You just shook hands and walked out of the building. Right now you can still hear the exact wording of the fourth question the hiring manager asked, the three seconds of silence before you answered it, and the small change in her face when you mentioned the migration project. Forty minutes from now, on the train, half of that is gone. By Thursday you will have a vague feeling and two headline moments, and you will have lost everything useful.
That is the entire argument for post interview reflection. It is not the thank-you email, which is a separate job that goes out the same day. It is not the polite chase you send when the recruiter blows past the date they gave you, which is also a separate job. This is a private file with exactly one reader: you.
Ten minutes now beats an hour three days later
Conversation memory is bad and it fades quickly. What survives is not the useful part. You will keep the moment you felt clever and the moment you felt stupid, and you will lose the ordinary middle where the actual information lives - the phrasing of a question, the job title of the quiet person taking notes, the throwaway remark about the last person who held the role.
Worse, delayed recall does not just fade, it rewrites. Three days later your brain has smoothed the whole thing into a story: it went well, or it went badly. That story is usually wrong and useless for preparing the next round.
So the rule is physical, not motivational. Do not get on the train. Do not call anyone. Sit on a wall, find a cafe, stay in the car park. Open your phone and spend ten minutes. Type badly. Nobody is reading it.
The template
Copy this and keep it somewhere you can duplicate in one click.
- Company, role, date, round number. Which stage this was, and what they called it.
- Who was in the room. Full names, job titles, who ran the conversation, who stayed quiet. Add whether they are in your reporting line.
- Every question, verbatim where you can manage it. Not a summary. The actual words. "Tell me about a time you disagreed with a decision that had already been made" is different from "tell me about a conflict", and the difference matters next time.
- The questions that made you stumble. Be honest here. Mark the ones where you rambled, went blank, or answered a slightly different question because you did not like the real one.
- The answer you wish you had given. Write it out properly, in full sentences, while the question is still fresh. This is the single highest-value line in the document.
- What they seemed to care about most. What did they return to twice? Delivery speed? Handling a difficult stakeholder? Cost? Somebody leaving badly?
- Tone and warning signs. Did they read your CV beforehand? Did anyone contradict a colleague? Did they dodge a question about the team? Was anyone visibly exhausted?
- Their questions to watch. Anything they asked that revealed a worry - about your notice period, your salary, your last departure.
- Next steps as they stated them. Who contacts you, by when, how many stages remain, and who you would meet.
- Your own gut read, one line. Do you actually want this job. Answer before you know whether they want you, because the answer changes once you have an offer or a rejection.
A worked example
Northbridge Logistics, Data Analyst, second round, Tuesday.
In the room: Priya Menon, Head of Ops, would be my manager. Tom Ashworth, senior analyst, on the team.
Questions asked: "Walk me through a dashboard you built that nobody used." "How do you decide something is a data quality problem and not a process problem?" "What would you do in your first month if the warehouse team ignores you?"
Stumbled on: the dashboard-nobody-used question. I got defensive and explained why it was actually fine.
Answer I wish I had given: I should have named the shipping-delay dashboard, said adoption was under ten percent because I built it from what the data allowed rather than what the supervisors were asked at 6am, and that the rebuild started with sitting in on two shift handovers.
They cared most about: getting the warehouse team to trust analytics. Priya came back to it three times. Tom mentioned the last analyst never left the office.
Warning signs: Priya said "we are still deciding what this role owns". Tom hesitated when I asked about the reporting line.
Next steps: Priya sends a task by Friday, final conversation the following week with the finance director.
Gut read: yes, if the ownership question gets a straight answer.
Using the notes for the next round
Before the next stage with the same company, reread that file instead of rereading the job ad. You already know what they push on, so you can prepare the specific thing rather than a generic set of stories. You know who you will meet and what they each seemed to want. And you can repair the stumble, because you wrote the better answer down while it was still recoverable - a second round is a legitimate place to say "I gave you a thin answer on that dashboard question last time, and I have thought about it since."
Patterns across several interviews
One debrief is a note. Twenty debriefs are data. Once you have six or seven, read them in a single sitting and look only at the stumble line.
If the same question has defeated you three times, that is not bad luck. That is a hole in your preparation, and it is costing you offers. The three that turn up most often in interview self review notes are the gap in your employment history, the reason you left the job you left badly, and the salary expectation. Sit down and build a real answer once. Say it out loud until it sounds like a sentence rather than a confession.
You will also spot patterns on their side. If four employers in the same sector all pushed on stakeholder management, that is what the market is buying, and your CV summary should say so.
Notes make you money at the negotiation
The people who negotiate well are usually not the boldest, they are the ones who took notes. When you get to money, you are not arguing about your worth in the abstract. You are pointing at the thing they told you they were struggling with, in their words, on a specific date, and explaining that this is the part you are good at.
The same file tells you what to ask for beyond salary. If they mentioned twice that the team is short-handed until spring, you know the start date has value to them.
Keep one running document
One spreadsheet for the whole search. Columns: company, role, date, stage, interviewers, key questions, stumbles, what they cared about, next step, next-step date, gut read.
The date column doubles as your follow-up tracker, so you stop wondering whether it has been too long. And when three processes are running at once, you stop mixing them up, which is the sort of small error that ends candidacies.
When it is a rejection
Open the file, spend fifteen minutes, and close it.
Harvest: which stumble showed up again, whether the gut read was right, whether the warning signs you wrote down were real. Then add anything you learned from the rejection itself - who sent it, how fast, whether they gave a reason.
That is it. The debrief is not a place to relitigate the interview at midnight. It exists so the next one goes better, and it can only do that if you write it while you can still hear their voice.