Interview prep
How to practice interview answers out loud
Reading answers silently does not prepare you to speak them. Three short passes do: a skeleton, a clock, and a proof check.
Rehearse the structure, then the clock, then the evidence
Say your answers out loud, three times, and check something different each time: first the skeleton of the answer, then how long it takes, then whether it names one real thing from your resume. Do not memorize sentences. Reading an answer silently feels like practice, but the interview tests speaking: starting without a script, choosing what to leave out, and stopping.
The usual failure is not a bad answer. It is a good answer that runs four minutes, circles back twice and never says what you personally did. Recording yourself is common advice; what this guide adds is what to check on the recording: a time budget by question type, one checkable detail, and a rule for when to stop.
Pick your five most likely questions and do all three passes on those, rather than one pass on twenty.
Pass one: write the skeleton, not the script
For each question, write four lines and no more: the situation in one sentence, what you personally did, the result, and the one thing you want the interviewer to remember. Then cut those lines down to five cue words on a card and say the answer from the cue words.
Here is an invented example for "Tell me about a time you pushed back on a deadline":
Situation: launch date set before the design review. Did: showed the two flows that would break and proposed cutting the second one. Result: shipped one week later with the first flow solid. Remember: I bring a cut, not just an objection. Cue words: review, two flows, cut, week, solid.
The skeleton holds a decision and an outcome, not phrases. Said aloud twice, it will come out in different words each time, which is the point. A memorized paragraph falls apart the moment the interviewer interrupts. A skeleton survives, because you only need to find your place. Test that: have a timer or a friend cut you off at 45 seconds, then restart from the cue words without repeating yourself. It counts only if you still land the result in the time left.
If you cannot fill the "what I personally did" line, the answer is not ready, and no amount of delivery practice fixes that. Pick a different story.
Pass two: put each answer on a clock
Start a timer, answer, stop it. Compare the time with a budget for that kind of question. The budgets in the table below are working defaults from experience, not research findings, and interviewers vary, so adjust them if an interviewer tells you what they want.
Speaking pace varies a lot by person, and nerves change it, so measure your own: read 150 words of anything aloud with a timer and note the time. At the 140 words a minute used in the table below, 90 seconds is about 210 words, and a 400-word answer runs close to three minutes. Use your own rate instead.
Over budget? Cut the setup, not the result. Most overlong answers spend the first minute on context the interviewer did not ask about. Far under budget? You probably skipped what you did. Stopping on time matters because an interviewer can always ask for more, and finds it awkward to ask you to stop.
Choose your last sentence in advance, usually the result, so you land instead of trailing off.
Pass three: check the proof and count the fillers
Record the answer and listen once, asking two questions. First: did you say one specific thing a recruiter could check against your resume, such as the project, the tool, or a number that is true on the page? "Improved the process" does not pass. "Cut the review loop from two rounds to one" does, if it is true. Second: how often did you say "um", "like" or "you know"?
Do not aim for zero fillers. They fill silence, and a one-second pause reads as thinking. As a working default, not a research finding: if the same filler opens most of your sentences, or you hear one about every ten seconds, drill the pause. Stop, breathe, then start the next sentence. You are practicing the pause, not the absence of a word.
Use this stop rule: move to the next question when two takes in a row land inside the budget and name the proof point. Recording past that point usually makes the answer worse, because you start polishing sentences instead of keeping the skeleton loose.
The proof check also catches a quieter problem, an answer that sounds fine but claims more than your resume supports. If you describe a project differently out loud than it appears on the page, fix one of them before the interview. Interviewers often have your resume open next to your answer.
Running the passes in Lamr's Prep tab
If the role is saved in Lamr, you do not have to invent the question list. The Prep tab generates about 8 to 10 likely questions for that posting (2 credits per run), each with why it is likely and a suggested angle from your real experience. The angles lean on the gaps found when your documents were tailored, so an application that was only screened has no gaps to use yet. For the questions about requirements you do not meet, see how to answer interview questions about a requirement you don't meet.
From the same tab you can answer a question out loud. Practicing is free. Your browser turns speech into text (in Chrome and Edge; elsewhere you type the answer), and Lamr receives only the text. No audio is stored. It shows how long you took, your words a minute and a count of filler words, all counted by code rather than judged by a model. For 1 credit you can also ask for written feedback grounded in your resume: what to keep, what to change, and one thing from your resume the answer left out. That last line prompts pass three, but you still need to listen for whether you named one checkable detail, and treat the filler figure as a count, not a verdict.
Two limits. It is not a live conversation, and Lamr does not talk back beyond optionally reading the question aloud. And speech-to-text is not perfect, so treat the counts as a guide, not a score. You can save and track the role in the application tracker first, then practice from its Prep tab.
What to stop worrying about, and the night before
- A flawless delivery. A clear answer with a small stumble beats a smooth one that never says what you did.
- Every filler word. Cut the habit that you can hear without trying, and leave the rest.
- Memorizing "tell me about yourself". Keep a skeleton: where you are now, the one or two experiences that matter for this role, and why this role.
The night before, run your three hardest questions once each, in order, at the budget. Then stop. A fourth take is worth less than sleep.
Quick answers
How many times should I practice each answer?
Until two takes in a row land inside the time budget and name the proof point. Expect three to five takes for a new answer and fewer for a story you have told before; that is a working default. More than that usually adds polish you do not need and takes the life out of the answer.
Should I memorize my answers?
No. Memorize the skeleton and a few cue words. Word-for-word answers break when the interviewer interrupts or rephrases the question.
Is it normal to feel awkward talking to yourself?
Yes. The first few takes tend to feel stiff and slow. Do the first take on an easy question you will not use, so the stiffness is spent there.
Does Lamr store my audio?
No. Your browser turns speech into text and Lamr receives only the text. Speech to text works in Chrome and Edge; in other browsers you type the answer.
Can Lamr run a live mock interview?
No. Practice in the Prep tab is one question at a time. Lamr shows your time, words a minute and filler count, and can give written feedback on an answer, but it does not hold a live conversation.