Common interviewer mistakes: the habits that skew hiring decisions, and the fix for each

The most common interviewer mistakes are deciding in the first few minutes and then looking for confirmation, leaving the agreed questions, asking leading or off-limits questions, talking more than listening, writing down conclusions instead of the candidate's words, letting one strong answer color every score, comparing each candidate with the previous one, and writing feedback days later. The U.S. Office of Personnel Management's structured interview guide lists most of these as common rating errors and interviewing mistakes, and each has a practical fix.

Below are twelve, grouped by when they happen, each with what it looks like, why it matters and the fix. Most are habits rather than character flaws, which is good news: habits respond to structure, practice and feedback.

Which interviewer mistakes happen most often?

Mistakes before the interview

1. Not knowing what the job needs

What it looks like: the interviewer reads the resume five minutes before and asks about whatever stands out. Why it matters: OPM lists "not knowing the job" as a common mistake: interviewers without a clear picture of the requirements form their own idea of the best candidate and judge against that. Fix: ten minutes with the scorecard, the competencies you own in this loop and their anchors. The resume tells you where to probe, not what to assess.

Mistakes during the interview

2. Deciding in the first minutes

What it looks like: a strong handshake or a familiar employer, and the rest of the hour is spent confirming it. Why it matters: in a field study of 79 audiotaped interviews, Dougherty, Turban and Callender (1994) found interviewers' first impressions, formed from application and test information, were related to how they behaved: with a positive impression they showed more positive regard, sold the company and gathered less information. Swider, Barrick and Harris (2016) found impressions from rapport-building related more strongly to ratings of early structured answers than later ones. Fix: keep the warm-up short, write notes rather than verdicts during the interview, and score each competency only after the interview ends.

3. Selling instead of assessing

What it looks like: the interviewer likes the candidate and spends half the slot on the company's story. Why it matters: it is the confirmation behavior Dougherty and colleagues observed, and it leaves less evidence on the scorecard for exactly the candidates you are most likely to hire. Fix: questions first, selling last, and the same selling time for every candidate.

4. Going off the agreed questions

What it looks like: "I like to just have a conversation." Why it matters: standardized questions are one of the factors that raised interviewer agreement in the meta-analysis by Conway, Jako and Goodman (1995). In experiments by Dana, Dawes and Peterson (2013), people formed impressions just as confidently from an unstructured interview in which the interviewee's answers were random as from one with real answers. An unstructured conversation feels informative whether or not it is. Fix: ask the core questions as written and in order; improvise only the follow-ups. See what is a structured interview.

5. Leading follow-ups

What it looks like: "So you'd say you're comfortable with conflict?" The candidate says yes, and the interviewer writes it down as evidence. Fix: follow-ups that ask for a specific example and the candidate's own part in it: "Tell me about the last time that happened. What did you say?"

6. Off-limits questions

What it looks like: friendly small talk about children, health, age or where someone's accent is from. Why it matters: the U.S. Equal Employment Opportunity Commission says employers are explicitly prohibited from making pre-offer inquiries about disability, and that questions which reveal race, sex, national origin, religion or age may be used as evidence of intent to discriminate unless they have a business justification (EEOC). Fix: know the list in illegal interview questions, and when a candidate volunteers personal information, acknowledge it briefly, do not follow up, and leave it out of your notes. This is general information, not legal advice.

7. Talking too much

What it looks like: long questions with the answer built in, filling every pause, answering your own question when the candidate hesitates. Why it matters: every minute you talk is a minute of evidence you did not collect. Fix: ask, then wait several seconds. If you have a transcript of your own interviews, look at how much of it is you.

8. Notes of conclusions instead of the candidate's words

What it looks like: "strong communicator," "seems smart." Why it matters: Google's re:Work guidance tells interviewers that statements like "seems smart" do not help a hiring committee decide, and not to restate rubric wording in place of examples (Google re:Work). In a laboratory study, Middendorf and Macan (2002) found note-taking improved recall accuracy, and that being able to review notes improved judgment accuracy for conventional note-takers. Fix: write what the candidate said, in their words, and keep your interpretation for the scorecard. The method is in evidence-based interview feedback.

Mistakes when scoring

9. Halo: one answer colors every score

What it looks like: a brilliant answer on product strategy, and suddenly every competency is a 4. OPM's guide defines the halo effect as letting performance on one competency influence ratings on others. Fix: score one competency at a time, and for each, point to the evidence that belongs to that competency only.

10. Contrast with the previous candidate

What it looks like: an average candidate after a weak one gets a 4. OPM's guide warns that the order of candidates can affect ratings. Fix: score against the anchor before you see the next candidate, and never write "better than the last one" as evidence.

11. Everyone gets a 3, or everyone gets a 4

What it looks like: an interviewer whose scores never move. OPM lists central tendency, leniency and strictness as separate errors, along with similar-to-me ratings for candidates who resemble the interviewer. Fix: once a quarter, put your scores beside the panel's. If your share of 4s is double everyone else's, bring it to the next calibration session; inter-rater reliability shows how to measure the gap.

Mistakes after the interview

12. Late, vague or shared feedback

What it looks like: scorecard written three days later, after a hallway chat with the hiring manager. Why it matters: Google says its internal research found feedback submitted within 48 hours is significantly higher in quality, and talking first turns independent evidence into an echo. OPM also names "pressure to hire" as a mistake: deciding fast on a limited sample of information because the seat is empty. Fix: block 15 minutes after every interview, submit before reading anyone else's feedback (see independent interview feedback before the debrief), and let the agreed decision rule, not the deadline, make the call.

How can you spot these mistakes in your own interviews?

You are the worst-placed person to notice your own habits in the moment. Three ways to see them afterwards:

• Ask to be observed. A colleague sitting in with the checklist from interviewer shadowing will spot leading probes and skipped questions in one session.
• Read your own transcript or notes against the kit: which core questions did you ask as written, how often did you lead, how much did you talk?
• Compare your scores with the panel's over a quarter, by competency.

Tools can help with the skipped-question problem in the moment. In Interview Signal, for example, must-ask questions from the job description stay pinned on screen until they have been asked.

INTERVIEWER SELF-CHECK (after each interview)
[ ] I read the scorecard and anchors before the interview
[ ] I asked every core question as written
[ ] My follow-ups asked for examples, not agreement
[ ] I did not ask about, or note, personal topics
[ ] My notes are mostly the candidate's words
[ ] I scored each competency on its own evidence
[ ] I did not compare this candidate with the last one
[ ] I submitted my scorecard today, before reading anyone else's