# Mock interviews: rehearsing thought, not answers

> A mock interview rehearses three things at once: an unfamiliar question, a run of follow-ups and what you do with a hint. Here is how to run one and fix it.

[Formatted version of this page](https://www.frontiervue.com/interviews/guides/mock-interview)

A mock interview is only worth doing if three things are in the room at once: **an unfamiliar question, a run of follow-ups, and a hint handed over halfway**. Remove one and you are doing problems, not interviewing. This page builds that condition: **how to practise alone** with a recording and a timer, **who to practise with** and what each person is good for, the **six stages of one forty minute mock** with the four instructions for whoever plays interviewer, and the step most people skip, which is **turning feedback into a three column log and then into a behaviour you can run next time**.

## Key takeaways
- One question tells you whether a session counted: were unfamiliarity, pursuit and a hint all present? If one was missing, you were doing problems.
- Practising alone works, but only if you pay three costs: video rather than audio, a hard timer rather than free time, and three passes of playback for sentences, posture, and how well you took the hint.
- A practice partner is worth what they pursue, not what they are called. A subject teacher gives depth, a parent or classmate tests whether you make sense, and someone outside your subject is best at following up.
- Fix at most three things per mock, and write each as something countable. Feedback you cannot attach a timestamp to on playback is not feedback.
- The five dimension rubric in our own Frontier ACE, snapshot dated 22 October 2026, is useful as a self review sheet you reuse across sessions.

### How to do a mock interview, in six lines

| Item | Value |
| --- | --- |
| What is rehearsed | Unfamiliar task, pursuit, and a hint, all at once |
| Practising alone | Record on video, use a timer, play it back three times |
| Choosing a partner | Judged by whether they follow up, not by status |
| One 40 minute mock | Warm up, unfamiliar task, statement pursuit, why this course, reverse questions, close |
| Test for feedback | Can it become a countable behaviour next time? |
| What a practice score means | It describes your own session, it predicts nothing |

> The durations and intervals here are practice advice. Dates and formats follow your invitation email.

## Contents
1. What a mock interview is actually rehearsing
2. Practising alone: video, timer, three passes
3. Practising with people: who suits which layer
4. One 40 minute mock: the six stages and the clock
5. Turning feedback into the next behaviour
6. Frequency, pacing, and what to change near the day
7. Three mocks that look like practice and are not

## 1. What a mock interview is actually rehearsing

Three things happen at once in a live interview, and none of them survive a revision desk.

The first is **unfamiliarity**. You have not seen the task and you cannot choose it. You may not even know which topic it belongs to, because interviewers deliberately work the seam between what you have covered and what you have not.

The second is **pursuit**. Every sentence you offer gets picked up and pressed one layer down: why that thought, what the conclusion depends on, what changes if you alter the case. Pursuit is not a trap; it is a test of whether you hold reasons or only impressions.

The third is **the hint**. When you stall, the interviewer hands you a direction and then watches whether you take it, use it, and can say what you changed because of it. Oxford describes the interview as an academic discussion about your subject; Cambridge states it assesses understanding, critical thinking and whether you can be taught to a higher level.

So a mock is not rehearsal for difficulty. It is rehearsal for how much thinking survives when all three pressures are on at once, and solving problems alone removes all three: you pick the question, nobody follows up, nobody redirects you halfway.

- **Unfamiliar:** Not seen, not chosen (Use a question outside what you have rehearsed)
- **Pursuit:** Every sentence gets one why (Practise giving the reason, not repeating the claim)
- **The hint:** Someone changes your direction (Then name what you changed because of it)
- **Out loud:** The idea has to land before they can follow (Writing it is not the same as saying it)

> Pick the questions you cannot tell whether you can do. If the first move is obvious from the wording, the problem is rehearsing fluency, not interviewing; save those for the day you just want to warm up.

## 2. Practising alone: video, timer, three passes

What solo practice lacks is pursuit, not company. So all the technique in practising alone goes into **putting the follow-up back in**, and the cheapest way to do that is not another person but a recording that plays the silent second role.

There are three layers you can add: **a timer** (under a minute to think, five to speak, and you stop when it says stop, because without a clock people linger exactly where they are already strong), **giving yourself the hint** (change the numbers, drop the assumption, reverse the case; whether you do this decides whether the session was a mock), and **three passes of playback**, unpacked below.

1. **Record it, on video** (Not audio alone)
On a video call, the lens, your eye line and the angle of your paper are all observable, and an audio recording cannot show any of that. A front camera is enough. Set the paper where it can be seen, which doubles as rehearsing the room.

2. **Fix the clock** (One minute, five minutes)
No more than a minute from reading to first words, then five to eight minutes of structured talking. Stop even if you have not finished, because an interview will not give you unlimited time or skip the next layer of pursuit.

3. **Three passes of playback** (One question per pass)
Pass one, listen without watching: are the sentences complete, or do half-clauses hang unfinished? Pass two, watch without listening: eye line at the lens, posture closing up, whether your hand blocks your face while writing. Pass three, watch normally and count two things only: how often you stated an assumption, how often you changed position after a challenge.

4. **Write one change down** (One per run)
When playback ends you are not allowed to write "practise more". Write one concrete rule instead, such as: next time the first sentence must contain "I will assume that". One countable rule beats ten impressions.

5. **Next day, same problem, new condition**
Rerun the previous problem with the condition altered, and check whether your new rule actually fired. Repeating a problem is the highest-yield drill you have: the task costs you no comprehension, so all attention goes on yourself.

> Cambridge’s own preparation advice gives a very usable shape: ask a teacher, or someone who does not know you well, or question yourself for half an hour. Practising alone is not a fallback; it is one of the three legitimate forms.

## 3. Practising with people: who suits which layer

A practice partner is worth the depth they push to, not the title they hold. The only right question is: after I finish speaking, can this person press one layer further?

On that standard the four kinds of partner suit completely different layers, and using the wrong one wastes a session; the split is in the table below.

If your school runs mock interviews, go to one. Imperial’s own interview advice tells applicants to find out whether their school offers mocks and to attend if it does, and the reason is structural: a school can put several adults in a row pursuing the same thing, which is very hard to reproduce at home.

|  | Practice partner | The layer they are best for | Where it tends to go wrong |
| --- | --- | --- | --- |
| Subject teacher | Depth, and being pushed past the syllabus edge | Knows you too well, so it drifts into a lesson check; rarely interrupts |
| Someone who has read the subject | Where this subject really goes, and how to talk about your reading | The temptation to hand you their old questions, which bends practice towards guessing |
| A parent | Making it intelligible: can a lay ear follow your first move | Cannot pursue the subject matter, and is easily convinced by you |
| An adult outside your subject | Sustained pursuit: why, and what if it were the other way round | Once it reaches technical depth they have no direction to give, so it works for the opening half |
| A classmate also applying | Cheap, frequent, and someone to time and review you | Two people nodding at each other; the weakest pursuit pressure of the five |

> Choosing by who can push one layer further beats choosing by title. The sturdiest pairing is two mocks: one with someone from the subject for depth, one with an outsider for clarity.

## 4. One 40 minute mock: the six stages and the clock

1. **Warm up** (Minutes 0 to 5)
The interviewer asks two things only: why this subject, and one thing you have read or done that belongs to it. The stage has one purpose, to get your voice into the vocabulary of the subject. Nothing is scored and nothing is pursued.

Cambridge describes its interviews as “academic conversations about the subject you’re interested in studying”, and this stage rehearses the first minute of that conversation.

2. **One unfamiliar problem** (Minutes 5 to 17)
The interviewer picks it and you see it only in the room, aimed at the seam between what you have covered and what you have not; one problem from the [original subject questions](/interviews/questions) is enough. Three rules: **open with an assumption**, **if you stall past sixty seconds say where you are stuck**, and at about minute twelve the interviewer hands over one hint, after which you name what you changed because of it.

3. **Pursue the personal statement** (Minutes 17 to 25)
The interviewer takes one sentence from your statement and presses three layers down: **which part of it you actually did**, **why you did it that way**, and **what you would change if you did it again**. Imperial’s own advice starts with rereading the personal statement, because it is the material most likely to be picked up, and a sentence that will not hold shows here first.

4. **Why this course** (Minutes 25 to 30)
One sentence of reason first, then one pursuit: **would this reason hold equally at another university**. An answer that survives gets specific about a course, a way of teaching or a direction; one that cannot name anything specific will not survive the real room either.

Do not prepare this as a script. Rehearse the same meaning at three lengths: one sentence, one paragraph, one paragraph with an example.

5. **You ask the questions** (Minutes 30 to 35)
Now you ask. In this stage the interviewer **asks nothing back and only answers**, so what is being rehearsed is the ability to raise a real question: how the subject is taught, what the first year projects look like, which parts you can choose. Most people discover on their first run that they cannot ask anything, which is the most useful information the stage produces.

6. **Close: timestamps only** (Minutes 35 to 40)
The interviewer reads back the timestamps written on the sheet, offering no verdict and no solution; you pick one of them and write it as the rule to be tested next time. Stop at forty minutes. **A mock that overruns never rehearses a close**, and the close is the only thing that gives the next session a direction.

Three objects run the whole session: **a problem you see only in the room**, **one A4 sheet for the interviewer** (the six stage times in the top left, blank space on the right for timestamps), and **a printout of your own personal statement**. The timer sits on the interviewer’s side, the recording starts at the warm-up, and that sheet may hold times and nothing else.

Difficulty has exactly three dials, which is enough for a dozen sessions: **unfamiliarity** (a further chapter of your own subject, or a different subject), **hint strength** (a direction, a counter-question, or nothing), and **how many layers of pursuit** (two or four). Move one dial per session. Move two and the next playback cannot tell you which one did the work.

If the real interview is online, run the same sequence twice in front of a lens: once writing on paper lifted toward the camera, once on a shared screen or whiteboard. Cambridge’s own online checklist includes practising a stylus and tablet for annotating a shared document or whiteboard when a College says you will need one, and a College may also hold a tech check first.

### 1. Pursue, do not judge

Before the closing five minutes, sentences like “not bad” or “that went wrong here” are not allowed. Only three are: **why that thought**, **what if the condition were the other way**, **which of your last sentences are you least sure of**.

Verdicts wait until the timestamps, and the solution stays in the interviewer’s pocket. The moment they give one, the mock becomes a lecture.

### 2. Say so when you lose the thread

The interviewer does not need to be a specialist in the subject. When they lose the thread they should say so plainly: “I lost that part. Say it again from your last sentence.”

This is the most valuable line in the whole brief, because it forces you to retell one thing in a second way, and **the second way** is the ability an online interview actually spends.

### 3. Keep to the stage clock

The six stage times are on the sheet and you switch when each runs out, even mid-sentence when it is going well. In a real interview the other person sets the pace, so one of the most useful things to rehearse is **how to stop a sentence cleanly when it is cut off**.

Only one stage may borrow time: if the unfamiliar problem is still being pursued at minute seventeen, it may take five minutes from the reverse question stage.

### 4. In the question round, only answer

During the five minutes of reverse questions the interviewer answers the question and nothing else, without adding “and what do you think about that”. This stage rehearses **asking a real question**, and any counter-question turns it back into a second round of subject pursuit.

If a question stumps them, they answer honestly, and that answer becomes part of what you review afterwards.

> The two stages people cut are the fifth, reverse questions, and the sixth, the close. They are the two that give the next session its direction: a new problem measures what you know, the close measures whether the last session worked. The length matches too, since Cambridge says most applicants have one or two interviews totalling about 35 minutes to an hour, and forty minutes sits inside that.

## 5. Turning feedback into the next behaviour

The real output of a mock is not a score. It is a short list of changes that can be executed next time, and most feedback dies before the second session for one reason: it was never written as an action. Sentences like “you sounded nervous” or “make it clearer” cannot be checked on any playback, because there is nothing in them to check. Working feedback has three fixed shapes: **a countable rule**, **a locatable timestamp** (at 04:12 you went silent for eleven seconds), **a condition you can replay** (same pursuit, different problem).

One discipline is easy to miss: **never more than three changes at once**. Someone handed five or six adjustments after a mock keeps none of them, because no single one gets tested on its own.

On paper those disciplines become one row with three columns: **what was observed**, something you could show a stranger on the recording; **the explanation**, which is allowed to be wrong; **the action for next time**, which has to be countable. They stay in separate columns because they must never be merged. Almost all useless feedback is a second column written without a first one, and “you rush” is an explanation with no timestamp behind it. Keep the same sheet across sessions, since the action column from last time is the only line worth checking next.

Then there is dependency. A partner supplies **direction, not verdict**: the judgement stays with your own playback and your own sheet. By the third mock you should be able to write the two observable facts yourself. If every session still needs somebody else to tell you what went wrong, the practice is training nothing independent.

|  | Log entry | What was observable | Your explanation | The action next time |
| --- | --- | --- | --- | --- |
| The silence at 04:12 | Twenty three seconds with no sound from 4:12 | Probably lost the task, possibly just deciding the next move | Name the stuck point once it passes sixty seconds; count how often it happens next time |
| The hint at minute twelve | The hint was repeated in its own words, then the original plan carried on unchanged | Not a failure to hear it, but a method decided before it arrived | After a hint, say “because of that I am changing” before taking two steps |
| The project in the statement | Ninety five seconds containing four “we” and no “I” | Could be nerves; more likely no sentence ever built for the part you actually did | Write three sentences whose subject is “I”, then count them on the next playback |
| Why this course | Ninety seconds that named no course, no teaching method, no direction | The reason is too large to be specific, or the syllabus genuinely has not been read | Rewrite it to survive three layers of pursuit; say it once next time with no extra length |

> The table carries across sessions: the action column from last time is the only line you check this time. If the observable column is empty, do not write the explanation yet.

### Adjective feedback

- Wrong: “Look more natural. More confident.”

- Right: “Next time no answer opens with ‘I think’. Open with what you are going to test.”

- How to spot it: Test it: can you count this on the next playback? If not, rewrite it.

### Judging content only

- Wrong: “You did not solve number three.”

- Right: “On number three you spent four minutes calculating before naming an assumption.”

- How to spot it: Test it: does it name an observable action? A verdict on the result alone cannot steer the next run.

### Ten changes at once

- Wrong: “Fix the pace, the structure, your eye line, your examples and your hands.”

- Right: “This session changes one thing: assumption before speaking. We add another next time.”

- How to spot it: Test it: more than three changes in one go means none of them will be checked separately.

### Impressions without timestamps

- Wrong: “The second half felt messy.”

- Right: “Between 08:20 and 09:05 you said the same thing three times.”

- How to spot it: Test it: no timestamp means you only felt it was bad, so next time it will be bad in the same place.

### Asking the partner for the model answer

- Wrong: “What was the right approach? I will copy it.”

- Right: “Where did my reasoning break, and at what second?”

- How to spot it: Test it: if a model answer for the problem ends up in your notes, the mock has become recitation.

### The same answer, before and after one round of feedback

The demonstration task (an original practice question written by us): a block of ice floats in a glass of water at room temperature and melts slowly. Someone says, “the bigger the ice, the slower it melts, so the rate is a property of the ice”. Say how you would test that claim rather than giving a verdict.

Both answers below were written by us to show what one round of feedback can do. The first is roughly what a first session sounds like; the second is the same task after exactly one change.

1. **Round one, and what it left behind**
“Um … I think roughly, the bigger the ice, the smaller its surface is against its volume, so heat gets in more slowly, it is actually about heat transfer, so a big block just melts slower. Sorry, I am a bit all over the place, that is about it.”

Only two things go into the observable column: **nineteen seconds of silence between the end of the question and the first words, with nothing said in it**, and **seventy four seconds containing no assumption and closing on “that is about it”**. Anyone watching the recording can point at both.

2. **One action for this session**
The rule is written so it can be counted: **the first sentence of every answer must contain “I will assume” or “first let me see whether it changes under the same conditions”**. At the next mock you count six answers and four hits is a pass. Nothing else moves: no posture, no pace, no extra heat transfer theory, and no model answer for this task.

3. **The next day, same task, condition changed**
Swap room temperature water for salt water at the same temperature and keep everything else. The repeat exists for one reason: to check whether the rule entered behaviour. If the first sentence is again “I think roughly”, it has not, so run the same rule again instead of adding a new one.

**Answer:** The same task in round two (again written by us): “I will assume the water stays about the same temperature, because the glass is much larger than the ice. Then any difference has to come from the ice itself, so first I want to see whether the rate changes as the ice gets smaller while everything else is held: if it does, size is a real clue, and if it barely does, I need to look on the water side instead.”

Same task, one change: the opening went from nineteen silent seconds to a stated assumption, and the ending went from “that is about it” to a comparison that can actually be tested.

> That is the size of one round: a countable rule plus one repeat under a changed condition. Two rounds will not make you a different person, and neither answer here is a script to memorise. The test stays the same: **if a third person cannot tell within two minutes whether you did it at the next mock, it was not feedback.**

> Our own practice product Frontier ACE carries a five dimension rubric you can copy straight into your log sheet (snapshot dated 22 October 2026): **technical accuracy 35, structure 20, relevance 20, engagement 15, delivery 10**, and for a puzzle or calculation task **45, 15, 20, 10, 10**. Score every session with it and the next one has the same ruler.

## 6. Frequency, pacing, and what to change near the day

### More than a month out: one full mock a week

At this stage you are building structure, not accumulating sessions. One full forty minute mock a week, with two or three fifteen-minute solo runs, is enough; the rest of the time goes on your own subject material and the edges of the syllabus.

Weekly spacing is the point: the previous rule gets tested at the next mock instead of being buried under new content.

### The last two weeks: twice a week, conditions changed

Now change the three variables a live interview actually changes: **unfamiliarity** (cross-subject, chosen at random by the other person), **hint strength** (sometimes one line of direction, sometimes a counter-question first), and **format** (record on video if yours is online, use an unfamiliar room if you are attending in person).

In these two weeks hand the pursuit over to someone outside your subject, because whether you can make it intelligible is closer to the real room than whether you can get it right.

### The final week: short mocks only, no new sources

Switch to ten or fifteen minute mocks that rehearse only the opening three minutes and one pursued problem. New material in this week has low yield and high cost: one problem you cannot shift will directly tax the mindset you need on the day.

Rereading your own notes, the three rules on your old sheets and the problems you already reran beats three fresh ones.

### 48 hours out: stop unfamiliar problems

Inside 48 hours, do three things only: **equipment and room** (camera, microphone, connection, quiet, paper and pen, the angle of your writing on camera), **the opening three minutes** (say your reasons for the subject once, to a lens), and **body clock** (move your alertness into the interview slot).

The only thing an unfamiliar problem produces this late is the thought that you still have gaps, and that thought is a liability in the room.

### Several interviews in a row: change content, keep process

Many Oxford applicants have more than one interview and some are seen by different Colleges; at Cambridge most have one or two, roughly 35 minutes to an hour in total. When interviews sit one or two days apart, do not invent a new approach for each.

Keep the six-stage process fixed and swap only the subject material, so that in the second interview you still open with assumption then next step rather than improvising a technique.

> A rough but workable ratio: with four weeks or more to go, full mocks and solo runs sit at about one to three. In the final two weeks reverse it, so full mocks dominate. The closer the day, the more your practice should look like the event rather than add up to more of it.

- **The online room:** A quiet space, device on full battery (Test camera, microphone and connection beforehand)
- **Paper, pen, writing:** Keep paper and pen ready; use only moves you have rehearsed (A College asks you to practise a stylus and tablet only when it says you will annotate a shared document or whiteboard, and it may run a tech check first)
- **Contact details:** Keep the College’s phone number and email to hand (Raise anything wrong before the interview; reported afterwards it usually cannot count)
- **The in person trip:** Find the location on a map, then plan the journey (Arrive on time, phone on silent)
- **The dates you protect:** Cambridge 7 to 18 December 2026, Oxford in December 2026 (Cambridge asks you to keep the period free and cannot usually reschedule for personal reasons)
- **When to schedule the mocks:** Before the invitations land (Oxford sometimes gives about a week’s notice, and that week closes mechanics rather than hosts a first recording)

## 7. Three mocks that look like practice and are not

**A mock that does not work**: Three common shapes

- The rehearsed mock: Questions shared in advance, answers memorised
  - Sign: not one second of hesitation, and never a stated assumption
  - Cost: the first real question sends you back to zero
  - Fix: your partner picks it and you see it only in the room
- The smooth run: No stall, no pursuit, from start to finish
  - Sign: your partner only nods when you finish
  - Cost: you rehearsed delivery inside your comfort zone
  - Fix: after every answer, one demand for why you think so
- The hint-free mock: Only complete tasks; jump on when you stall
  - Sign: no moment anywhere where you took a direction
  - Cost: never rehearsing the coachability they watch for
  - Fix: after sixty stuck seconds a direction is handed over

All three share one fault: they turn interview practice into saying out loud what you already know. The value of a real interview sits outside that part, and both universities describe it in the same place: they are watching how you handle the unfamiliar and whether you can use a hint to move forward.

So judge a mock not by whether you performed well but by whether three things happened: **did you genuinely stall**, **did you take a hint and say what you changed**, **did you leave with one rule you can count next time**. If none of the three occurred, that session rehearsed comfort.

One warning deserves its own line: do not treat practice tasks as a list to memorise. Every problem on this site is written by us as original practice, has no connection to any university’s unpublished interview material, and cannot be predicted. What can be rehearsed is your first sentence on an unfamiliar task, and the three moves after it.

## Questions
**Do I need another person for a mock interview?**

No. What is missing is pursuit, not company. Solo, a recording plus a timer with you playing the hint-giver covers most of the work, and Cambridge’s own preparation advice explicitly allows questioning yourself for half an hour. Another person helps because the pursuit is less polite, so the sturdiest pattern is mostly solo practice with one live mock kept each week.
**How long should one mock interview last?**

Run the full session as specified here: forty minutes, six stages. For scale, Cambridge says most applicants have one or two interviews totalling about 35 minutes to an hour, so forty minutes sits inside the real thing and is still enough for an unfamiliar problem, one pursuit of your statement and a round of questions back at the interviewer. If time runs short, cut the warm-up and the chat, never the last two stages.
**How often should I run a mock interview?**

With four weeks or more, one full mock a week plus two or three fifteen-minute solo runs is enough; raise it to twice a week in the final fortnight; in the last week switch to ten to fifteen minute mocks and open no new sources; inside 48 hours rehearse the opening and the equipment only. Frequency should be set by whether your last rule gets tested next time, not by how much you covered.
**My teacher does not know my subject. Is the mock still useful?**

Yes, and some layers only that kind of partner can train. Someone outside the subject will ask only why, and can you say it in plain terms, which are the two things that bite hardest in an interview: whether your reasoning has grounds, and whether anyone can follow it. If a lay listener tracks your first move, you explained it. Find someone who has read the subject for the depth session.
**Is a school mock interview worth doing?**

Yes. Both Imperial and Cambridge tell applicants to rehearse, and Imperial specifically says to check whether your school offers mock interviews and attend one. A school can assemble several different staff pressing the same thread, which is hard to reproduce at home; its weakness is familiarity and a reluctance to interrupt. Take the feedback from it and feed it back into your own six-stage loop.
**If I do badly in mocks, does that mean the interview will go badly?**

That inference does not hold. Doing badly is a normal product of practice: the value is that stalling, missing a hint and failing to explain surface in a recorded session rather than in the interview room. The case to worry about is the opposite one, where every mock goes smoothly and you never stall. That kind of mock carries no information.

## Official sources
- [University of Cambridge: how to prepare for your interview](https://www.undergraduate.study.cam.ac.uk/apply/after/interview-preparation)
- [Imperial College London: undergraduate interviews (usually November to February)](https://www.imperial.ac.uk/study/apply/undergraduate/process/interviews/)
- [University of Oxford: a guide to undergraduate interviews](https://www.ox.ac.uk/admissions/undergraduate/applying/guide-for-applicants/interviews)

## Related pages
- [Oxbridge interview hub](https://www.frontiervue.com/interviews)
- [Oxbridge interview mistakes and tactics](https://www.frontiervue.com/interviews/guides/oxbridge-interview-tips)
- [Oxford interviews](https://www.frontiervue.com/interviews/oxford)
- [Cambridge interviews](https://www.frontiervue.com/interviews/cambridge)
- [Original subject questions to pick from](https://www.frontiervue.com/interviews/questions)
- [Mathematics interview questions](https://www.frontiervue.com/interviews/questions/mathematics)
- [How Imperial interviews are arranged](https://www.frontiervue.com/interviews/imperial-college-london)
- [Thinking aloud under pressure](https://www.frontiervue.com/interviews/guides/interview-thinking-aloud)
