# The Behavioral Story Bank Run, From Dated Records to a Loop Allocation Sheet

*Build 6 to 15 true STAR stories from dated records, size them to your loop, and assign them so every competency is covered and no story repeats.*

- Canonical URL: https://www.refolk.ai/candidates/guides/behavioral-story-bank-allocation-sheet
- Pillar: Positioning and materials
- Format: Playbook
- Published: 2026-11-21
- Last reviewed: 2026-11-21
- Reading time: 22 min
- Keywords: how to build a behavioral interview story bank, how many STAR stories do I need, map STAR stories to competencies, can I reuse the same story in different interview rounds, story matrix for leadership principles interview, behavioral interview prep from performance reviews

## Key takeaways

- Size a behavioral story bank by loop math: interviewers times competencies per interviewer, which gives 8 to 15 slots for an Amazon loop reported at 4 to 5 interviewers scoring 2 to 3 Leadership Principles each.
- Published story-count advice ranges from 5 to 15 stories because each guide assumes a different loop, so the right number comes from the recruiter, not from an article.
- Every figure in a story should trace to a dated document such as a self-review, offer letter, launch post, or the resume you actually sent, because a remembered 40% that the record shows as 25% fails the first measurement probe.
- Write each story at 200 to 300 words with about 60% of it on your own actions, then prepare written answers to OPM's three probe families: Situation/Task, Action, and Result.
- Sources disagree on whether interviewers share notes before the debrief, so the safe default is an allocation sheet where no story ID appears twice within one loop.
- In Refolk's index, UK profiles list Competency Based Interviewing 8.5 times as often as Behavioral Interviewing, so UK candidates should search for prep material and frameworks under competency and behaviour terms.

This playbook is for anyone facing a behavioral or competency-based interview loop who wants a bank of true "tell me about a time" stories, not improvised ones. It runs from dated work records to a per-interviewer allocation sheet. The output is 6 to 15 story cards, with every figure tied to a named document, every card tagged and prepared for follow-up questions, and no card used twice in one loop. Expect roughly three hours for the build once your records are in hand, plus the statutory wait if you have to request a personnel file.

## What a behavioral story bank is and what it has to cover

A behavioral story bank is a fixed set of true, rehearsed work stories. Each one is mapped to a competency the employer's loop will score, so every "tell me about a time" question gets a prepared answer you have not already given another interviewer in the same loop. A bank has a size set by the loop, a source for every fact, and a plan for which story goes to which interviewer. A list of favourite anecdotes has none of those.

The terms this guide uses:

- **Loop:** the full set of interviews for one role at one employer, from first behavioral round to final panel.
- **Competency:** a named behaviour the employer scores. Amazon calls them Leadership Principles. The UK Civil Service calls them behaviours, under its Success Profiles framework.
- **Card:** one story written in STAR form (Situation, Task, Action, Result), with the record that backs it.
- **Probe:** a follow-up question an interviewer asks to test an answer.
- **Allocation sheet:** a table that assigns each card to one interviewer and one competency for a specific loop.

Most published advice says "write 6 to 10 STAR stories and tag them". That advice leaves out three things. It does not say where the facts come from. It does not say how to work out the right number. And it stops before the loop, where the actual risk is: running out of stories, or repeating one to two interviewers. This playbook covers all three.

#### The evidence chain behind every answer

1. **Allocation slot** - one interviewer, one competency, one primary card and a backup
2. **Story card** - 200 to 300 words in STAR form, tagged, with three probe answers
3. **Event line** - date, record, what happened, figure as the record states it
4. **Dated record** - self-review, offer letter, launch post, the resume you sent

*Each answer you give should trace down through a card and an event line to a dated document, and any break in that chain is where a probe will land.*

## How many STAR stories do I need?

You need one card for every interviewer and competency slot in your loop, plus at least two cards per competency, so you always have a backup. Work it out by multiplying the number of interviewers by the number of competencies each one scores. That product is your slot count, and it is the honest answer to "how many stories".

Published advice is all over the place, because each source is quietly assuming a different loop:

| Source | Stories recommended |
|---|---|
| thetailorcv.com | 5 to 6 |
| cvpilot.pro | 5 to 8 |
| extern.com / aced.io | 8 to 10 |
| techscreen.app | 8 to 12 |
| jobsbyculture.com | 12 to 15 |

The top of that range is three times the bottom. Loop math explains the gap. A 5-story count fits a single structured interview. A 15-story count matches the ceiling of a reported Amazon loop. Neither number is wrong. Each one is the right answer for a loop you may not be facing.

| Context | Interviewers | Competencies each | Slots |
|---|---|---|---|
| Amazon (coach reports) | 4 to 5 | 2 to 3 | 8 to 15 |
| OPM single interview | 1 | 4 to 6 | 4 to 6 |
| UK Civil Service (coach report) | panel | 3 to 5 | 3 to 5 |

These figures come from different places. The OPM figure is official: the US Office of Personnel Management says structured interviews typically assess 4 to 6 competencies, unless the job is unique or high level. Amazon confirms officially that different interviewers are assigned different Leadership Principles to evaluate, but the per-interviewer and per-loop counts come from candidate and coach reports, not from Amazon. Amazon's own SDE II prep page says each interviewer will typically ask two or three behavioral questions, which is consistent with those reports. Loops are reported at 4 to 5 interviews of 45 minutes each.

#### How Amazon's principles narrow to one interviewer

| Stage | Figure | Note |
| --- | --- | --- |
| Leadership Principles | 16 | the full published list |
| Scored in one loop | 8 to 15 | 4 to 5 interviewers times 2 to 3 each, per coach reports |
| Assigned to one interviewer | 1 to 3 | interviewing.io says 1 to 3, coaches say 2 to 3 |

*Amazon has 16 Leadership Principles, coach reports put 8 to 15 of them in a single loop, and each interviewer is assigned only a few.*

> **Rule:** Size from the loop, not the article
>
> Get the interviewer count and the competency list from the recruiter, multiply, and use the higher figure wherever sources disagree. A bank sized from the low end of a range leaves slots with nothing to say.

Two adjustments turn the slot count into a bank size:

1. **Floor of 6.** Even a single interview scoring 4 to 6 competencies should have at least 6 cards. One card per competency leaves you stuck if the interviewer asks for a second example or if the first story turns out to be a weak fit.
2. **Two cards per competency.** Every competency column needs at least two cards against it. That is extern.com's rule, and it matches Amazon's own advice to prepare multiple examples for each Leadership Principle. Secondary tags count towards the two, so you do not need twice as many cards as competencies.

On a reported Amazon loop, that comes to 8 to 15 cards. For the loops in the table above, the result lands between 6 and 15.

## Which work records can I still get, and how fast?

The records already in your hands are the ones to start from: the resume you sent, self-reviews you kept, offer letters, and public launch posts. Your formal personnel file is a different matter. In the US, federal law does not guarantee you the right to see it, so access depends on state law, and some state windows close over time.

A dated record does two jobs. It fixes the date, which stops two similar projects from blurring into one story. And it fixes the figure, which is what an interviewer is testing when they ask "how did you measure it?" Amazon says outright that behavioral answers should include metrics or data where applicable, and a figure you cannot source is a liability rather than a strength.

The resume you sent matters most of all, because it is the document your interviewers are holding. Every claim on it is a claim they may ask you to tell as a story. If [Refolk](/candidates) tailored your resume for this posting, extract from that tailored version, not from a master copy, so your stories match the lines the panel has in front of them.

Where state law gives you rights to your personnel file, the rules differ a lot:

| State | Former-employee access | Turnaround |
|---|---|---|
| California | Inspect and copy performance records; one request per year | 30 days |
| Washington | Counts as former if separated within 3 years of the request | 21 calendar days, no cost |
| New York | Labor Law 210-b, check whether it is in force | 5 business days |
| Louisiana | None, unless the handbook grants it | Not applicable |

Typical personnel file contents include applications, offers of employment, pay records, and performance reviews. Those are useful for dates and for review language. They will not hold the operational figures that make a story concrete, the kind that live in tickets, launch documents, and dashboards. Nobody has published data on which of those records people usually lose when they leave, so I assume they are all gone. Before you leave a job, write your own dated notes of the figures you were responsible for, and check your employer's policy before copying any internal document.

I have not verified how UK subject-access rules apply to personnel records. UK readers should check the current guidance from the UK data protection regulator before relying on a request.

> **Watch out:** The window closes after you leave
>
> Washington stops treating you as a former employee three years after separation, and California employers only have to honour one request per year from a former employee. Collect copies of your own reviews and figures while you still work there.

Turn the records into an event list, one line per event, before you write a single story. A good line looks like this: "Q3 self-review, dated; migrated billing service; error rate figure as stated in review." Sources disagree on where to start. One says to pull 5 to 6 stories from your resume rather than from memory alone. Another starts from a raw memory dump and reconstructs from there. I start from the records and use memory only to find events that I then go and source. That way the memory dump finds candidates and the records decide which ones survive.

## What the loop spec is and how to get it

The loop spec is two facts: how many interviewers you will see, and which competencies the loop will score. The recruiter is the fastest source for both. Without the spec you cannot size the bank, and you cannot allocate it.

At Amazon, the recruiter is also the person who tells you which principles your loop will focus on. Cody Nelson, in Amazon recruiting, recommends building your stories from the job description, which gives you a second source when the recruiter is vague. For the UK Civil Service, the behaviours are specific to the grade or level of the role, so the advert and the grade descriptor are part of the spec.

Ask the recruiter for:

1. The number of interviews in the loop, and which ones are behavioral.
2. The competencies, principles, or behaviours the loop scores.
3. If they will say, which interviewer covers which competencies.
4. The length of each behavioral interview.
5. The grade or level you are being assessed against.

If the recruiter gives you a range, take the top of it. If they give you nothing, size from the higher coach figure for that employer, and tag cards widely so a surprise competency still has something under it.

Once you have the competency list, set it against your history before drafting. Refolk scores how well your history fits a posting, and the requirements where that score is weakest are usually where your bank will be thinnest. Find those gaps now, while you can still go looking for a record, not in the room.

### Search under the employer's vocabulary

UK candidates should search for prep material and frameworks under "competency" and "behaviour" terms, because that is the language UK employers and their interviewers use. In Refolk's index of professional profiles, the split is sharp:

| Skill listed | Country | Profiles |
|---|---|---|
| Behavioral Interviewing | United States | 13,836 |
| Behavioral Interviewing | United Kingdom | 985 |
| Competency Based Interviewing | United Kingdom | 8,333 |

**8.5x - UK profiles listing Competency Based Interviewing for every one listing Behavioral Interviewing, in Refolk's index**

The US has 14 times as many Behavioral Interviewing profiles as the UK, and the two skills may overlap on the same profiles.

UK public-sector frameworks such as Success Profiles use competency or behaviour language, and that shows up in how the people running the interviews describe themselves. In small 25-profile samples, Gartner appeared twice among UK Behavioral Interviewing holders and Agora Talent twice among UK Competency Based Interviewing holders. Samples that small are only indicative, not a ranking of employers.

## The procedure: from dated records to an allocation sheet

The full run has nine steps. Two of them take real time: requesting records, where the statutory turnaround runs from 5 business days to 30 days, and drafting the cards. Everything else takes minutes to an hour. One prep source budgets 90 minutes for writing inside a three-hour build, which leaves about 90 minutes for tagging, probes, and allocation.

#### Behavioral story bank run

1. **Collect dated records** - Gather the resume you sent, self-reviews, offer letters, and public launch posts, and request your personnel file where state law allows it. Done when every role you will draw from has at least one dated document.
2. **Extract one line per event** - Write one line per event with date, record, what happened, and the figure as the record states it. Expect 1 to 2 hours; done when every candidate story has a source line.
3. **Get the loop spec** - Ask the recruiter for the interviewer count and competency list, and read the advert and grade descriptors. Done when you can write down both numbers and the named competencies.
4. **Size the bank with loop math** - Multiply interviewers by competencies per interviewer, using the higher figure where sources disagree. Done when you have a target of 6 to 15 cards and a list of slots.
5. **Draft 200 to 300 word cards** - Write each story as a STAR card with about 60% on your own actions, citing the record for every figure. Done when each card fits the template and has no unsourced number.
6. **Tag primary and secondary competencies** - Give each card one primary and at most two secondary competencies, matched to your grade's descriptor. Done when every competency has at least two cards.
7. **Prepare the three probe families** - Write short answers to the Situation/Task, Action, and Result probes for every card. Done when you can say all three without notes.
8. **Build the allocation sheet** - Assign a primary and a backup card to every interviewer and competency slot, with spares held in reserve. Done when no competency is uncovered and no card ID repeats as a primary.
9. **Rehearse, log, and refresh** - Time every card out loud, log which card you used after each round, and update the sheet before the next one. Refresh the bank every 6 to 12 months or when a new loop starts.

## How to write a story card that survives three follow-ups

A card that survives follow-ups has three things. It runs 200 to 300 words. About 60% of it is about what you did, in the first person. And it has written answers ready for the three probe families OPM tells interviewers to use. The length keeps the answer inside the time an interviewer can give it. The Action weighting and the probe answers are what get it through the follow-ups.

### Length

Practitioner consensus puts a STAR answer at about 90 seconds to two minutes, or roughly 200 to 300 words at a normal pace. Speaking rate shifts that a little. At 150 words per minute, 300 words takes two minutes and 225 words takes 90 seconds. One source assumes 130 words per minute, which brings it down to 150 to 250 words. Adults reading aloud average about 183 words per minute in a meta-analysis of 77 studies. That is faster than most people speak in an interview, so time yourself talking, not reading. Deep-dive answers in Leadership Principle interviews can run 2 to 3 minutes, rarely more. OPM rating guides give no answer length.

### Shape

Put about 60% of the card on Action. That split is attributed to MIT guidance by one prep site. On a 250-word card it means about 150 words on what you did, and about 100 words between the Situation/Task and the Result. Write the Action in "I", not "we". OPM tells interviewers to narrow in on the candidate's specific role and actions, so a team story with no individual part will be taken apart by the second probe.

### Probes

OPM's interviewing guidance groups probes into three families. Interviewers are told to use very similar probes for every candidate, which makes them predictable enough to prepare for:

| Probe family | Example questions |
|---|---|
| Situation/Task | Who was involved? What led up to this? |
| Action | How did you respond? What was your role? |
| Result | What was the outcome? What would you do differently? |

Candidates report the same pattern at Amazon: "What did you do specifically?", "What would you change?" and "How did you measure it?" I prepare every card to three depths, one per family. That is my own convention, not a published standard. OPM leaves it to each interview to decide whether probing is none, limited, or unlimited, and no fixed depth is established publicly. Interviewers also differ in style even under one rubric, as interviewing.io reports from data on 500 Leadership Principles interviews. Depth on every card protects you better than breadth across many thin ones.

**Story card**

```
CARD ID: C-01
Title (five words or fewer):
Primary competency:
Secondary competencies (max 2):
Grade descriptor matched:
Record: [document name] | [date] | [where your copy lives]
Figure as the record states it:

SITUATION / TASK (about 50 words):
ACTION (about 150 words, first person, what I did):
RESULT (about 50 words, sourced figure, what changed):

PROBE 1, Situation/Task - Who was involved? What led up to it?
PROBE 2, Action - What was my role? What did I do specifically?
PROBE 3, Result - How did I measure it? What would I do differently?

Timed out loud: [seconds] on [date]
Used in: [employer] | [round] | [interviewer] | [date]
```

*One card per event. Delete any figure whose source line you cannot fill in.*

Before a card goes in the bank, put it through two tests: can you show the figure in a document, and is the Action yours? The matrix below sorts cards by those two tests.

#### Is this card bank-ready?

Horizontal axis runs from Figure recalled from memory to Figure tied to a dated record. Vertical axis runs from Action told as "we" to Action told as "I".

| Quadrant | What it means |
| --- | --- |
| Strong role, weak proof | find the record, or drop the number and describe the result |
| Bank-ready | tag it, time it, and allocate it |
| Cut | replace it with an event from your record list |
| Team story, good data | rewrite the Action around your own decisions |

*A card needs both a sourced figure and a first-person Action before it can be allocated, and anything short of that is either a rewrite or a cut.*

## The allocation sheet, and whether you can reuse a story

The allocation sheet has one row per interviewer and competency slot. Each row names a primary card and a backup, and no card ID appears twice as a primary within one loop. It is the last thing you build and the thing you carry into the loop.

Can you reuse the same story in different interview rounds? The evidence is split. Amazon's official pages do not address reuse. They recommend preparing multiple examples for each Leadership Principle. Third-party sources contradict each other. One says interviewers deliberately do not discuss you with each other beforehand, so a repeat is acceptable. Another says interviewers share notes before the debrief, so a repeat should be avoided. For Civil Service written statements, one third party says sifters check for reuse and that it weakens both statements.

> **Watch out:** Treat any repeat as visible
>
> You cannot tell which version of the note-sharing claim holds for your loop, so plan as if every interviewer will see every answer. A no-repeat sheet costs you a few extra cards, while a visible repeat can cost you the debrief.

Reuse across separate loops is a different question. No employer has published guidance on it, so it is not established either way. I treat each new loop as a fresh sheet built from the same bank. If you are interviewing again at an employer you have been through before, I would not assume your earlier answers are gone from their records.

For Amazon loops, plan for the Bar Raiser seat by keeping at least one unallocated card per competency in reserve. Any interviewer can push past your primary, and a reserve card means you never have to reach for a story you have already told.

**Loop allocation sheet**

```
LOOP: [employer] | [role] | [grade]
Spec confirmed by recruiter on: [date]
Slots = [interviewers] x [competencies each] = [n]

INTERVIEWER        | COMPETENCY     | PRIMARY | BACKUP
1 [name or role]   | [competency A] | C-03    | C-08
1 [name or role]   | [competency B] | C-01    | C-11
2 [name or role]   | [competency C] | C-05    | C-02
2 [name or role]   | [competency D] | C-07    | C-04
3 [name or role]   | [competency E] | C-09    | C-06

RESERVE (unallocated): C-10, C-12
CHECK 1 - every competency has a primary: [yes / no]
CHECK 2 - no card ID repeats in the PRIMARY column: [yes / no]
```

*One sheet per loop. Fill the slot count first, then the rows, then run the two checks at the bottom.*

Backups can appear more than once on the sheet, because a backup is only told if the primary is rejected. Once you use a backup, it becomes a primary. Strike it from every other row before the next interview.

## How behavioral story banks go wrong

Most banks fail in one of seven ways, and each one looks fine until a probe hits it. Each entry below covers what the failure looks like, why it gets past a quick review, and the check that catches it.

1. **Unsourced figure.** A number recalled from memory differs from the record. The false positive is a confident "40%" that the self-review shows as 25%. It survives rehearsal because nobody in your kitchen asks "how did you measure it?" Check: every card cites a dated document, and the figure matches it word for word.
2. **"We" answers.** The headline sounds strong, then the answer fails OPM's "What was your role?" probe because the Action describes the team. Check: count the uses of "I" in the Action section, and rewrite any card where "we" outnumbers them.
3. **Wrong loop count.** Coach reports range from 1 to 3 and from 2 to 3 Leadership Principles per interviewer. A bank sized from the lower figure looks complete on paper and leaves slots empty in the room. Check: confirm the count with the recruiter and size from the higher figure if they will not say.
4. **Cross-round repeat.** A story told twice in one loop. It slips through because each round feels separate while you are in it. Sources split on whether notes are shared before the debrief, so the safe default is to treat any repeat as visible. Check: no duplicate card IDs in the primary column, and backups struck out once used.
5. **Grade mismatch.** The story is true and well told but pitched at the wrong level. In one coach's words, the same story can be a strong Level 2 and a weak Level 4. This happens most with Civil Service behaviours, which are specific to grade. Check: hold each card against the descriptor for the grade you are applying for, not the grade you hold now.
6. **Length drift.** An answer that runs past 2.5 minutes. It usually means the Situation has taken over, or you are telling two stories at once. It passes on paper because word counts look fine and the overrun comes from ad-libbing. Check: time every card out loud and split any card that holds two events.
7. **Record not retrievable.** The story depends on a document you can no longer get. A Louisiana employee may have no right to their file at all, unless the handbook grants one. Check: secure copies of your own records before you exit, and drop or soften any figure whose record is gone.

> A number you cannot point to in a dated document is a number you should not say out loud.

The first two failures cause the most damage, because they look like strengths. A precise figure and a big team result both sound impressive right up to the follow-up. That is why the matrix test above comes before tagging, not after.

## Keeping the bank current after each round

The bank is never finished. After every round, log which card you used, with whom, and for which competency. Before the next round, update the sheet so no used card stays in a primary slot. Refresh the whole bank every 6 to 12 months, or whenever you start a new loop.

After each round, within the hour:

- Write down every question asked, the card you used, and any probe you were not ready for.
- Mark the card as used on the allocation sheet and promote its backup if needed.
- Add any probe you fumbled to that card's probe section, with a written answer.
- Note any competency the interviewer seemed to be scoring that was not on your spec, and ask the recruiter about it.

When a new loop starts, start from the loop spec again. The competency list, the interviewer count, and the grade can all change, and the same bank produces a different sheet. Re-extract from the resume you sent for that posting, because a resume tailored for a new role puts different claims in front of the panel. When you add a new role to your history, add its records and event lines straight away, while the documents are still yours to keep.

#### Before you walk into the loop

- [ ] Every card cites a dated record, and each figure matches that record exactly.
- [ ] The bank holds at least as many cards as interviewers times competencies per interviewer, and at least 6.
- [ ] Every competency on the recruiter-confirmed list has at least two cards tagged against it.
- [ ] Every card has one primary competency and no more than two secondaries.
- [ ] Every card matches the descriptor for the grade you are applying for.
- [ ] The Action section of every card is told in "I", with your own decisions named.
- [ ] Every card has written answers to the Situation/Task, Action, and Result probes.
- [ ] Every card times out loud at under 2.5 minutes.
- [ ] No card ID appears twice in the primary column of the allocation sheet.
- [ ] At least one reserve card per competency is held back unallocated.

## Frequently asked questions

### How many STAR stories do I need for a behavioral interview loop?

You need one story per interviewer and competency slot, plus a backup for each competency. Multiply interviewers by competencies each one scores: a reported Amazon loop of 4 to 5 interviewers at 2 to 3 principles each gives 8 to 15 slots, while a single OPM-style structured interview scoring 4 to 6 competencies needs fewer. Published advice ranges from 5 to 15 because each source assumes a different loop, so confirm the count with your recruiter.

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

Avoid it. Amazon's pages say to prepare multiple examples per Leadership Principle but do not address reuse, and third-party sources contradict each other on whether interviewers compare notes before the debrief. Because you cannot know which is true for your loop, treat any repeat as visible and make sure no story ID appears twice as a primary answer on your allocation sheet. Reuse across loops at different employers has no published guidance either way.

### Can I build behavioral interview stories from my performance reviews?

Yes, and self-reviews are one of the best sources, because they date an event and state its figures in writing. Pull each event into a one-line record with date, document, and figure, then draft the story from that line rather than from memory. If you no longer have the reviews, US access depends on state law: California and Washington give former employees rights to performance records, while Louisiana gives none unless the handbook grants it.

### How long should a STAR answer be?

Aim for 90 seconds to two minutes, which practitioners put at roughly 200 to 300 words at a normal speaking pace. Deep-dive answers for Leadership Principle interviews can run 2 to 3 minutes, rarely more. If an answer passes 2.5 minutes, the Situation has usually taken over or you are telling two stories at once. Time each card out loud, because reading speed is faster than interview speech.

### What is a story matrix for a Leadership Principles interview?

It is a grid with your stories as rows and the competencies or principles as columns, marking one primary and up to two secondary tags per story. Its job is gap detection: every column needs at least two stories before you allocate. The allocation sheet then turns the matrix into a per-interviewer plan, assigning a primary and backup story to each slot so nothing repeats within the loop.

### What if I cannot get records from a previous employer?

Use what you already hold: the resume you sent, offer letters, public launch posts, and any self-reviews you kept. For figures with no surviving record, either drop the number and describe the result qualitatively, or state it as an estimate and say how you would have measured it. Do not say a precise figure you cannot trace, because the measurement probe will expose it.

---

*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/behavioral-story-bank-allocation-sheet*
