How do I know my answer is too long?

What the practice interview knows about length, and what it does not

The tool puts a hard ceiling on length: three minutes per question, with a countdown on screen. When 30 seconds are left the timer turns amber, and at zero the answer is sent, even mid-sentence. The limit is deliberately generous. It is not there to rush anyone, but a practice answer that runs past it has usually stopped being an answer and turned into a presentation.

The rating works differently from the clock. After the conversation an AI model reads the transcript of your answer and judges it from the point of view of the panel member who asked: substance, structure and fit, each from 0 to 10, plus a short explanation and a note on what a strong answer would have included. No duration appears anywhere in that rubric. Overlong answers still show up, mostly under structure, because a transcript with three false starts, two side stories and a missing conclusion reads exactly the way it sounded.

There is a live signal too. The simulated interviewer may ask one short follow-up when an answer stayed vague. If you are asked, after two minutes of talking, what your own contribution actually was, you did not say too little. You said a lot, just not the part that mattered.

Four warning signs you can notice yourself

One: the point comes late. If your answer opens with background (how the project started, who was on the team, what had gone wrong before) the listener is waiting for something they have not heard yet. A good answer says in its first sentence where it is heading and tells the story afterwards, as evidence.

Two: you add a second example. That almost always comes from doubt about whether the first one landed. A second example does not double the effect; it splits attention between both. If the first was well chosen it is enough, and an interviewer who wants more will ask.

Three: you repeat yourself in different words. Phrases like "so, as I said" or "just to sum up again" in the middle of an answer are a reliable sign you are looking for the thread. A summary at the end is fine; one halfway through is not.

Four: you lost the question. After a practice answer, try a simple test. Can you still say the question word for word, and did your last sentence answer it? Someone asked about a conflict in their team who ends up on the merits of agile methods has not answered at length; they have answered something else.

How long is right? Guides by question type

There is no fixed rule, and no study sets an ideal number of seconds. Interview preparation has settled on some useful guides, though. Spoken aloud, a minute is roughly 120 to 150 words, so a 90-second answer is about half a page of text.

The opener, "tell me about yourself", takes 60 to 90 seconds: where you are now, the two steps that lead to this role, and why you are applying here. A behavioural question answered with the STAR method (situation, task, action, result) usually needs 90 to 120 seconds, because it tells a small story. Technical questions built on a scenario from the job advert can run slightly longer when you walk through steps, but they should still open with the outcome.

Factual questions are the most common trap. "What is your notice period?" or "What are your salary expectations?" deserve one sentence, two at most: the period, or a range stated as an annual salary with a note on whether the 13th-month salary is included. Start explaining why the notice period is so long, or why the figure is justified, and you sound unsure even though every word is true.

The Swiss bar: brief and backed by evidence

In Swiss interviews a sober answer you can back up generally goes down better than an embellished one. Recruiters want to know what you did and how anyone could tell, not how enthusiastically you describe it. A long answer full of adjectives easily reads as self-promotion here, and that rarely helps.

Company size moves the bar a little. At a corporation with more than 200 employees a panel tends to expect structured STAR answers and the formal register, and a well-organised two-minute answer is appropriate. At a small firm of five to fifty people pragmatism and ownership count for more, at a micro company first names come quickly, and the same answer can feel overly thorough. The practice interview asks for company size when you set it up and rates with exactly that difference in mind.

If you are arriving from abroad, or have worked mainly outside Switzerland, your habits are not a bad guide, but they are not a safe one either. If you learned that a detailed build-up shows rigour, the Swiss version is to lead with the result and offer the build-up rather than deliver it.

A worked example: the same answer, twice

This is a constructed example, not real output from the tool. The question: "You led a team of four; why do you want to go back to hands-on work?" The first answer runs close to three minutes. It opens with the company's history, explains the restructuring, mentions two projects, moves on to work-life balance, returns to one of the projects and ends with "so yes, that's more or less the reason". Everything needed is in there somewhere.

A plausible rating for that version: substance 7, structure 3, fit 5, overall 5 out of 10. The explanation would say something like: the motivation is recognisable but arrives late, with no clear link to the advertised role. The note on a strong answer: the reason in the first sentence, and evidence that the specialist role is a choice, not a fallback.

The cut version takes about 80 seconds: "Because I realised the specialist work matters more to me than managing. In two years as team lead, the project I'm proudest of is the migration I built myself. Your role is exactly that kind of work, responsible for the result but without line management." Same substance, one structure, a clear link. Structure 8 and fit 8 would now be plausible. Nothing important was cut, only everything that came before the reason.

How to cut: point, evidence, link, stop

The simplest cutting rule is an order. First the point in one sentence, answering exactly the question asked. Then one piece of evidence, a concrete example with a result someone could check. Then a sentence that connects it to the role. Then stop, even if the silence feels odd for a moment.

That silence is where many answers grow long. People keep talking because no reaction comes, and they add clauses. In a real interview the pause is often just someone taking notes. Ending deliberately on a clear last sentence gives the other side room to follow up, and a follow-up question is a good sign, not a gap.

When practising, answer the same question twice: once freely, once with the intention of stopping right after the link to the role. The "Redo interview" button keeps the position, company size and panel, though the questions are generated afresh, so compare the attempts on the structure score rather than word for word. To repeat the exact same question, a phone recording is the simplest route.

What the score cannot see

The rating reads text. It cannot tell whether you spoke calmly or rushed, whether you left long pauses, or whether a real interviewer would have checked the time after a minute. An answer can look tight in the transcript and still be too long out loud, because it was delivered slowly and with filler words that a transcript smooths over.

For that part another tool is often better: a voice memo on your phone played back against a stopwatch, or a friend who will honestly tell you when they stopped listening. Both are free. A coach is worth paying for if you can recognise the length but still cannot brake in the moment, because then the issue is nerves more than structure.

And the main limit: a good structure score says nothing about whether you will get the job. It measures your answer against a rubric, not against the other candidates or the mood in the room. What it does show reliably is whether your answer has a recognisable core, and that is the question hiding behind "too long" almost every time.