Interview answer length, from both sides of the call

On 28 September 2026, during live test calls on our staging system, our AI interviewer kept cutting in on the test candidate before an answer was finished. It was waiting 1.2 seconds of silence before deciding an answer was complete, and people pause for longer than that in the middle of a thought, especially when the question is about their own work. We changed the rule that day. On conversational jobs the interviewer now waits 1.5 seconds after a finished sentence, and up to 2.8 seconds when a sentence is left hanging, because a sentence that trails off usually means the candidate is still thinking.

That change sits at the heart of the question in this post. How long should interview answers be sounds like a question about talking time. In practice it is about three things together: how long the candidate talks, how long they need to think, and how much time the interview has in total. Most advice covers only the first. This guide covers all three, using the same question-by-question time budget we use to size a whole round, for candidates preparing an answer and for interviewers who have to keep a round on time without cutting anyone short.

How long should an interview answer be?

Most interview answers should run 1 to 2 minutes. Factual questions need 20 to 45 seconds, an introduction 60 to 90 seconds, and a behavioural “tell me about a time” answer 90 seconds to 2 minutes. Deal-breaker questions need one clear sentence. A pause to think does not count against you.

The ranges are wide because questions are different. A notice-period question has a one-line answer, and padding it makes the candidate sound unsure. A question about a difficult project needs a situation, the candidate’s own part and a result, and that cannot be done well in thirty seconds.

Interviewers can use the same table in reverse. If a candidate answers a behavioural question in fifteen seconds, the answer is not finished: ask what they did themselves. If they spend three minutes on a deal-breaker, they are probably avoiding it: ask for the number.

The arithmetic of a 15-minute interview

Answer length only becomes a real problem when the interview has a fixed length, and most first-round interviews do. A HireQwik interview gives a heads-up at 15 minutes and closes at 20. A recruiter’s screen usually lives inside a 30-minute calendar block. Either way, every answer is drawing on the same small budget.

Here is what that budget looks like for a typical screen. Take two deal-breaker questions and four core questions, each core question with one follow-up:

• Opening and introduction: about 2 minutes.
• Two deal-breakers: under a minute between them.
• Four core questions at 2 minutes each, plus about 45 seconds per follow-up: 11 minutes.
• Candidate’s questions and the close: 2 minutes.

That is roughly 15 minutes, and it only works if the core answers stay near two minutes. If the first answer runs to six minutes, the interview does not get longer. Something else gets shorter: a follow-up is dropped, the last question is rushed, or the candidate’s own questions disappear.

In a time-limited AI interview the stakes are concrete. Questions that are never reached cannot count in the candidate’s favour, and if too little of the screen is covered before the call ends, HireQwik marks the interview Incomplete rather than giving it a low score. We explained that rule, and why a partial call must not read as a weak one, in what an incomplete AI interview means. The better outcome for everyone is answers sized so that every question gets asked.

A pause is part of the answer

Back to the 1.2 seconds. The reason it failed is that conversation and interviews run on different clocks.

Researchers at the Max Planck Institute recorded question-and-answer pairs in ten languages from five continents, and found the same pattern in all of them: people avoid overlapping and keep silences short. Across the whole dataset the average gap before an answer was about 208 milliseconds, and even the slowest language in the study, Danish, averaged under half a second. That is the rhythm of everyday talk, and it is why a two-second silence in an interview feels enormous to the person who is silent.

But an interview question is not everyday talk. “Tell me about a decision you got wrong” is a question people have to think about, and the honest answer often starts with a pause. An interviewer, human or AI, who treats a pause as the end of the answer is not measuring the candidate’s communication. They are measuring how fast the candidate can start talking, which is mostly nerves.

So we made our interviewer more patient than conversation is, on purpose. The 2.8 seconds it waits after a sentence that trails off is roughly six times the slowest everyday average in that study. It is still short enough that a finished answer does not hang in silence.

For candidates, the practical advice is simple. Pausing for two or three seconds before a hard question is fine. If you need longer, say “let me think about that for a second”; it tells the interviewer you are not finished. For interviewers: count to three before you speak. You will hear the second half of a lot of answers you used to cut off.

What happens if you talk too long in an AI interview

Candidates ask this more than almost anything else about AI interviews, so here is the exact answer for HireQwik.

There is no timer on individual answers. The interviewer waits for a pause before it speaks, and the length of an answer is never a reason for it to interrupt. On conversational jobs it is the patient version described above. What is limited is the whole interview. At 15 minutes (17 on conversational jobs) the interviewer starts to wrap up, and on a standard interview you can talk over that heads-up and finish your point. At 20 minutes the interview closes with a fixed thank-you, and that closing line cannot be interrupted.

That design has a consequence worth knowing. A long answer is never punished directly, but it is paid for indirectly, by the questions that come after it. If you spend eight minutes on the first question, the interviewer may reach the end with two questions unasked. Those questions are not scored against you, but they cannot be scored for you either.

The interviewer also keeps track of what it still has to ask. On conversational jobs it gets a short timekeeper note every three minutes, telling it how far into the interview it is and which of the hiring team’s questions are still open. When there is plenty of time, the note pushes it to dig in: ask once more about what the candidate did personally, for an example, or for the figure behind a claim. So a short answer early on does not end the topic. It usually earns a follow-up.

If you want the full picture of what happens at each limit, including what HR sees afterwards, we wrote it up in what happens when an AI interview hits its time limit.

How interviewers can redirect a long answer kindly

Most long answers are not arrogance. They are nerves, or a candidate who was never told what the interviewer wanted. Cutting them off bluntly makes both worse. Redirecting them well takes one sentence.

The trick is to interrupt with a question, not a stop sign. Wait for a breath, then narrow the question:

• “That’s helpful. What was your own part in that, specifically?” It moves a team story to the candidate’s actions.
• “Can you give me the number for that?” It turns a description into a fact.
• “Let’s pause there, because I want to make sure we get to everything. The next question is…” It is honest about time without blaming the candidate.
• “If you had to pick one example, which one would it be?” It ends a list of five half-examples.

What not to do: look at the clock, sigh, or say “we’re short on time” without moving on. That transfers the interviewer’s planning problem to the candidate.

This is also exactly what we taught our AI interviewer to do. When time is short, its timekeeper note tells it to move on to what is still unasked rather than dig further into an answer it has already heard.

A two-minute answer structure for candidates

Candidates who give good two-minute answers are usually not more confident. They have a structure. The one below works for any “tell me about a time” question and fits comfortably inside two minutes:

1. The situation, in one or two sentences (about 15 seconds). “In my last role I handled renewals for about 80 small business accounts.”
2. What you did yourself (about 60 seconds). Use “I”, not “we”. Name the steps.
3. The result, with a number if there is one (about 20 seconds). “Renewals went from roughly 70% to 85% over two quarters.”
4. What you would do differently, or what you learned (about 15 seconds). This is the part interviewers remember.

Then stop. The silence after a finished answer is the interviewer’s turn, and a good interviewer will use it for a follow-up. If you are preparing for a HireQwik interview specifically, our candidate briefing covers the setup and what to expect, and if nerves are the problem, how candidates can handle nerves in an AI interview is written for exactly that.

Signs an answer is too short

Long answers get all the attention, but a short answer is just as costly in a time-limited interview, because the time it saves is rarely spent well. Three patterns are worth recognising, whichever side of the call you are on.

The one-line answer to a story question. “Yes, I have handled escalations.” That is a claim, not evidence. Interviewers should ask for the last time it happened, and candidates should assume that is the question they were really asked.

The “we” answer. “We launched the new process and we hit our targets.” It may be a perfectly good story, but nobody can tell which part was the candidate’s. One follow-up, “what did you do yourself?”, usually doubles the length and triples the value.

The answer without a number. “Sales went up a lot.” If there was a number, say it; if there was not, say roughly how big. “About 20 more orders a week” is checkable. “A lot” is not.

None of these are signs of a weak candidate. They are usually signs of a candidate who does not know how much detail the interviewer wants. A good follow-up fixes all three in under a minute, which is why our interviewer is told to use spare time on exactly these follow-ups. The same follow-ups, written as situational interview questions, work for human interviewers too.

Write questions that have a natural answer length

Half of the answer-length problem is created by the person writing the questions. Three habits fix most of it.

Ask for one thing at a time. “Tell me about your last role, why you are leaving and what you are looking for” is three questions, and the candidate will spend four minutes trying to answer all of them. Before a conversational HireQwik interview starts, the system splits compound questions like that into separate parts, so each gets its own answer and a later part can be skipped if the candidate already covered it.

Ask for one example, not examples. “Give me an example of handling an angry customer” gets a two-minute story. “How do you handle angry customers?” gets a three-minute philosophy.

Write the strong answer down before the interview. Google’s guide to structured interviewing recommends grading every candidate on an identical scale with defined answer levels. Once you know what a strong answer contains, you also know roughly how long it takes to say, and you can tell a candidate who is done from one who is still getting there.

The short version

For candidates: one to two minutes for most answers, ninety seconds for your introduction, one sentence for a deal-breaker, and pausing to think is allowed. For interviewers: budget the interview from the questions, wait longer than feels natural before you speak, and redirect long answers with a narrower question.

We built both halves into HireQwik: a patient interviewer that waits out a candidate’s thinking, no timer on any single answer, and a whole-interview limit that keeps a screen at volume to about fifteen minutes. You can hear the pacing yourself in the five-minute live demo. Try pausing mid-answer and see what the interviewer does.