To track your interview practice progress, keep a short log of every practice session, score each answer on the same simple scorecard, and count a few things you can measure: answer length, filler words, structure, specific results and how you handle follow-up questions. Review the log once a week and pick one fix for the next week; real progress is that fix holding, not one good score.
This guide gives you a practice log template we wrote, a per-answer scorecard, the metrics worth counting and how to measure them, and a 20-minute weekly review routine.
Why track interview practice at all?
Memory flatters. After a session you remember the answer that went well and forget the three that wandered. Without a record, you tend to practice the questions you already like and repeat the same habits, such as a long wind-up or a missing result, week after week.
A log fixes three things. It shows whether practice is changing anything. It tells you what to practice next. And it gives you an honest answer to "am I ready?", which our guide to how many mock interviews to do before a real interview ties to signals rather than a fixed number.
A practice log template we wrote
Keep one row per session in a spreadsheet or notebook. The columns matter more than the tool.
| Date | Session type and settings | Questions | Average score | Longest answer | Filler words | Fix of the day | Did the last fix hold? |
|---|---|---|---|---|---|---|---|
| Mon | AI mock, behavioral, mid level, 5 questions | Conflict, failure, ambiguity, initiative, leadership | 5.8 / 10 | 3:40 | 14 | State the result in one sentence | First session |
| Wed | AI mock, behavioral, mid level, 5 questions | Same competencies, new wording | 6.6 / 10 | 2:50 | 11 | Situation in two sentences | Yes |
| Fri | Friend, 30-minute rehearsal | 6 questions from the job description | Friend rubric: 3 of 4 | 2:20 | 9 | Pause instead of "so" | Yes |
The rows above are an example we wrote. Two rules make the log useful. First, compare sessions only when the settings match: a hard technical round and an easy behavioral round will not give comparable scores. Second, write one fix per session, not five. You cannot work on five things at once, and a single fix makes the "did it hold?" column meaningful.
A per-answer scorecard
Score each answer on the same 1-to-5 scale, whether the feedback comes from you, a friend or an AI. You do not need every row for every answer; pick the rows that match the question type.
| Dimension | 1 (weak) | 3 (acceptable) | 5 (strong) |
|---|---|---|---|
| Point first | The point appears only at the end, or never | The point appears after some background | The first sentence answers the question |
| Structure | No clear order | STAR (or another structure) is present but uneven | Clear structure; most of the time spent on what you did |
| Specificity | General claims ("I'm a team player") | One real example, few details | One real example with names of decisions, tools and trade-offs |
| Result | No outcome | Outcome stated vaguely | Measurable or clearly observable outcome, plus what you learned |
| Length | Over 3 minutes, or under 30 seconds | Slightly long or short | Fits the question; about 1.5 to 2 minutes for behavioral |
| Follow-up handling | Repeats the original answer or freezes | Answers, but with hedging | Answers the follow-up directly in the first sentence |
For the full method of scoring and getting feedback from other people, see how to get feedback on interview answers.
Which interview practice metrics should you track?
Scores are opinions. These metrics are counts, so they are harder to fool yourself about. Measure them from a recording and its transcript; our guide to recording yourself for interview practice covers the setup.
- Answer length. Time each answer. Common guidance for behavioral answers is about 1.5 to 2 minutes; our page on how long a STAR answer should be explains the split. If your longest answer keeps creeping past three minutes, how to stop rambling in interviews has specific cuts.
- Time to the point. How many seconds before your first sentence that actually answers the question? A long wind-up is one of the easiest habits to fix with practice.
- Filler words. Count "um", "uh", "like", "you know" and sentence-opening "so" in the transcript. Track the trend for the same type of question; a few fillers are normal speech.
- "I" versus "we". In the Action part of a behavioral answer, count how often you say what you did. A string of "we" hides your contribution.
- Specific results. Does each story end with a number or a concrete change? Mark yes or no per answer.
- Follow-up recovery. After each follow-up, mark whether you answered it directly or restarted the story.
- For coding rounds: time to a working solution, whether you stated complexity, whether you tested edge cases, and how often you went silent for more than about 30 seconds. Our guide to practicing thinking out loud in coding interviews covers that last one.
Choose three or four metrics that match your weaknesses. Tracking everything turns review into a chore, and you will stop doing it.
A 20-minute weekly review routine
- Look at the trend (5 minutes). Compare this week's average score and metrics with last week's, using sessions with the same settings.
- Check the fixes (5 minutes). Which fixes held for two sessions in a row? Retire those. Which keep returning? Those need a different approach, such as writing one strong answer by hand and saying it aloud several times in your own words.
- Find the pattern (5 minutes). Read the notes column. If "no result" or "too long" appears three times, that is next week's fix.
- Plan the week (5 minutes). Choose one fix, the sessions you will run, and one session with a person if your interview is close. Update your story bank if a story keeps scoring low; our guide to building a STAR story bank with AI shows how to replace weak stories.
How do you know you are ready for the real interview?
No score guarantees an offer, but these signals suggest your practice has done its job:
- Your average score has been steady across your last three comparable sessions.
- Your recent fixes hold without you thinking about them.
- You can tell every story in your bank without notes, at about two minutes each.
- Follow-up questions no longer push you back to the start of a story.
- A person, not only software, has watched you answer under some pressure.
If you are close but unsure, the free interview readiness quiz gives a quick check, and our list of common interview mistakes is worth one final read. If you practice with both software and a person, our comparison of an AI mock interview vs practicing with a friend shows how to split the sessions.
Practice it
The easiest way to fill the log is to use a practice tool that scores every answer the same way. Rehearse out loud with a free AI mock interview built from your résumé and the job description: CoPilot Interview's AI interview assistant has a Mock Interview mode that reads each question aloud, scores each spoken answer out of 10 with what worked, what to fix and a model answer, and ends with a report card. Keep the level, mix and number of questions the same so your scores stay comparable. It is free for 15 minutes a day, with no account.
Our 7-day AI practice routine pairs well with the log, and our explainer on AI interview assistants covers what these tools can and cannot judge. They do not replace a human read on your delivery, so keep one session with a person in the plan, and in real interviews follow each employer's rules.
Get a consistent score for every answer
Answer interview questions out loud and see each one scored out of 10 with what to fix. Log the score, choose one fix, and run it again.
Try a free AI mock interviewFAQ
What should I track when practicing for interviews?
Track the date and type of each session, a score for each answer on the same scale, answer length, filler words, whether the answer had a clear structure and result, how you handled follow-ups, and the one fix you are working on. A fix that holds for two sessions in a row is the clearest sign of progress.
How do I count filler words in my interview answers?
Record your answer, get a transcript, and count words such as um, uh, like, you know and so at the start of sentences. Compare the count for the same kind of question from week to week rather than aiming for zero.
How long should my practice answers be?
Common guidance is about 1.5 to 2 minutes for a behavioral answer and 30 to 60 seconds for simple factual questions. Technical and design answers run longer, but each step should still be stated clearly before you go deeper.
How often should I review my practice log?
Once a week for about 20 minutes. Look at the score trend, check whether the last fix held, find the most common problem in your notes, and choose one fix for the next week.
Is an AI mock interview score a reliable measure of progress?
It is a useful trend if you keep the settings the same, such as level, question mix and number of questions. A single score is noisy, so compare averages over several sessions and pair the score with things you can count yourself, like answer length.
How do I know I am ready for the real interview?
Good signs are scores that stay steady across your last three sessions, fixes that hold without reminders, every story in your bank told without notes, and follow-up questions that no longer throw you off.