Interviewer talk-time ratio: how much should the interviewer talk, and how to measure it

In an assessment interview the candidate should do most of the talking, because the interviewer's job is to collect evidence. A recruiter screen that explains the role runs closer to even, and a sell call is mostly the interviewer. Measure it from a speaker-labelled transcript: interviewer words (or seconds) ÷ total words (or seconds) × 100, with the opening and the candidate's questions counted separately. There is no research-based ideal share; the only published benchmark we found, a vendor's own analysis published July 29, 2026, put the median interviewer share at 46.8% across 2,657 interviews.

Talking too much is one of the habits in common interviewer mistakes. This page is about measuring it, reading it by interview type, and knowing its limits.

How do you measure talk-time ratio?

1. Get a transcript with speaker labels, recorded or transcribed with the candidate's knowledge and any consent your jurisdiction requires; see the interview recording consent script.
2. Check the labels. If both people speak into the same microphone, a tool that labels by audio source can attribute everything to one side. Spot-check a few turns.
3. Mark the segments: opening, assessment questions, candidate's questions and close.
4. Count words per speaker per segment. In a spreadsheet with the speaker in column A and the text in column B, this formula counts the words in a turn: =IF(LEN(TRIM(B2))=0,0,LEN(TRIM(B2))-LEN(SUBSTITUTE(TRIM(B2)," ",""))+1). Then total by speaker with SUMIF.
5. Divide the interviewer's words by the total for each segment and for the whole interview.

Words or seconds? Timestamps give a true share of airtime; words are easier to count. They differ slightly: in the vendor analysis cited below, interviewers spoke faster than candidates (a median of 172.2 against 158.6 words per minute), so a word-based share reads a little higher for the interviewer than a time-based one. Pick one method and keep it.

If you use Interview Signal, "Save transcript as .md" on the review screen writes the transcript with each turn marked Interviewer or Candidate, which pastes into a spreadsheet for the count above.

Worked example: one 45-minute interview

The whole-interview figure, 41.7%, hides a well-run assessment section at 27.1%. The opening and the candidate's questions are meant to be interviewer-heavy: you explain the format, then answer what they ask. Comparing whole-interview ratios between an interviewer who leaves fifteen minutes for questions and one who leaves five would be unfair to the first.

What does a low interviewer share look like in practice?

The assessment segment above is mostly short questions and short probes around long answers. An invented excerpt, with word counts:

That stretch runs at about 8% interviewer share, and every interviewer word moves the answer toward something scoreable. Compare a 60-word question that explains why the topic matters, offers two example situations and then asks two things at once: the share goes up and the evidence goes down.

How do you measure it in a panel interview?

Count each panelist separately as well as the panel as a whole. A panel's combined share can look fine while one panelist does most of the talking and another asks nothing; both are worth a conversation. Agree in the plan who asks which questions, so the split is deliberate, and compare each panelist's share only for the questions they owned. The panel interview tips for interviewers cover how to divide the questions.

What should the ratio be, by interview type?

These are working rules from how each interview is designed, not research thresholds. Set your own expected range per interview type, then look at interviews that fall well outside it.

Is there a published benchmark?

The only one we found is a vendor's analysis of its own customers' interviews. Metaview's "Who talks more in an interview, you or the candidate?", published July 29, 2026, reports from 2,657 interviews with speaker-attributed transcripts in its corpus:

• a median interviewer share of 46.8%, and 53.2% for the candidate;
• interviewer share of 27.9% at the 10th percentile, 36.8% at the 25th, 56.7% at the 75th and 65.1% at the 90th (percentiles from 2,688 interviews);
• 5.6% of interviews where the interviewer held more than 70% of the talking (from 2,542 interviews).

Read it for what it is: one vendor's customers, limited to interviews where speakers could be separated cleanly, with no split by interview type and no link to hiring outcomes. It shows how wide the spread is between interviewers more than it sets a target. Your own baseline, by interview type, is the comparison that matters.

Why does the interviewer's share creep up?

• Long questions with context, examples and two asks in one sentence.
• Filling silences. A candidate thinking for five seconds is not a gap to fill.
• Answering their own question when the candidate hesitates, which also leads the answer; see leading questions in interviews.
• Selling too early. Pitching the company before the assessment is done.
• An early decision. In a field study of 79 recorded interviews, Dougherty, Turban and Callender (Journal of Applied Psychology, 1994) found that interviewers with positive first impressions showed positive regard, sold the company and gave job information, and gathered less information. A share that jumps halfway through an interview is worth a second look.

What can the ratio not tell you?

• Whether the questions were good. A low share with weak questions still produces weak evidence.
• Whether each competency was covered. Check the scorecard evidence, not the talk share.
• Whether the candidate was rambling. A very low share can mean the interviewer never steered back to the question.
• Language and nerves. Candidates speaking a second language, or nervous candidates, may give shorter answers that need more prompting.
• Interruptions and tone. Two interviews with the same ratio can feel completely different to the candidate.

Use it as a coaching signal alongside scorecard quality, never as a performance target on its own; a target invites interviewers to stop probing.

How do you bring your share down?

1. One question per sentence, read from the kit.
2. Wait. Count to five after a question before rephrasing.
3. Probe in a few words: "What happened next?", "What did you say?" More in probing questions for interviewers.
4. Move the pitch to the end, or leave it to the recruiter.
5. Keep summaries short: one sentence to check understanding, then the next question.

How do you use it in coaching?

Review two or three interviews per interviewer each quarter, using the assessment segment only, and track the trend rather than a single interview. Pair the number with the transcript: find the longest interviewer turn and ask what it achieved. The routine in how to coach recruiters with call reviews works for hiring managers too, and how to evaluate your interviewers covers the other measures to put beside it.