Interview story bank: coverage is the constraint, not polish

---
name: interview-story-bank
description: Turns a career history into a story bank: a matrix of competencies against stories, each story written once in a fixed form with its scope, its decision and its outcome, and every thin column named as a gap rather than papered over. Carries the coverage targets and the reuse cap, the word budgets for a two minute spoken answer, the four gates a failure story has to pass, the three probe types, and the in-room rule for choosing which story to spend on which question. It does not prepare technical interviews and it will not write an episode you did not live. This skill should be used when a behavioural or competency interview loop has been scheduled and the preparation so far amounts to one or two favourite stories.
---

# Interview story bank

## The claim this skill is built on

You finish holding three things. A coverage matrix, with the competencies as columns and your stories as rows, every cell marked primary, secondary or empty. A card per story, written once, in a fixed form. And a gap list: the columns your career does not currently cover, written as sentences you can say out loud.

The obvious approach is to pick your two best stories and make them excellent. It fails for a structural reason rather than a quality reason. A behavioural loop scores a small fixed set of competencies, chosen before you arrive, across several interviewers who then compare evidence in a debrief. Four interviewers, each spending most of an hour and asking two or three behavioural prompts after introductions and your questions, ask somewhere between eight and twelve prompts between them. If you hold three stories, at least one of them gets told more than once.

Repetition is invisible to you and obvious to them. In a serial loop each interviewer hears exactly one telling. The debrief is where the same project surfaces on three sets of notes, and by then you have left the building. It reads as a thin career, and it is the most avoidable failure in a loop.

There is a second mechanic worth stating, because it changes what a good answer is. Structured interviews are scored per answer against an anchored scale rather than as a single overall impression, which is what behaviourally anchored rating scales are for. So an excellent story told against the wrong competency does not score well somewhere else. It scores nothing in the column it was asked for. Coverage and correct filing beat polish, every time.

## Step 1. Recover the real competency set before writing anything

Three sources, in this order.

**A published framework, where one exists.** The clearest public example is the UK Civil Service Success Profiles, published on 18 June 2018 and last updated on 29 January 2025. It sets out five elements: behaviours, strengths, ability, experience and technical. The behaviours framework names nine: seeing the big picture, changing and improving, making effective decisions, leadership, communicating and influencing, working together, developing self and others, managing a quality service, and delivering at pace. Two features of it matter more than the list. Each behaviour is written out separately at six grade bands, from administrative grades up to director general, so the same named behaviour asks for a different scope at each level. And the guidance states that recruiting managers choose a selection of behaviours suited to the role, and that a candidate will not be asked to demonstrate all of them. That is public confirmation of the assumption the matrix rests on: the set is small, fixed in advance, and picked before you walked in.

Private employers publish their own versions. One large online retailer publishes sixteen leadership principles on its public pages, and dozens of companies have copied the format. Be precise about what such a page is: it presents the principles as how the company operates day to day, and it does not say which of them appear in which interview. The mapping from a published principle to an interview question is your inference, not the company's statement, so treat it as a strong hypothesis rather than a rubric.

**The posting's own verbs.** Strip the job description down to its verbs and its nouns of responsibility. "Own the roadmap for", "partner with", "define the standard for", "unblock", "operate with limited direction". Each of those is a competency in disguise, and the ones repeated across sections are the ones the hiring manager actually cares about.

**The recruiter.** Ask directly which competencies the loop covers and whether there is a written rubric. Some will tell you. Some will send the framework. Asking costs nothing and it is the single highest-yield minute in the whole preparation.

If none of the three yields anything, use a default set of seven axes as scaffolding to be replaced the moment better information arrives: ownership and scope; dealing with ambiguity; influence without authority, including conflict; failure and what changed afterwards; prioritisation under constraint; raising the standard when nobody asked; and customer or user judgement.

**Adjust the whole set for level.** At an individual contributor band, delivering at pace is a story about your own week. Two bands up it is a story about somebody else's quarter. Write each story at the scope of the band you are applying for, or the panel scores it as a level miss rather than a competency miss, which is the harder rejection to argue with afterwards.

## Step 2. Inventory episodes before you write a word of prose

Write stubs, not stories. One line each: year, role, what happened, who else was involved, and what you personally decided. Target eighteen to twenty-five stubs before anything gets written out.

Sweep the last three roles, plus any earlier role with unusual scope. Mine the sources rather than your memory, because memory returns the same four episodes every time: old performance reviews, promotion documents, planning docs, your calendar for the quarter of a launch, ticket or commit history, and decks you presented.

The qualifying test is a single question: can you name a decision you personally made? If you cannot, it is context rather than an episode, and it belongs in the background of somebody else's story.

Do not write prose at this stage. Writing prose early is exactly what produces a bank of three excellent stories, because polishing is more satisfying than remembering and it consumes the same evening.

## Step 3. Build the matrix, then read the holes

Draw the grid: competencies across the top, stubs down the side. Mark each cell P where the story is the strongest evidence you have for that competency, S where it is real but not the best, and blank otherwise.

Four coverage targets, each with the reason attached, because the reason tells you when to break it.

- **Two primary stories per competency.** The first is spent the moment an interviewer asks for it, and the next interviewer asks a variant of the same question.
- **No story tagged primary for more than two competencies.** A story that seems to fit four columns is generic, not versatile, and the same telling would have to be repeated to different people.
- **At least half of your primaries from the last three years.** A loop assessing seniority hears an old story as a story about who you used to be. The exception is real: an older episode is the right answer when it is the only one with genuine scope, and you say the date out loud rather than hoping nobody notices.
- **Every competency needs at least one story whose outcome carries a number.** Not every story. One per column.

Now read the holes. Any column with zero or one primary is thin. Do not fix it by re-tagging an existing story: that is the same story with a new label, and it fails on the first probe. Go back to the inventory, then back to the career, then to an older role, in that order.

**The gap branch.** If a column still has zero primaries after a second sweep, that is a real gap and it gets written down as a sentence you will actually say: "I have not run a team through a reorganisation. The closest I have is the depot consolidation, where I kept forty people through a move but nobody left involuntarily, and here is what I would want to get right if I did." A stated gap plus the nearest real thing beats a stretched story, because the stretched story dies on the first follow-up and takes your credibility for the rest of the hour with it.

## Step 4. The story form, and why STAR is not the missing knowledge

Every candidate knows situation, task, action, result. Most stories still fail. STAR is a container, and an empty container is still empty. Three things decide whether a told story scores.

**The scope is stated, in numbers, in the first three sentences.** How many people, how much money, how long, how many customers, how many sites. Without them the interviewer cannot tell whether this involved two people or two hundred, and an anchored rating scale cannot be applied to an unsized event. This is the single most common defect in an otherwise good answer.

**The decision is visible.** Not what you did: what you chose between. "I chose to cut the second region rather than delay the launch, because the forecast for that region was three months old and the launch date was contractual." What you did is the least interesting part of the story. The alternative you rejected, and why, is the assessment.

**The outcome carries a number, or an honest admission that it does not.** If you do not have the figure, say so and give the observable instead: "I do not have the retention number, and what I can point to is that the escalation queue was empty for the next quarter." An invented number dies on the second probe, and it dies badly, because now the question is about your honesty rather than your judgement.

**Word budgets.** A spoken answer of 350 to 450 words runs roughly two to three minutes at a normal speaking pace of about 150 words a minute. Allocate it: 60 to 80 words of situation carrying the scope numbers, 30 to 40 of task, 180 to 250 of action including the rejected alternative, and 60 to 80 of result including what changed afterwards.

**The written card is bigger than the spoken answer and shaped differently.** Roughly 250 to 350 words in note form, plus the probe block from step 6, plus two header lines naming which competency this is primary for and which it is secondary for. Never write the card as a script. Write it as five bullets and the numbers, because a memorised paragraph is audible and a bulleted card forces you to build the sentences live, which is what the interviewer is listening for.

## Step 5. The failure story is built differently, and you need two

This is the most reliably botched question in the set, and it is asked in almost every loop.

Four gates, all of which must pass.

- **It is real.** An invented failure has no aftermath, and the aftermath is the entire content.
- **It is yours.** Caused by a decision you made, not a decision made near you.
- **It cost something.** Money, a deadline, a customer, a person's trust, six months. If nothing was lost, it is not a failure.
- **It changed something specific.** Name the practice, threshold or habit that is different now, and how you know the change held.

The shape inverts. In a normal story the action gets most of the words. In a failure story the action gets 80 to 100 and the aftermath gets 200 to 250: what you noticed, when you noticed it, what you told whom and how quickly, what you changed, and what evidence there is that the change stuck.

**Tells of a fake failure story, as they look from the other side of the table.**

- *The disguised strength.* "I care too much about quality, so I over-invest." Reads as an unwillingness to be assessed, which is worse than the failure would have been.
- *The success in a coat.* Everything goes wrong for two minutes and then the candidate saves it and is congratulated. No cost survived to the end of the story.
- *Somebody else's failure.* A vendor missed a date, a reorganisation broke the team, a colleague left. Interesting, and not evidence about you.
- *The trivial cost.* A typo in a deck, a meeting that ran long. Signals either no scope or no candour, and the interviewer cannot tell which.
- *No change named.* "I learned to communicate better." Nothing checkable, so nothing scoreable.

Two failure stories, not one, for the same reason as every other column: the second interviewer asks a variant, and "I already told your colleague about that one" is not an answer to a question about your judgement.

## Step 6. Probe hardening, which is where the assessment actually happens

Interviewers probe because the prepared answer tells them very little and the unprepared follow-up tells them a lot. Three types, and you should write one answer to each on every card.

**The counterfactual.** What would you do differently. A weak answer is "nothing". An equally weak answer is a total repudiation, which reads as having no conviction. A strong answer names one specific thing and says why the rest of the decision still stands.

**The other mind.** What did the other person think, how did the team take it, what did your manager say at the time. This tests whether other people were genuinely present or whether you have narrated a solo performance in a company of one.

**The omission.** What would have happened if you had not been there. This is the probe that exposes a story where you were in the room but not on the hook, and it is the one candidates least expect.

**The rule: three probes.** A story you cannot survive three probes on is a story you did not live, or did not live at the level you are claiming. Demote it to secondary or drop it. Do this at the desk. The alternative is doing it in the room, where it is called being caught.

## The decision rule in the room

Which story do you spend on the question in front of you?

- **One story is primary for that competency and has not been used in this loop.** Use it.
- **Two or more are available.** Use the one whose outcome carries a number and whose scope matches the band you are interviewing for.
- **The best story has already been told to another interviewer.** Use the second, and say so plainly: "I talked about the migration with your colleague earlier, so let me use a different example." Panels compare notes in the debrief, and naming the constraint costs nothing while silently repeating the story costs a whole column.
- **No story is primary but one is secondary.** Use the secondary, and re-anchor it in the opening sentence to the competency being asked about rather than the one you originally tagged it for. Say what it is evidence of before you tell it.
- **You cannot tell which competency is being asked about.** Ask. "Do you want the version about how I handled the disagreement, or the version about how I made the call?" Guessing spends two minutes and one story on the wrong column, and because answers are filed per competency, a brilliant answer in the wrong column scores nothing. A clarifying question costs eight seconds and is itself evidence of the thing several frameworks call making effective decisions.
- **The question is hypothetical rather than behavioural.** Answer the hypothetical on its own terms first, then offer the nearest real episode as support. Substituting a story for the answer reads as evasion.

## Worked example

Invented throughout: a senior operations manager at a mid-size logistics company, applying for a similar band at a freight software business. The recruiter names five competencies for the loop: ownership, dealing with ambiguity, influence without authority, prioritisation under constraint, and learning from failure. Four interviewers.

**Inventory.** Twenty-one stubs across three roles. Nine pass the qualifying test, which is that a decision the candidate personally made can be named.

**First matrix, six stories.** Depot consolidation, primary for ownership, secondary for prioritisation, three sites into one, forty staff, eleven months. Warehouse system migration, primary for ownership, secondary for ambiguity, eight months ago. Carrier rate renegotiation, primary for influence, six per cent of freight spend, no line authority over the counterparty. Peak season staffing model, primary for prioritisation, secondary for ambiguity. Damaged goods claims backlog, primary for failure, a customer left. New region launch with no demand forecast, primary for ambiguity.

**Reading the holes.** Influence has one primary. Failure has one primary, against a loop of four people. Prioritisation has one primary and three secondaries, which is the pattern that means the candidate does a lot of it and has never framed any of it as a decision.

**Filling them.** A second sweep of the inventory produces nothing new for influence. A third sweep reaches back four years to a smaller role: persuading a finance team to move a month-end cut-off that was corrupting the depot count, no authority, three attempts and a data exercise before it moved. Smaller scope and older than the recency guideline, and it is admitted as the second primary for influence with the date said out loud, on the grounds that a real old story with a visible decision beats a recent story stretched into a column it does not fit.

For failure, the claims backlog is the only episode that passes all four gates. A second candidate exists and had been excluded because it is more uncomfortable: a hiring mistake that cost six months and a team's goodwill, and that changed the process by adding a work sample step. It passes all four gates. It goes in.

**The probe pass.** Peak season staffing collapses on the omission probe. Asked what would have happened without them, the honest answer is that an analyst built the model and the candidate approved it. Demoted from primary to secondary. Prioritisation is now short again, so the new region launch is promoted to primary for prioritisation as well as ambiguity, which is its second column and therefore still legal, and a stub about cancelling a project two quarters in is promoted to a full card as the second prioritisation primary. Ambiguity is now short, so the warehouse migration moves from secondary to primary for ambiguity, its second column.

**Verdict.** Twenty-one stubs became nine qualifying episodes and nine written cards, eight of which carry a primary tag and one of which sits on the bench as a secondary after failing a probe. Five competencies, two primary stories each, ten primary tags spread across eight stories, none tagged primary for more than two columns. Two cards are older than three years and both will be introduced with their date. One story was demoted at the desk on a probe the candidate had never been asked before, which is the entire reason for doing the probe pass in writing. Total output: nine cards of about three hundred words, twenty-seven probe answers, and no gap sentence, because every column was filled with something real.

## Failure modes

**Single Story Syndrome.** Three interviewers write down the same project. The candidate never learns this happened, and the debrief conclusion is not "the migration was impressive" but "we only ever heard about the migration".

**Scopeless Narrative.** A well told story that never says how many people, how much money or how long. The interviewer has no way to place it against a band, so it gets scored at the bottom of the range they were considering, because the anchored scale needs a size and the story did not supply one.

**The We Drift.** The whole story is told in the first person plural. "We decided, we shipped, we saw a lift." No individual contribution is visible. The omission probe exists precisely for this, and the candidate hears it as a hostile question rather than as a request for the missing information.

**Outcome Vacuum.** The story ends at the launch. What happened afterwards is unstated, so the interviewer cannot tell whether the decision was right, only that it was made.

**Probe Collapse.** The prepared two minutes are fluent and the follow-up produces vagueness or contradiction. This is worse than a mediocre story told consistently, because it retroactively casts doubt on the polish.

**Rehearsal Rot.** The answer is so smooth it sounds recited. Identical phrasing on the repeat, no hesitation anywhere, no detail volunteered that was not in the script. It reads as untrue even when it is true, and the fix is to hold bullets and numbers rather than sentences.

**Competency Blindness.** The candidate prepares hard for the competencies they are already good at and skips the one they dread. The loop was designed to cover the set, so the dreaded column is guaranteed to be asked, usually by the interviewer who cares most about it.

**The Reused Failure.** One failure story, told twice, the second time with an apology for repeating it. The second telling is scored lower than the first for a reason nobody says out loud: the panel now knows the size of the bank.

## What this skill does not do

- It does not prepare technical or skills-based assessment of any kind. Coding rounds, system design, case interviews, take-homes, portfolio reviews and live exercises are judged on the work in front of the interviewer, not on your account of past work, and nothing in this file helps with them.
- It cannot tell you the employer's real rubric. It works from what is published, what the posting says and what the recruiter tells you, and all three can be wrong or out of date.
- It will not write an episode you did not live. Where a column has no material, the output is a gap sentence, and a gap sentence said honestly is a better answer than a stretched story.
- It cannot interrupt you, and interruption is the mechanism that trains probe resistance. A person who cuts in at the wrong moment does more for you in an hour than any written bank does in a week.
- It stops at the loop. The recruiter screen has its own disqualifiers and compensation turns on information gathered long before the offer call. Both are separate conversations that this material does not touch.
- It does not rank you against other candidates. Full coverage of the matrix removes an avoidable failure. It does not make the stories more senior than the career that produced them.