# The Post-Round Interview Record, Graded Pass or Fail Before the Next Call

*Grade any post-round note pass or fail on eight fixed fields, written within 15 minutes, so two people grading the same note reach the same result.*

- Canonical URL: https://www.refolk.ai/candidates/guides/post-round-interview-record-standard
- Pillar: Interviewing
- Format: Standard
- Published: 2026-10-16
- Last reviewed: 2026-10-16
- Reading time: 22 min
- Keywords: what to write down after a job interview, how to keep interview notes across multiple companies, can I reuse the same story in different interview rounds, is it legal to record my job interview, interview debrief template for candidates, how to request my interview feedback under GDPR

## Key takeaways

- A post-round note passes only if all eight fields meet their criterion and the file timestamp is earlier than the start of your next call. One missing field fails the whole note.
- In conversation-memory studies, people recall only 10 to 14% of idea units five minutes after a conversation, so the record has to be written in the gap before the next call, not that evening.
- Recall of your own statements drops to 6% after 3 to 4 days (Ross and Sicoly, 1979), which is why every figure you said aloud must be logged exactly as spoken while it is still fresh.
- Amazon's debrief starts with 10 to 15 minutes of everyone reading every written assessment in silence, so a team size or outcome you told differently in two rounds is read side by side.
- Under UK GDPR an employer must answer a subject access request within one month, extendable by two months for complex requests, and under the CCPA within 45 calendar days, extendable by 45 more with notice.
- In Refolk's index there are about 420 US technical recruiters for every profile matching interview coach, against 24.8 in the UK, so most candidates will be grading their own notes without anyone checking them.

This is the pass/fail standard for the note a candidate writes after an interview round, before the next call starts. It is for anyone running several loops at once who needs a record they can use in the next round, the follow-up email, and the offer conversation. It sets out eight required fields, the time window for writing them, what makes a note fail, and the consent and data-access rules that decide what you may capture and what you can later ask to see.

Most interview-notes templates are written for the interviewer filling in a hiring scorecard. This one covers the candidate's side. It is built around the panel debrief, the meeting where interviewers compare notes, because that is where contradictory facts and numbers do the most damage.

## What counts as a usable post-round record?

A usable post-round record is a written note, timestamped before your next call, with eight fixed fields filled to a stated criterion. If any field fails its criterion, or the timestamp comes after the next call started, the note fails.

Each criterion is a check another person can run on the page without asking you anything. That is what makes two graders agree. "Good notes" is an opinion. "Every commitment has a date" is a fact you can look up.

### The eight fields and their pass criteria

1. **Interviewer names and roles.** Pass: a name and a role for every interviewer, good enough to address a follow-up email from the note alone. *Proves:* you can follow up and can tell which interviewer heard which story. *When it lies:* "Sarah, eng" looks filled in but can't address an email.
2. **Questions asked.** Pass: each question written out as it was asked, and a count at or above the floor for the interview length. Interview Genie puts behavioral interviews at 2 to 6 questions per hour, so 2 per hour is the floor. *Proves:* you know exactly what was tested. *When it lies:* themes such as "leadership stuff" pass at a glance but hide which competency was probed.
3. **Story and facts used per question.** Pass: a story ID from your bank against every question, plus the facts you used, with at least one figure per entry. *Proves:* you can retell the story the same way next time. *When it lies:* an ID with no facts lets the team size or outcome drift in the next telling.
4. **Figures stated aloud.** Pass: every number you said (pay, team size, metrics, dates) logged exactly as you said it and compared with your master facts sheet, with zero unexplained differences. A master facts sheet is the single document holding the canonical version of every number in your history. *Proves:* your numbers will match across rounds. *When it lies:* a range said aloud gets logged as a single number.
5. **Concerns the interviewer raised.** Pass: each concern written in the interviewer's words, or an explicit "none raised". *Proves:* you know what to address in the follow-up and the next round. *When it lies:* a blank field reads as "none" when you simply didn't write it.
6. **Commitments made.** Pass: every promise you made has a due date. *Proves:* nothing you offered gets dropped. *When it lies:* "will send samples" is a commitment with no date, and it fails.
7. **Next step and date.** Pass: the date they gave you, or today plus five business days if they gave none, entered in your calendar. *Proves:* you will follow up at the right time. *When it lies:* "they'll be in touch" is not a date.
8. **Consent status of any recording.** Pass: one of four values: yes, no, opted out, or none notified. *Proves:* you know whether a recording exists and whether any transcript you hold is safe to keep. *When it lies:* "N/A" written when a notice email was actually sent.

### The grading rule

- **Pass:** all eight fields meet their criterion and the file timestamp is earlier than the start of your next call.
- **Fail:** any field misses its criterion, or the timestamp is late.
- **A failed note gets fixed or labelled, not thrown away.** Fill whatever you still can, and mark the note "reconstructed" so you know not to treat its figures as verbatim.

> **Rule:** Blank is not none
>
> An empty field always fails. If the interviewer raised no concerns or you made no commitments, write "none raised" or "none made", so a grader can tell a checked field from a forgotten one.

## Why does the record have to exist before the next call?

Because memory of a conversation collapses within minutes, and the next call is exactly the kind of interfering activity that speeds forgetting. If you write tonight, you are reconstructing, not recording.

The research on how people remember conversations is blunt. These figures come from Brown-Schmidt and Benjamin's review, "How We Remember Conversation":

| Study | Delay | Recall |
|---|---|---|
| Samp & Humphreys 2007 | 5 min | 14% gist |
| Stafford & Daly 1984 | 5 min | 10% gist |
| Ross & Sicoly 1979 | 3 to 4 days | 6% own, 3% partner |

Two details matter for this standard. In Stafford and Daly's study, the best of 128 participants recalled only 40% after five minutes, so even an unusually good memory loses most of the conversation. Ross and Sicoly also measured recall of your *own* statements separately, and it was barely better than recall of what the other person said.

**6% - Recall of your own statements after 3 to 4 days**

Ross and Sicoly (1979) found people retained the gist of only 6% of what they themselves said, and 3% of what their partner said.

That second point is why the figures field exists. You can't rebuild later what pay range or team size you gave. If it isn't logged while fresh, it's gone, and the interviewer's write-up becomes the only version of what you said.

### Where the 15-minute window comes from

No study names a specific window, and I won't pretend one does. The 15-minute buffer is a policy choice built on the five-minute recall data and on the finding that rest after learning reduces forgetting while interfering activity speeds it up. Another call is about the strongest interference there is. The firm rule is the timestamp: the note must exist before the next call starts. Fifteen minutes is simply the smallest gap that leaves room to write it.

## How do panel debriefs turn small inconsistencies into a fail?

In a panel debrief, interviewers write their feedback independently and then read each other's write-ups side by side. Any fact you stated differently in two rounds shows up in the same reading pass, attached to your name.

Amazon's process is the most fully documented example. The Bar Raiser, a trained interviewer from outside the hiring team, has been part of the loop since the program launched in 1999. The Bar Raiser runs a debrief that one former Bar Raiser describes as bringing "all the interviewers in the same room debating their data points". It takes about 30 minutes with 5 or more interviewers, and Amazon's 16 Leadership Principles are split among them.

#### How a panel debrief reads your round

1. **Interviews** - Each interviewer covers assigned competencies and hears your stories
2. **Independent write-up** - Each submits written feedback before talking to anyone
3. **Silent read** - Everyone reads all assessments in silence for 10 to 15 minutes
4. **Debate** - The Bar Raiser leads discussion of the data points and gaps
5. **Decision** - Inconsistencies between retellings are weighed as a signal

*Because each interviewer writes alone and everyone then reads every write-up, a figure you told two ways ends up side by side in the same 15 minutes.*

Practitioner accounts say what gets flagged is about stories: the same story used for two competencies, or details that change between retellings. A coaching source says that "inconsistencies between retellings are treated as a signal." What has not been published is a list of the categories interviewers flag. Nobody has documented whether dates, team sizes, pay or reasons for leaving get checked more than anything else. I treat every number as in scope, because the silent read makes every number comparable.

### Can I reuse the same story in different interview rounds?

Yes, but how often depends on whether the interviewers compare notes, and the facts must be identical every time. The sources disagree on how often:

- **Interview Genie:** you can't use the same story twice in one interview.
- **A residency coach:** a story can be reused "several times (3 - 5) without any problem" across a day.
- **An Amazon-focused source:** avoid telling one story to more than two interviewers in a loop.

The disagreement mostly goes away once you ask whether the interviewers compare notes. Where each interview is scored on its own, reuse costs little. In a debrief, every repeat of a story is a chance for its details to disagree. Recommended story banks range from 3 to 5 stories (MIT, as cited by one prep source) up to 30 or more for senior Amazon loops, where you may face 25 to 35 behavioral questions.

#### Story reuse by loop format

Horizontal axis runs from Interviews scored separately to Panel debrief compares notes. Vertical axis runs from Story told once to Story told to several interviewers.

| Quadrant | What it means |
| --- | --- |
| Repeated, scored separately | Acceptable within 3 to 5 tellings a day; log facts each time |
| Repeated, debriefed | Cap at two interviewers; facts must match the log exactly |
| Single telling, scored separately | Low risk; still log the story ID |
| Single telling, debriefed | Safe; the log stops it repeating in a later round |

*The stricter reuse rule applies whenever interviewers compare notes, and logged facts are what keep any repeat safe.*

This guide covers recording which story you told and what facts it carried. Choosing which story to lead with for a given question is a separate decision.

## Is it legal to record your own job interview?

It depends on the state, and the conservative default is to not record without every party's consent. Several states require every party to agree, and no source I found addresses a candidate running their own AI transcriber. This section is not legal advice. It tells you what to check.

The Reporters Committee for Freedom of the Press (RCFP) recording guide lists these states as requiring the consent of all parties:

- California
- Delaware
- Florida
- Illinois
- Maryland
- Massachusetts
- Michigan (at least for recordings made by a third party who is not in the conversation; Michigan courts have held that a participant may record their own conversation)
- Montana
- New Hampshire
- Pennsylvania
- Washington

Lists range from 9 to 13 states depending on how hybrid rules are counted. Oregon is one of those hybrids: for conversations through a video conferencing program, all parties must be specifically informed before recording, and the Ninth Circuit, sitting en banc, upheld that rule. The stakes are real. California Penal Code § 632 allows civil damages of up to $5,000 per violation.

### The consent asymmetry

Employers face consent rules too. Greenhouse requires that "the candidate must provide consent for interview recording and transcription", and BrightHire sends candidates a notice email before the interview with an opt-out. BrightHire reports that 2 to 4% of candidates at Greenhouse customers opt out. So the employer may hold a lawful recording of your interview while you hold none. Two practical details: Greenhouse Notetaker transcribes in English only, and an interview in another language may produce no transcript at all.

#### Who holds which record of your round

1. **Employer recording or transcript** - Exists only if you consented; you can request a copy later
2. **Interviewer written feedback** - Written independently, read in the debrief, often accessible by request
3. **Your post-round note** - Written by you before the next call; the one record you control

*The employer may hold a recording and written feedback, but the only record you fully control is the note you write yourself.*

That asymmetry is why the consent field sits in the standard. It tells you whether an employer recording exists, and whether any transcript of your own is safe to keep.

> **Watch out:** An AI transcript is not automatically your record
>
> If you ran a transcriber without every party's consent in an all-party state, the transcript itself may be unlawful to have made. Treating your own transcriber as a recording is the cautious reading, not settled law. Check the consent field before keeping any transcript, and check the RCFP guide for every state involved.

## How do I write the record in the 15-minute window?

Run the eight steps below in order. Step 1 happens before the call, steps 2 to 7 fill the buffer after it, and step 8 is optional and can wait. Only the raw dump is under real time pressure; the rest is checking.

#### Writing the post-round record

1. **Log recording consent before the call** - Check whether a recording or notetaker notice arrived and whether you opted out. Done when the consent field reads yes, no, opted out, or none notified.
2. **Block a 15-minute buffer between calls** - Put a gap of at least 15 minutes after every interview on your calendar. Done when the gap exists before the next call starts.
3. **Raw-dump names, roles, and questions** - Inside the buffer, write every interviewer's name and role and every question asked, word for word where you can, in 5 to 8 minutes. Done when the question count matches what you remember being asked.
4. **Map each question to a story and its facts** - Tag each question with a story ID from your bank plus the facts you used, including at least one figure. Done when no story ID appears twice in the round.
5. **Log every figure and diff it** - List every number you said aloud exactly as you said it, ranges included, and compare it with your master facts sheet. Done when there are zero unexplained differences.
6. **Record concerns and dated commitments** - Write each concern the interviewer raised and each promise you made, each with a due date. Done when every commitment has a date, with documents and the thank-you due within 24 hours.
7. **Set the next step and date** - Write the next-step date they gave you, or today plus five business days if they gave none. Done when that date is in your calendar.
8. **Optionally request the employer's records** - Send a UK GDPR subject access request or a CCPA right-to-know request for your interview notes. Done when the request date and the response deadline are logged.

### Notes before the thank-you

Sources disagree on the order. Some coaches put the thank-you, due within 24 hours, ahead of full notes. I put it after step 5. A thank-you written first can repeat a figure or promise a deliverable you haven't checked against your master facts sheet yet. Both still fit comfortably inside the 24 hours. One recruiter agency's covenant sets the same window for documents: "Documentation should be turned around within 24 hours upon request."

### What the master facts sheet is checked against

The figures diff in step 5 only works if your master facts sheet matches the resume each company actually received. If you tailor your resume per posting, each company may be holding a slightly different set of numbers. [Refolk](/candidates) builds each tailored resume from the same underlying history, so the version a company saw and your master sheet stay in step, and the diff checks against what that company has in writing.

**Post-round record (one per round)**

```
COMPANY / ROLE / ROUND:
Call start - call end:
Next call starts at:            (file timestamp must be earlier)

1. INTERVIEWERS (name :: role)
-
-

2. QUESTIONS ASKED (as asked)          Count:     Floor (2/hr):
Q1:
Q2:

3. STORY MAP (question :: story ID :: facts used, incl. at least one figure)
Q1 ::  ::
Q2 ::  ::
Repeated story IDs this round: none / list

4. FIGURES SAID ALOUD (exactly as said :: master sheet value :: match? / explanation)
-

5. CONCERNS RAISED (their words, or "none raised")
-

6. COMMITMENTS (what :: due date)   Thank-you due:
-

7. NEXT STEP :: DATE (stated, or today + 5 business days)   In calendar: y/n

8. RECORDING CONSENT: yes / no / opted out / none notified
Inbox searched for "[company] record": y/n

STATUS: pass / fail / reconstructed
```

*Copy one per round; fill top to bottom inside the buffer, and never leave a field blank.*

## How post-round notes fail

Most bad notes don't fail on sight. They fail as false passes: the field looks filled in but can't do its job. Each failure below comes with the check that catches it.

- **Names field: first name and a department.** "Sarah, eng" with no surname or role looks complete. *Check:* could you address a follow-up email using only the note? If not, it fails.
- **Questions field: themes instead of questions.** "Asked about conflict and ownership" hides what was actually tested. *Check:* compare the question count with the floor of 2 per hour. A count below the floor almost always means you paraphrased.
- **Story field: an ID with no facts.** You log "S7" but not the team size or outcome you gave. The next retelling shifts the numbers, which is exactly what retelling comparisons catch. *Check:* every entry carries at least one figure.
- **Figures field: a range logged as a point.** You said "150 to 170" and logged "160". The interviewer wrote down what you said, not your average. *Check:* log exactly what came out of your mouth.
- **Consent field: "N/A" when a notice existed.** Notetaker notices arrive as routine scheduling email and are easy to miss. *Check:* search your inbox for the company name plus "record".
- **Timing: written after the next call.** The five-minute recall data plus interference from the next conversation means the note is reconstructed. *Check:* the file timestamp must be earlier than the next calendar start. Later is a fail, however detailed the note is.
- **Commitments: no date.** "Will send samples" fails. *Check:* every commitment line ends in a date.
- **AI transcript treated as the record.** A transcript feels more complete than a note, but if consent wasn't obtained in an all-party state, keeping it may be the problem. *Check:* read the consent field first.

> A note that looks complete but can't address an email or reproduce a figure is worse than a blank, because you will trust it.

### The figure that sinks screens

Which commitments candidates miss most often has not been established publicly, so I won't rank them. One failure is named, though: a coach lists "gave comp numbers (and they are way off)" as a root cause of failed screens. That is why pay sits in the figures field rather than a separate section. If you said a range to the recruiter on Monday and a different one to the hiring manager on Wednesday, both versions are now in writing on their side.

### What this standard does not prove

A note that passes proves your record is complete and consistent. It doesn't prove the interview went well, and it doesn't prove your figures are true. It only proves they match your master sheet. If the master sheet is wrong, every passing note carries the same error. Audit the sheet itself against source documents whenever you update your resume.

## How do I request my interview notes under GDPR or CCPA?

In the UK, send the employer a subject access request (SAR), which is a request for a copy of the personal data an organisation holds about you. In California, send a CCPA right-to-know request. Both can cover interview notes, and both have statutory deadlines. Log the request date and the deadline in your record.

| Regime | Base deadline | Extension | Clock pause |
|---|---|---|---|
| UK GDPR | 1 month | +2 months if complex | Clarification, ID, fee |
| CCPA | 45 calendar days | +45 days with notice | Not sourced |
| EU GDPR | Not separately sourced | - | - |

What is established:

- **UK GDPR.** The employer "must respond without undue delay, and within one month of receipt of the request", and may extend by up to two further months if the request is complex. The clock pauses while it waits for clarification. The request doesn't need to cite Article 15.
- **Interviewers are told to expect it.** The ICO's Employment Practices Code tells employers to make sure "all interviewers are aware that interviewees may have a right to request access to their interview notes."
- **CCPA.** Businesses must respond within 45 calendar days and can extend by another 45 (90 total) if they notify you. The request covers the preceding 12 months and is free of charge. The CCPA's former exemption for job applicants has expired.
- **Redaction.** Expect other people's personal data, such as other candidates or interviewers, to be redacted.

What is not established here: the specific UK Data Protection Act 2018 exemptions an employer might rely on (such as management forecasting or negotiations) were not verified for this guide, and EU GDPR timing was not sourced separately from the UK rules. If you are in the EU, read your national regulator's subject access guidance before relying on the UK figures.

Use the request when it changes a decision: after a rejection you want to understand, or when your record and theirs might disagree on something that matters, like pay. Comparing their notes with your figures log is the only way to see what the debrief actually read.

**Subject access request for interview records (UK)**

```
Subject: Subject access request - [your full name], [role] recruitment process

Dear [name or Data Protection team],

I am making a subject access request for a copy of the personal data you hold about me in connection with my application for [role], including interviews on [dates].

Please include:
- Interview notes and written feedback from each interviewer
- Any scorecards, ratings, or debrief notes
- Any recordings or transcripts of my interviews
- Emails or messages about my candidacy

I understand you may need to redact the personal data of other people.

Please confirm receipt. I understand you are required to respond within one month of receiving this request.

[Your full name]
[Email used in the application]
```

*Send to the employer's data protection contact or recruiter; for California, call it a request to know under the CCPA.*

## Running the standard across several live loops

Adopt it as policy: one note per round, the same eight fields every time, graded by you within the buffer, with a master facts sheet shared across every company. Across three or four loops, consistency of format matters more than the depth of any single note.

You will most likely be grading your own notes. Refolk's index shows how lopsided candidate-side support is next to recruiter capacity:

| Country | Technical recruiters | Interview coaches | Recruiters per coach |
|---|---|---|---|
| United States | 21,424 | 51 | 420 |
| United Kingdom | 718 | 29 | 24.8 |
| US/UK multiple | 29.8x | 1.8x | - |

**420 - US technical recruiters per interview coach**

In Refolk's index of professional profiles there are 21,424 US profiles titled Technical Recruiter against 51 matching Interview Coach, compared with 24.8 recruiters per coach in the UK.

Treat the coach counts as rough. They are headline matches, and only 6 of 25 sampled US profiles and 11 of 25 UK profiles held that exact title. The direction is clear even if the exact ratio isn't. On the employer side, someone reads every write-up. On yours, a pass/fail standard you can grade alone is the closest thing to a reviewer you're likely to get.

### Using the record across loops

- **Before each round,** read the previous rounds' notes for that company: story IDs already used, figures given, concerns raised.
- **After each round,** add any new or changed figure to the master facts sheet only after you've confirmed it is true. The sheet is not a log of whatever you happened to say.
- **When an interviewer raises a concern,** carry it into the next tailored application as well as the next round. If the same concern comes up at several companies, it is a fit gap, not bad luck, and the fit score Refolk gives each posting is a quick way to see whether the gap is real before you apply again.

### Keeping the standard current

Three inputs here can change, and each has a place to re-check:

- **Recording consent.** Re-read the RCFP Reporter's Recording Guide for every state involved when you start a loop in a new state. Hybrid rules like Oregon's are where the lists disagree.
- **Access deadlines.** Re-check the ICO's guide to subject access and the California Attorney General's CCPA page before sending a request.
- **Employer notetaker practice.** Read the scheduling emails. Vendors' notice and opt-out flows are where the consent field gets its answer.

#### Before you close a post-round note

- [ ] The file timestamp is earlier than the start of my next call.
- [ ] Every interviewer has a name and a role I could address an email to.
- [ ] The question count is at or above 2 per hour, and the questions are written as asked.
- [ ] Every question has a story ID and at least one figure, and no ID repeats within the round.
- [ ] Every figure I said aloud is logged exactly as said, ranges included, with zero unexplained differences from the master facts sheet.
- [ ] The concerns field lists each concern in the interviewer's words or says "none raised".
- [ ] Every commitment has a due date, with documents and the thank-you due within 24 hours.
- [ ] The next-step date is in my calendar, either as stated or as today plus five business days.
- [ ] The consent field reads yes, no, opted out, or none notified, and I searched my inbox for a recording notice.
- [ ] Any transcript I hold was made with every party's consent, or I have deleted it.

## Frequently asked questions

### What should I write down right after a job interview?

Write down eight things before your next call: each interviewer's name and role, every question asked, the story and facts you used for each one, every figure you said aloud, any concerns they raised, any commitments you made with due dates, the next step and its date, and whether the call was recorded. Write the questions as they were asked, not as themes, and log figures exactly as you said them, ranges included.

### Can I reuse the same story in different interview rounds?

Yes, as long as the facts are identical every time. How often depends on whether the interviewers compare notes. Advice ranges from never twice in one interview to 3 to 5 times across a day. In a loop that ends in a panel debrief, use the stricter rule: tell a story to no more than two interviewers and keep every figure in it exactly the same.

### Is it legal to record my job interview?

It depends on where you and the interviewer are. A group of states (counts run from 9 to 13 depending on the source) require every party's consent, including California, Florida, Illinois, Pennsylvania and Washington. I have found no source that addresses a candidate running their own AI transcriber. The safe reading is to treat it as a recording and ask first. This is not legal advice, so check the RCFP recording guide for every state involved.

### How do I request my interview feedback under GDPR?

Send the employer a subject access request asking for a copy of your personal data from the hiring process, including interview notes, scorecards and any recordings or transcripts. You don't have to cite Article 15. Under UK GDPR they must reply within one month. They can extend that by two months if the request is complex, and the clock pauses while they wait for clarification. Expect other people's personal data to be redacted.

### Should I send the thank-you note before or after writing my interview notes?

Write the notes first, then send the thank-you. Both fit inside the usual 24-hour window. Some coaches put the thank-you first. The risk is that you repeat a figure or promise a deliverable before checking it against your master facts sheet. Once the figures log is clean and every commitment has a date, the thank-you can confirm those dates instead of improvising them.

### How do I keep interview notes straight across multiple companies?

Keep one note per round in a fixed format, keep one master facts sheet shared across every company, and record a story ID against every answer. The fixed fields make notes comparable, the master sheet makes figure drift visible, and the story IDs show which stories each company has already heard. Store the notes with timestamps you can check, so you can prove each one was written before the next call.

---

*From the Refolk guide library. I revise these guides rather than replacing them, so the current version is always at https://www.refolk.ai/candidates/guides/post-round-interview-record-standard*