Interview statistics: what 2,088 practice interviews show about how people answer
These are our own numbers: 2,088 mock job interviews that 1,822 people ran in Colloqa between May and November 2026 — how long they lasted, how long people talked per answer, which skills the AI interviewer rated lowest, and how many sessions ended “interview ready”. They are practice sessions, not real hiring decisions, and every figure says what it is based on.
Free: 1 full practice interview a day, no card.
Key figures
Data as of November 2, 2026. Sample sizes are under each figure.
- 9%of practice interviews ended “Interview Ready”61% “On Track”, 30% “Keep Practicing” — the AI interviewer’s verdict on 730 sessions with feedback.
- 75 wordsmedian length of an answerAbout half a minute spoken. A quarter of answers were under 40 words. Based on 6,482 answers.
- Depth & structurethe lowest-rated skillsDepth 51 and structure 52 out of 100; ownership (64) and relevance (63) scored highest. 142 sessions.
- 18 minmedian length of a finished practice interviewHalf took between 12 and 27 minutes. 759 finished sessions.
- 49%pasted the real job posting before practicingOf 2,088 sessions. Those with a posting ended “Interview Ready” more often: 10% vs 6% (730 sessions with feedback).
- 3.6%of behavioral interviews ended “Interview Ready”The lowest of any interview type (mixed 9.5%). 140 behavioral sessions with feedback.
Where these numbers come from
Colloqa is an AI mock interview app: you set up the role, level and interview type (and, if you have one, paste the job posting), then answer an AI interviewer out loud or in writing. It asks follow-up questions on your answers and, at the end, gives a verdict and feedback. Every session is stored, so we can count across them.
The sample: 2,088 practice interviews by 1,822 people, May 2 – November 2, 2026, with our own test accounts removed. 759 of the sessions were finished — the person reached the end and got feedback; the rest were left part-way, which is normal for practice. Most sessions (92%) were in English. Figures about answers and feedback use finished sessions only.
How long people talk per answer
The median answer was 75 words; half of all answers fell between 38 and 134 words. At a normal speaking pace, 75 words is roughly 30 seconds — on the short side for a “tell me about a time…” question, which usually needs one to two minutes to cover the situation, what you did and the result.
| Answer length | Share of answers |
|---|---|
| Under 15 words | 8.8% |
| 15–39 words | 17.0% |
| 40–200 words | 62.1% |
| 201–300 words | 8.1% |
| Over 300 words | 4.0% |
6,482 answers in 759 finished sessions. Spoken answers are counted as transcribed, typed answers as typed.
Which skills scored lowest
At the end of a session the AI interviewer scores eight skills from 0 to 100. Averaged across sessions, the weakest were not confidence or motivation but the shape of the answers: depth, structure and clarity.
| Skill | What it measures | Average score |
|---|---|---|
| Depth | Real command of the subject: reasoning about trade-offs, causes and edge cases, not just naming things | 51 |
| Structure | A beginning, a concrete middle and a clear ending; answering the question that was asked | 52 |
| Clarity | How easy the answer is to follow: short sentences, plain words | 52 |
| Specificity | Facts, numbers and named situations instead of general claims | 54 |
| Motivation | A reason for wanting this role, not just a job | 57 |
| Confidence | Committing to an answer and staying composed under follow-up questions | 58 |
| Relevance | Whether the answer matches what this role and level need | 63 |
| Ownership | Owning decisions and mistakes — “I decided”, not “it was decided” | 64 |
142 sessions by 116 people, scored since June 29, 2026 (when skill scores were added). AI ratings on practice answers — a measure of the answers, not a hiring prediction.
How often a session ended “Interview Ready”
The verdict has three levels: “Interview Ready” (would likely pass this interview), “On Track” (a solid base, a few things to sharpen) and “Keep Practicing” (real effort, clear areas to work on). Below, by how the session was set up.
| Session | Sessions | Interview Ready | Keep Practicing |
|---|---|---|---|
| All sessions with feedback | 730 | 8.6% | 30.4% |
| Pasted the job posting | 447 | 10.1% | 27.1% |
| No job posting | 283 | 6.4% | 35.7% |
| Behavioral interview | 140 | 3.6% | 27.9% |
| Mixed interview | 538 | 9.5% | 28.6% |
| Technical interview | 52 | 13.5% | 55.8% |
| Junior level | 343 | 8.2% | 25.7% |
| Mid level | 292 | 7.5% | 34.2% |
| Senior level | 95 | 13.7% | 35.8% |
These are differences between groups, not causes: people who paste a job posting may also be further along in their preparation. Technical and senior rows are small samples.
Which jobs people practice for
The role each person entered for their latest session, grouped by keyword. Customer service, administrative, software and hourly retail or food-service jobs lead; nearly a third of people practiced for something outside these groups — from flight attendant to electrician.
| Role group | People | Share |
|---|---|---|
| Administrative & assistant | 187 | 10.3% |
| Software & IT | 182 | 10.0% |
| Customer service & call center | 181 | 10.0% |
| Retail, food service & hourly | 169 | 9.3% |
| Healthcare & nursing | 148 | 8.1% |
| Finance, accounting & analysis | 129 | 7.1% |
| Sales | 107 | 5.9% |
| Management (other) | 70 | 3.9% |
| Marketing | 46 | 2.5% |
| Teaching | 31 | 1.7% |
| Other roles | 536 | 29.5% |
| Not specified | 31 | 1.7% |
1,817 people. Most sessions were set up as a mixed interview (65%), then behavioral (23%) and technical (12%); 48% chose junior level, 40% mid, 13% senior.
What this means for your own preparation
Practice data cannot tell you what a specific employer wants, but the same gaps came up again and again.
- Give your answers more room. A quarter of answers were under 40 words — too short to show any evidence. For a “tell me about a time…” question, aim for one to two minutes: the situation in a sentence or two, what you did, and what happened. Behavioral questions and the STAR method →
- Fix the shape before the content. Structure and clarity scored lower than confidence. A simple frame — the point first, then one example, then the result — does more than rehearsing more examples. A structure for “Tell me about yourself” →
- Prepare for the second question, not only the first. Depth was the lowest-rated skill: answers that name something but cannot explain why or how when asked. For each example you plan to use, know the reasoning behind it. Competency-based questions with example answers →
- Practice against the actual job posting. Sessions set up with the real posting ended “Interview Ready” more often. Whether you practice with a tool or a friend, use the posting to choose the questions and the examples you will need.
- Spend the most time on behavioral questions. Only 3.6% of behavioral sessions ended “Interview Ready” — the lowest of any type. They are also where structure and specific examples matter most. How to answer “What are your strengths and weaknesses?” →
Method and limitations
Source: Colloqa’s session records, May 2 – November 2, 2026, queried on November 3, 2026. Accounts we identified as internal or test accounts (97) were removed. Only counts and averages were computed; no individual answer was read or quoted for this page, and nothing here identifies a person.
Definitions. A session is one practice interview. Finished means the person reached the end and the session was closed. Answer length is the word count of each answer as recorded. The verdict and the skill scores are produced by the AI interviewer at the end of a session; skill scores exist only for sessions since June 29, 2026. Roles are what people typed at setup, grouped by keyword, so the groups are approximate.
Limitations. These are practice sessions by people who chose to practice, mostly in English, not real interviews or hiring outcomes. AI ratings measure the answers given in practice, not whether someone gets hired. Differences between groups are associations, not causes.
We plan to refresh these figures every quarter. You are welcome to cite them — please link to this page as the source.
See where your own answers land
The fastest way to find your gaps is to answer real questions out loud and get told where the answer was thin. In Colloqa you set up the role and level, paste the job posting if you have one, and answer an AI interviewer that asks follow-up questions on what you actually said. At the end you get the same verdict and skill scores this page is built from.
Find out which of these gaps are yours — run a practice interview for your next job.
Start a practice interviewInterview statistics: FAQ
In our data the median answer was 75 words — about 30 seconds spoken — and a quarter were under 40 words. For most questions, especially “tell me about a time…”, one to two minutes is a better target: long enough for the situation, what you did and the result, short enough to leave room for follow-up questions.
In 142 scored practice sessions, the lowest-rated skills were depth (51/100), structure (52) and clarity (52); the highest were ownership (64) and relevance (63). The gap is usually how the answer is built and how far it goes when questioned, not confidence.
Of 730 practice interviews with feedback, 8.6% ended with the AI interviewer’s verdict “Interview Ready”, 61% “On Track” and 30% “Keep Practicing”. This is a rating of practice answers, not a hiring outcome.
The median finished practice interview in Colloqa lasted 18 minutes; half took between 12 and 27 minutes, with a median of nine turns from the interviewer and eight answers.
Practice sessions set up with the real job posting ended “Interview Ready” more often than those without it (10.1% vs 6.4%) and less often “Keep Practicing” (27.1% vs 35.7%). That is an association — people who paste the posting may also be better prepared in other ways — but it is consistent with the usual advice to prepare from the posting.
Behavioral interviews: only 3.6% of 140 behavioral practice sessions ended “Interview Ready”, against 9.5% for mixed interviews. Behavioral questions need a specific example with a clear structure, which is exactly where answers were weakest.
From 2,088 practice interviews run by 1,822 people in Colloqa between May 2 and November 2, 2026, with test accounts removed and only aggregate counts computed. You are welcome to cite the figures with a link to this page.
Yes. In Colloqa you run a full mock interview by voice or text with an AI interviewer that asks follow-up questions, and at the end you get the verdict and skill scores used on this page. The free tier is one full practice interview a day, no card.