work.

The second worst was run by somebody I had promoted the month before. He asked a candidate whether she was planning to have children, because he wanted to know whether she would still be there in two years. He was not being malicious. Nobody had ever told him that the question was off limits, and in a company with no HR department there was nobody else in the room to stop it.

Both of those are training failures, not character failures, and they are the same failure: somebody was handed responsibility for a hire and given nothing to run it with. This piece is what I would teach that person now, in what order, and how to check that it landed. I build the people and records tooling for companies without an HR department at FirstHR, and I have watched this go wrong in a lot of small teams. This is general information rather than legal advice.

What a First-Time Interviewer Actually Needs to Know

A first-time interviewer needs four things: the short list of questions that create legal exposure, a fixed set of questions with a written scoring scale, the discipline to record what the candidate said rather than how they came across, and awareness of the three habits that reliably produce bad hires. Everything else can wait until the second or third search.

That list is much shorter than what interview training usually contains, and the reason is arithmetic. A manager at a small company might run six interviews this year and then none for eight months. Anything taught as theory decays completely in that gap. Anything taught as an artifact they can hold, a printed question set with a scale on the back, survives because the artifact is still on the desk when the next search starts.

The last row is the one that surprises people. In most interview training, notes get a passing mention. In the small companies I have seen get into trouble, the note is the exhibit. The question that was asked is usually defensible or at worst ambiguous, and then somebody wrote three words in the margin that were not.

The other thing worth saying out loud before you teach anybody: be clear about who owns the decision. If the founder is going to overrule the hiring manager anyway, training the hiring manager is theater. The split of authority between founder, hiring manager, and anybody else in the loop is worth settling first.

The Legal Minimum They Must Not Get Wrong

There are three things a new interviewer must not get wrong: asking questions that reveal a protected characteristic, asking different candidates different questions for the same job, and writing notes that record impressions instead of evidence. Almost every interview problem a small company has traces back to one of those three.

Start with the categories. Federal law prohibits employment discrimination on the basis of race, color, religion, sex including pregnancy, sexual orientation and gender identity, national origin, age from 40 up, disability, and genetic information including family medical history (EEOC). Coverage thresholds differ by statute: Title VII and the Americans with Disabilities Act generally start at 15 employees, the Age Discrimination in Employment Act at 20, and many state laws start lower or at one employee.

Then teach the mechanism, because a list without a mechanism does not stick. Asking the question is rarely the offense by itself. The damage is that the question puts the answer in your possession, so when the candidate is rejected there is now a direct line between what you learned and what you decided. The employer is left arguing about a state of mind. That is a bad position and an expensive one.

The questions that actually get asked almost never sound like discrimination. They sound like friendliness. Are those your kids in the photo. Where is that accent from originally. What year did you graduate. Is your husband local as well. Do you have a church around here. Every one of those arrives during the warm-up, from a nervous interviewer trying to make a stranger comfortable, and every one of them puts a protected characteristic on the table.

State law layers on top, and the layers move. Many states and cities prohibit asking about salary history, several restrict when criminal history can be raised, and a growing set require pay ranges to be disclosed in the posting or on request. Those rules are state by state and change often enough that they belong on a sheet you refresh each year rather than in somebody’s memory.

One rule carries most of the weight for a beginner, and it is worth teaching before the list. If the answer would tell you something about the person rather than something about doing the job, do not ask it. Then give them the list anyway, because the rule catches ninety percent and the list catches the rest.

Consistency Across Candidates Is the Part Small Companies Lose First

Consistency means every candidate for the same job gets the same core questions, in the same order, scored on the same scale. It is the cheapest legal protection available to a company with no HR department, and it is also the only way two interviewers can produce results that are comparable to each other.

The legal reason is worth explaining to the person you are training, because it makes the discipline feel less arbitrary. An interview is a selection procedure. Under the Uniform Guidelines on Employee Selection Procedures, any selection procedure that has an adverse impact on a protected group is treated as discriminatory unless it is validated or otherwise justified, and where two equally valid procedures exist the employer is expected to use the one with less adverse impact (29 CFR 1607.3). An unstructured conversation cannot be validated, because there is nothing stable to validate. A fixed question set can.

The practical reason is that inconsistency destroys your ability to compare. If candidate A was asked about a failed project and candidate B was asked about their proudest achievement, you have two impressions and zero comparison. Small companies lose this fast, because the fourth interview of the week runs long, the interviewer is tired, and they cut two questions from the middle without telling anyone.

The confidence gap in that callout is the thing to name in training. A new interviewer will feel that the free-flowing conversation went better, because it did feel better. Feeling and accuracy point in opposite directions here, which is why the process has to be a rule rather than a preference.

Two operational rules make consistency survive contact with a busy week. First, the core questions are fixed and the follow-up probes are free, so the interviewer still gets to be curious without losing comparability. Second, if a question gets cut for one candidate, it is cut for everyone, decided before the next interview rather than after.

Note-Taking That Will Not Embarrass You Later

Teach one rule: write what the candidate said and did, not what you thought of them. Interview notes are records, they are discoverable, and the EEOC requires covered employers to keep personnel and employment records for at least one year from the date the record was made or the personnel action was taken, whichever is later (EEOC). If a charge is filed, everything connected to it must be kept until the charge or any resulting lawsuit is finally resolved.

New interviewers write two kinds of dangerous note without noticing. The first is the impression note, which records a conclusion with no evidence attached and often smuggles a protected characteristic in the process. The second is the shorthand note, where the interviewer compresses a real observation into two words that read completely differently a year later in a deposition binder.

Two more rules go with the table. Notes get written during the interview or within ten minutes of it, because a note written the next morning is memory rather than record. And every interviewer keeps notes for every candidate, not only the ones who were interesting, because a file with notes on three candidates and nothing on the other nine is its own kind of evidence.

Then decide where the notes live. Loose paper in a drawer and a personal notes app both fail the same test: you cannot produce them on request and you cannot prove what was there. One folder per requisition, with the scorecards and the notes together, is enough.

How to Run a Calibration Session

A calibration session is a 45-minute meeting where everybody who will interview scores the same sample answers independently, reveals their scores at the same moment, and then discusses only the answers where the gap was large. Run one before the first interview of any search where more than one person will be in the loop.

The reason to run it is that a shared scale is not actually shared until it has been tested. Two people can both be looking at a four-point scale, both be acting in good faith, and be two points apart on the same answer, because one of them is scoring against the best person they ever worked with and the other is scoring against the current team average. Nobody discovers this until a real candidate is on the table and the debrief turns into an argument about a person instead of about a standard.

The three sample answers matter more than the format. One should be clearly strong, one clearly weak, and the third should sound impressive and contain nothing checkable. That third answer is the whole exercise. It is the fluent, confident, well-structured response that describes a strategy without ever saying what the person themselves did, and untrained interviewers score it high almost every time.

Where do the samples come from at a company that has never done this? Three places. Written answers from candidates in your last search, with names removed. Your own team answering the same question about a project they ran, which has the useful side effect of showing what a real strong answer sounds like in your business. Or answers you write yourself, deliberately, to hit the three levels.

Do not skip the anchor writing at minute thirty. The output of a calibration session is not agreement, it is text. If nothing gets written down, everybody leaves feeling aligned and the alignment is gone within a fortnight. Anchor text written in the words somebody actually said in the room gets recognized later; anchor text written in polished competency language does not.

Shadowing and Reverse-Shadowing Are the Cheapest Training You Have

Shadowing is the new interviewer sitting in on your interview and scoring it silently. Reverse-shadowing is you sitting in on theirs, saying almost nothing, and holding your feedback until it is over. Two cycles of each cost about four hours of time you were already spending, and they produce more skill than any course you can buy.

The rule that makes shadowing work is independent scoring. The shadow does not observe and reflect. They score, in writing, on the same sheet, and they hand it over before the two of you discuss anything. Then you compare question by question. Every gap of two points or more is a teaching moment with real content, and it is a much better conversation than asking a beginner what they noticed.

The rule that makes reverse-shadowing work is that you do not rescue. This is genuinely hard. The interview goes sideways at minute twelve, the candidate misunderstands a question, and every instinct says step in and fix it. If you do, the person you are training learns one thing: that they cannot be trusted to run an interview, and that you will always be there to take over. Let it be imperfect. You can ask your own two assigned questions at the end.

What if there is genuinely nobody to shadow, because you are three people and the founder is the only person who has ever interviewed? Two substitutes that hold up. Have the new interviewer interview one of your existing employees about a project that person actually ran, with you scoring alongside. Or use written transcripts and score them together, which loses the live pressure but keeps the calibration value.

Recording is the other option, and it is fine as long as you get written consent from the candidate first and you apply the practice to every candidate for that role rather than to a chosen few. Several states require all-party consent for recordings, so this is a rule to check for your state before you turn anything on.

How to Give a New Interviewer Feedback on Their Interviewing

Give feedback on four measurable things: what share of the interview they spoke, how many follow-up probes they asked, whether their score was written before the debrief started, and whether every score above the midpoint has a quoted example under it. Feedback on those four changes behavior. Feedback on style produces a defense and no change.

The difference is the same one that makes any performance conversation land or fail. You spoke for about half the interview is a fact that can be accepted without anybody losing face. You dominate conversations is a claim about who somebody is, and it will be argued with. The behavior-first structure here is the same one described in the SBI model, and it works for interviewing as well as it works anywhere else.

Timing matters more than most people expect. The feedback happens within an hour of the interview, while both of you can still quote what was said. A note in a document three days later is a review, not coaching, and the person has already run their next interview the same way.

Give it in a fixed order that puts the fixable things first. What went to plan, with one specific example. The four numbers. One habit to change before the next interview, one only. Then what they should keep doing, because a beginner who hears only corrections concludes that they are bad at this and starts reading from the script in a monotone, which is worse.

One thing to separate clearly: coaching your interviewer is not the same activity as writing feedback for the candidate. They use different language, go to different places, and carry different risk. New interviewers mix the two constantly and end up writing internal scorecards in the careful, hedged language of a rejection note, which makes the scorecard useless as a record.

The Habits That Produce Bad Hires

Four habits cause most bad hires by first-time interviewers: talking more than the candidate, deciding in the first few minutes and spending the rest confirming it, scoring similarity to yourself as if it were competence, and selling when you are supposed to be assessing. Each one has a mechanical fix that does not require anybody to be less biased by force of will.

The second habit is the one with the strongest research behind it, and it is worth quoting to the person you are training because it is so counterintuitive. Barrick, Swider and Stewart, publishing in the Journal of Applied Psychology in 2010, measured evaluations formed during the rapport-building phase before the real questions began. Those initial impressions correlated at .42 with end-of-interview ratings from the same interviewer and at .22 with actual job offers.

Read that again from the interviewer’s side. A judgment formed during the pleasantries, before a single substantive question, predicted a meaningful chunk of the final score. The interviewer experiences the remaining forty minutes as evidence gathering. It functions as confirmation, and no amount of good intent changes that, which is why the fix is a scoring rule rather than an instruction to be careful.

The third habit deserves one extra warning, because it hides behind a word that sounds like a virtue. Culture fit, used loosely, is where similarity scoring goes to look respectable. There is a defensible version, which is testing whether somebody works well under the way your company actually operates, and there is the version that means I would enjoy sitting next to this person.

A Two-Hour Curriculum You Can Actually Deliver

Six modules, 115 minutes, each producing something physical. Deliver it in one sitting in the week before the first real interview, not as a standing programme that never quite gets scheduled. The artifacts are the point: an interviewer who leaves with a question set, a scale, and a desk sheet has been trained, and one who leaves with notes from a talk has not.

What you leave out matters as much as what you put in. No general unconscious bias module, because it takes an hour and changes no observable behavior in the interview. No body language segment, because it teaches people to score first impressions, which is the exact habit you are trying to break. No employer brand deck, because selling belongs in a separate block. No tooling walkthrough on day one.

Module one is where people cut corners, and it is the module that carries the rest. If the role is written as a list of responsibilities rather than outcomes, the questions come out generic, the anchors cannot be written, and the calibration has nothing to calibrate against. If you are starting from a job posting rather than an outcomes page, rewrite it first using the structure in the guide to writing a job description.

After the two hours, the rest of the programme is the practice sequence: one shadowed interview, one reverse-shadow, one debrief on the four measurables, and one paperwork audit. That is the entire curriculum. Roughly six hours of your time per new interviewer, most of it during interviews you were already running, which is the reason I keep arguing that the no-training-function excuse does not hold.

Where this sits in the wider picture: interviewing is one skill inside a first-time manager’s job, and it usually arrives before any of the others get taught. If you are promoting people into management for the first time, it is worth folding this into the broader sequence covered in new manager onboarding.

How to Tell Whether the Training Worked

Check four things after the first search. Were scores written before the debrief started. Does every score above the midpoint have a quoted example under it. Did two interviewers land within one point of each other on most questions. And at 90 days, is the person you hired doing the job that was written on the outcomes page.

The first three are checkable the week after the search closes, and they are checkable from paper without anybody being asked how they felt about it. The fourth takes a quarter and is the only one that measures the actual outcome, which is why the first three exist. They are the leading indicators for a result you cannot see yet.

Each check has an obvious repair. If scores are not being submitted before the debrief, make submission the gate that opens the debrief invitation, so the meeting cannot start without them. If high scores arrive without quoted evidence, send the scorecard back once. In my experience that only has to happen a single time.

If two interviewers are landing two or three points apart at the end of a search, the anchors failed rather than the people. Rerun the 45-minute calibration on answers from that search and rewrite the anchor text for whichever questions produced the gaps. If the notes are still reading as opinion, sit down and rewrite two of them together, then re-audit the next search rather than assuming it took.

The 90-day check is the one that improves the process rather than the person. When a hire is not working out, the useful question is not whether the interviewer was bad at their job. It is which question should have surfaced this and did not, and whether the answer is a missing question, a missing follow-up probe, or an anchor that was set too generously.

Keep the loop small. One change per search, applied before the next one opens. A company that rewrites its entire interview process after every disappointing hire ends up with fourteen questions, three scales, and nobody willing to run any of it. The check on the hire itself belongs in the same conversation as the 30-60-90 day plan you should already be writing for the new person.

One last thing about ownership. Interview training decays in exactly the way that any process without an owner decays, so give it a name on the calendar. Whoever runs the next search delivers the two hours to whoever is joining the loop. That is the whole governance model a small company needs, and it survives longer than any document, including this one, sitting in a folder nobody opens.

Frequently Asked Questions