How do I improve an answer that scored 60%?

First: 6 out of 10, or 60 per cent?

The report carries two numbers, and they call for different work. Next to each question is a score from 0 to 10 for that one answer. At the top right is the pass probability in per cent, a verdict on the whole round that only appears once you finish the session.

So "I got 60%" means either a single 6 or an amber session. With a single 6 you are working on perhaps a minute of speech. With an amber session you first find which answers pulled the verdict down, then work on those. Mixing the two up is how people polish answers that were already fine and leave alone the one that bothered the panel.

The explanations on scoring and the criteria for a good score have their own pages; this section covers the precise adjustments you make to the answer.

What a 6 usually means

An answer around 4 typically has nothing checkable in it: adjectives instead of experience. A 6 is a different case. The example is there, the experience is real, the vocabulary is right, and still the answer does not quite land. There is nearly always one of three reasons.

The first is a missing thread: the answer opens with generalities, jumps to the example, drifts back to generalities and trails off instead of ending. The second is a missing result: you say what you did but not what was different afterwards. The third is fit: the answer does not suit the person who asked. The subject-matter expert wanted your method and got your motivation, or the other way round.

That gives the most important rule for a 6: do not rewrite the answer from scratch. Swapping a usable story for a different one starts you over with an untested answer and often throws away the part that was already working. Keep the core and change the shape.

Read the feedback as an edit brief

Under every score are two or three sentences of reasoning and one line on what a strong answer would also have included. That last line is the most useful thing in the whole report. Internally the score is built from three sub-scores, substance, structure and fit, but what you see is the overall score and the text, and the text tells you which of the three was short.

Turn that line into exactly one change you can say out loud. "A strong answer would have quantified the outcome" becomes one sentence with a number or a visible before and after. "A strong answer would have connected to the role as advertised" becomes a closing sentence that picks up the wording of the job advert. More than one change per attempt is too many, because you will not know which one did the work.

The conversation itself gives a second signal. The simulated panel may ask one short follow-up when an answer stays vague, and the follow-up and your reply are scored along with the rest. If the panel pushed ("And what came of it?"), that piece of information belongs in your first answer. In a real interview not every recruiter asks; some simply note the gap and move on.

Example 1: the content is right, the thread is missing

Setup: financial controller, corporation with more than 200 employees, the head of finance on the panel. Her question: "The month-end close shows a variance nobody can explain. How do you approach it?" The first answer: "Well, variances happen all the time, that's normal in controlling. I look at the postings first, then I talk to the cost-centre owners, and at my last employer it once turned out to be an invoice booked twice. You just have to be systematic. Yes, that's basically how I usually do it."

On the rubric that is a typical 6. The method is sound and a real case appears. But the case is buried in the middle, there is no outcome, and the answer ends on filler. In a corporate setup, where structured answers are expected, that costs points. The line under the score would read something like: a strong answer would have set out the approach in clear steps and shown it on the concrete case, with its result.

The rebuilt answer keeps everything and puts it in order: "I work in three steps: check the affected accounts down to individual postings, then ask the cost centre, then document the cause so it cannot recur. At my last employer it was a supplier invoice of around twelve thousand francs booked twice. I found it on day two, and since then invoice numbers are matched automatically before approval. That's what I'd bring to building out your close process." The only additions are an order, an outcome and a closing sentence. An answer like that typically lands at 8.

The example is designed to expose the underlying mechanism. The scores illustrate the evaluation criteria; they do not constitute a guaranteed outcome.

Example 2: the right answer for the wrong person

Setup: accounts assistant, company of 5 to 50 employees, the managing director on the panel, who also covers the HR questions. Her question: "What salary are you expecting?" The first answer: "The overall package has to be right, and I'm flexible. What have you budgeted for the role?"

The answer is polite and not wrong, but it dodges. In Switzerland the salary question is an ordinary part of a first interview, and naming no figure hands the work back to the employer. Many candidates arriving from the UK, the US or elsewhere in Europe are used to deflecting at this stage and find the directness uncomfortable. On the rubric an answer like this stays low on fit even when everything else is tidy: typically a 5 or 6.

The edit is a range, stated as gross annual salary, saying what it covers: "I'm looking at CHF 78,000 to CHF 84,000 gross a year at 100 per cent, including the 13th-month salary. That's based on my experience in accounts payable and on comparable figures for the region. I'm happy to talk about the exact number once I know the duties in detail." Three things make the difference: the figure; being clear about the workload percentage (the Pensum) and the 13th salary, which many Swiss employers pay and which changes what a monthly figure means; and an open close that suits the pragmatic, informal culture of a small firm.

The amounts are an example, not salary advice. Work out your own range from salary comparisons for your occupation, canton and experience before you practise the answer.

From a 60% pass probability to the next round

Open the "Questions" tab first and look for the one or two answers that sit clearly below the rest. Then the "Report" tab: the panel's concerns and the note from each panel member show what the round hinged on. The two trails often meet, for example a weak answer to the team lead's question and a concern in the team lead's note.

Work only on those, using the method above: keep the core, make one change, add a closing sentence. Say the new version aloud two or three times without reading it. An answer you have only written sounds different in conversation from how it looks on paper, especially in a language that is not your first.

Then choose "Redo interview". The new session keeps the position, company, company size and panel; you supply the CV and job advert again. The questions are generated afresh, so you will not get the same question in the same words. That is not a drawback. A real panel will not phrase its question the way you rehearsed it either, and the evidence that your edit has stuck is that you can carry it over to a similar question, not that you can repeat a script.

Compare the new number only with sessions in the same setup. A 60 at a corporation with a four-person panel and a 60 at a small firm with one interviewer are two different results.

Where the number stops helping

What gets scored is the content of your answer, meaning what you said. If your issue lies in the manner in which you say it (too fast, too quiet, or long nervous pauses), it only shows up indirectly, perhaps as a missing thread. For that, a video recording of yourself, a trusted friend who will give an honest appraisal, or a coach when a lot is at stake is the better mirror.

Nor do you need to spend scored questions on drafting. Write the new version, read it aloud, have a chat assistant or a colleague look it over. The scored question is worth using for the test after the edit. A free account gives five scored answers on the first day and five a month after that; the subscription at CHF 15 a month or CHF 95 a year allows forty a month, and eighty on the yearly plan. Checked on 16 September 2026.

And do not chase a 10. A memorised answer can score well on a transcript and still sound recited in the room. What you want is a frame (core, result, link to the role) that you fill freely for each question. Once your scores sit steadily at 7 or 8, move on to the next answer rather than repeating the same one.