Interview Skills

Why strong candidates give weak interview answers

An analysis of 3,700+ interview answers scored by our own model rather than by hiring panels, plus the 7,582 coaching notes written on them. The average answer scored 52.1 out of 100, and the most common note asked for a detail the candidate already had.

AM

Andre Mendes

·

August 25, 2026

·

9 min read

A nurse moving from emergency nursing into post-anesthesia care sat down to practice, and spent the opening of her answer explaining what she had not done. The coaching note that came back read: "You spend a lot of time explaining that your experience is not specifically PACU, instead of leading with the direct ways it applies." She was qualified for the job she was preparing for. The answer she gave was aimed at the question she was afraid of.

Strong candidates give weak interview answers because interviewing is a speaking task and almost all preparation is a reading task. Across 3,700+ answers scored by our own model rather than by hiring panels, the average answer scored 52.1 out of 100, and roughly three in ten coaching notes asked for a detail the candidate already had.

The average readiness score on a completed practice session was 43.9 out of 100, across 691 completed sessions. These are not scores from people who wandered in unprepared: they come from working professionals who went out of their way to practice, most of them within days of a real interview at a named employer. Against the standard our own model applies, the typical answer lands at about half.

What can you fix the night before an interview?

The four findings below reduce to four passes over material you already have, in the order they cost the most.

1. Put one number in each of your three best stories. Scope counts: how many people, how many cases, how many weeks.
2. Read each story for the word "we" and add the one sentence that says what you personally did.
3. Confirm every answer states an ending. If a story stops at the hard part, it is unfinished.
4. Say the first six words of each answer out loud and cut any negative you find sitting there.

Every one of those is an edit to something you already lived, which is why it fits in an evening.

Where these numbers come from, and their limits

Everything below was queried from the production database of Mock Interview Pro on August 25, 2026.

• Completed mock interview sessions · 691
• Scored answer blocks inside live sessions · 2,712
• Scored standalone practice answers · 550
• Scored answers from anonymous practice · 439
• Individually scored answers in total · 3,700+
• Improvement notes written on those answers · 7,582

Everyone in this sample chose to practice, and the direction of that bias is unknown: choosing to practice may select for the better prepared, for the more anxious, or for both at once. The scores come from the platform's own scoring model, not from the hiring panels these candidates faced, so a 52 here is not a hiring manager's 52. It is one instrument applied consistently across thousands of answers, and we have not validated it against hiring panel judgments, so treat comparisons inside the data as suggestive.

The improvement notes were grouped into themes by matching their titles and content, so the percentages describe how a coaching model chose to phrase its feedback, which is a weaker thing than an audit of what the candidates actually said. And nothing here connects a score to an offer, because hiring outcomes are not tracked at all. What survives those four limits is still several thousand real answers from working adults, each carrying a written diagnosis, and the diagnoses repeat.

About 22% of 7,582 coaching notes ask for more specific evidence

Of the 7,582 improvement notes, the largest single theme is a request for more specific evidence, at about 22%. The second largest is a request to quantify the result, at about 9%. Roughly three in ten notes in the whole set are some version of one instruction: make this concrete.

Specificity is not knowledge, which is what makes that number strange. Picture a district manager who says she turned around an underperforming store: she has the plan percentages in her head, she reported them monthly, and she said the abstract version anyway. The sentence-level repair for that habit is laid out in why interview answers come out vague and how to make them specific.

The notes that ask you to clarify your own role

The second cluster is stranger, and experienced candidates fall into it hardest. Three recurring note titles are "clarify your role" at 43 occurrences, "clarify your ownership" at 32, and the thinner "clarify personal contribution" at 8.

They fire on answers where somebody described a real project accurately and left themselves out of it. Picture a police sergeant, call him Marcus, describing a shift-scheduling overhaul: "we restructured the rotation, we brought the union in early, and we got sick-time usage down." Every word of that is true, and none of it tells the panel what Marcus himself did, which is the only thing they are trying to find out, since they are not hiring his department. Marcus is not hiding anything, he is describing the work the way he would describe it to another sergeant. Claiming credit in the first person is socially expensive in most workplaces, and the people who are best at collaborative work are the most trained out of it. A teacher who has spent years on a grade-level team thinks in "we" for good reasons, and a panel hiring an instructional coach still cannot tell from her answer whether she led the literacy block redesign or attended the meetings about it. In both cases the note asks for one sentence that separates the person from the group. Repairing that habit, along with four others drawn from the same corpus, is the subject of the five most common interview answer mistakes.

"Provide a complete answer" is the most frequent note of all

The single most frequent literal note title in the dataset is "provide a complete answer," at 117 occurrences.

That one rewards a careful read, because it fires on answers that stopped before finishing the question that was asked, whatever their length. Somebody asked about a time they handled a conflict with a colleague gives a rich account of the conflict and trails off without saying what they did or how it ended. Somebody asked about their first ninety days describes a philosophy of the first ninety days and never lands on week one.

We cannot see from the data why the answer stops.

The pattern is only that the setup is complete and the landing is missing. Two other frequent titles, "tighten the wording" and "lead with one example," sit on the other side of the same question about how much to say.

"Connect experience to role" and the candidates who apologize first

The last major cluster is tailoring, and it arrives as two note titles: "connect experience to role" and "name the target role." They fire when somebody describes their experience truthfully and never closes the loop to the job in front of them, which leaves the panel to do the translation.

It lands hardest on the people who are otherwise the most qualified in the room, the ones moving into a seat adjacent to one they have already proved: a different specialty, a different sector, a different level, or a different country. Listen to how the nurse from the top of this page opened her answer.

"Even though I don't have specific experience as a PACU nurse..."

The gap goes first and the relevance waits behind a conjunction, which is what the coaching note picked up. If you are interviewing for a seat you have not technically held, what to say when you are interviewing for a job you have not done picks the problem up from there.

86% of second attempts scored higher than the first

The dataset contains 264 cases where somebody answered the same question a second time after reading the feedback on the first attempt, across 58 candidates.

• Pairs · 264
• Candidates · 58
• Scored higher on the second attempt · 86%
• Scored worse on the second attempt · 14%
• First attempt average · 60.1
• Second attempt average · 68.2
• Average gain across all pairs · 8 points
• Average gain among those who improved · 11.6 points

The other 14% scored lower on the second attempt, which is what happens when somebody overcorrects and recites instead of talking. The sample is also 58 people motivated enough to retry, a scoring model can drift upward on a second look at similar content, and there is no offer data here whatsoever.

Their experience did not change between the two attempts, and they had written feedback in hand for the second one. An 8 point average gain in people whose experience did not change is consistent with a production problem rather than a knowledge problem. It does not prove it, for the reasons above.

What the four clusters can and cannot tell you

Line the four up and each one describes something missing from the answer. Whether it was also missing from the candidate is the question this data cannot settle: a note asking for specifics fits a person who had the detail and skipped it just as well as a person who never had it.

What the notes do describe consistently is the answer itself. Each of the four was written about what reached the interviewer, which is also the only material the interviewer had to work with.

That is one reason the standard preparation ritual moves the number so little. Reading lists of common questions rehearses recognition, and an interview tests production. You can read forty articles about the STAR structure and still, when a stranger asks about a difficult stakeholder, produce four minutes of context and no result, because your mouth has never done that task under that pressure.

Frequently asked questions

How were these interview answers scored?

Each answer was scored 0 to 100 by the platform's own scoring model, against the requirements of the specific role the candidate was preparing for, with written strengths and improvement notes attached. It is one consistent instrument applied across 3,700+ answers. We have not validated it against hiring panel judgments, so treat comparisons inside the data as suggestive.

Is a 52 out of 100 average bad?

It is lower than most people expect from prepared candidates, and what it grades is one spoken answer at a time against what that round is asking for. Most first attempts leave out evidence the person clearly has. The same candidates who came back for a second attempt at the same question averaged 68.2 on it, up from 60.1.

Does this data prove practice interviews help you get hired?

No, and it should not be cited that way. Hiring outcomes are not tracked, so nothing here links a score to an offer. The 264 retry pairs show that scores on the same question by the same person rose 86% of the time after feedback and fell 14% of the time. That is evidence about answer quality and nothing wider.

Why do experienced people score lower than they expect?

The habits that make somebody good at the job work against them in the room. Saying "we" is correct on a team and unreadable to a panel, declining to name numbers reads as modest at work and as unverifiable in an interview, and leading with what you have not done is honest in conversation while functioning as an argument against yourself in a twenty minute round.

What kinds of candidates are in this sample?

Working professionals preparing for a specific named interview, usually within days of it. The occupational mix of the 691 sessions was never queried, so the findings cannot be broken down by field. What can be said is that the paying customer base concentrates in nursing and healthcare, education, government and public safety, management, and finance more than in software.

What to do with a number like 52.1

The uncomfortable implication of a dataset like this one is that the preparation most people do cannot reach the thing that is actually broken. You already have the example and you already know the number. The step that has not happened yet is you saying it out loud, in order, to somebody who asks a follow-up when the answer goes thin.

You can put one of your own answers through the same scoring and read back what it says you left out. Whatever comes back will be about how you said it, which is the half of this that is still yours to change.

AM

About the author

Andre Mendes — Founder, Mock Interview Pro

André is the founder of Mock Interview Pro, where job seekers run realistic mock interviews and get coach-level feedback on every answer.