The Behavioral Story Bank, Built From Your Records to a Coverage-Checked Set

This playbook turns your work history into a behavioral interview story bank: a fixed set of true, tagged stories you can draw on for any "tell me about a time" question in every loop you have coming up. It is for anyone with more than one interview ahead who is tired of hunting for examples the night before each one. In one working session it takes you from a raw inventory of at least 30 experiences to 8 to 12 stories. Each has a fact sheet tied to a source of record, a 30-second and a 90-second version, and a prepared probe layer, and a coverage matrix shows every required competency covered twice.

What a behavioral story bank is, and why it deserves a full session

A behavioral story bank is a small, fixed set of real experiences from your history, each checked against documents and tagged to the competencies your target employers interview for, so that you choose your answers in advance instead of improvising them. It deserves a full session because structured interviewers score these answers against a rubric, and the follow-up questions are designed to find stories that do not hold up.

A few terms, defined once:

• Behavioral question: a question about what you did in the past ("Tell me about a time you...").
• Situational question: a hypothetical ("What would you do if...").
• Competency: a capability the employer has decided the job needs, such as ownership or stakeholder management, written into the posting or into the employer's published values.
• Source of record: the document that proves a claim, such as a dashboard, ticket, email thread, review, or shipped artifact.
• Probe: a follow-up question the interviewer asks to recover details your first answer left out.

The U.S. Office of Personnel Management (OPM) says every structured interview question should be "reflective of competencies derived from a job analysis", and that each question should be written "to elicit specific details about a situation, task, or context, the actions the person took or did not take, and the impact of these actions." Google's re:Work guide tells interviewers to use behavioral questions "to validate their resume claims". Put those together and the bank has to do two things. It has to be organised by competency, and it has to start from your resume and the records behind it. That is why this playbook mines your history rather than brainstorming examples.

Why behavioral questions dominate

Most published comparisons find that questions about past behavior predict job performance better than hypotheticals. The figures below are validity coefficients, which measure how strongly interview scores correlate with later job performance. Higher means more predictive.

Behavior-description questions lead in 4 of the 5 rows. The last row is real contrary evidence, so treat the advantage as a strong tendency rather than a law. Two findings matter for how you build:

• Experienced candidates should expect past-behavior questions. Huffcutt and colleagues found that job complexity lowers the validity of situational questions but does not affect behavioral ones. In the managerial sample, situational questions did not predict performance at all.
• Anchored scoring is where stories get tested. An anchored rating scale gives the interviewer a written description of what a weak, adequate, and strong answer looks like. With these scales, behavioral questions reached .63. If your story lacks the details the anchors describe, the interviewer will probe until they find them or score you down.

How many STAR stories do you need?

You need 8 to 12 stories for a full loop. Structured interviews typically assess 4 to 6 primary competencies, according to a U.S. Department of Energy candidate module, and two stories per competency leave room for a weak fit or a repeat across rounds. STAR stands for Situation, Task, Action, Result: the four parts a behavioral answer has to cover.

Published counts vary widely:

The median of the upper bounds is 8. The advice splits into two camps:

• Breadth. "Most behavioral interviews include 4-6 questions, so having double that gives you flexibility to choose the best-fit story without repeating yourself."
• Depth. "Four to six stories you know cold, each flexible across several prompts, beats twelve you know shallowly."

My position is that both camps are right about different failures. The depth camp is right that a shallow story collapses under follow-up questions. The fix for that is the probe layer, not fewer stories. The breadth camp is right that a loop asks you many questions. Amazon's guidance, as reported secondhand, has each interviewer ask two or three behavioral questions. A third-party report puts Google's Googleyness and Leadership round at roughly 45 minutes with around 4 to 6 questions. Across four or five interviewers, a bank of five stories forces you to repeat yourself or improvise.

So the formula is 4 to 6 competencies x 2 stories = 8 to 12, each one prepared deeply. No source sets counts by seniority. If you have a single interview this week, MIT's 3 to 5 is a reasonable first cut. Grow it to 8 to 12 before the full loop.

Where your competency list comes from

Your competency list comes from the target postings and the employer's published values, never from a generic list of themes. A generic list will miss the requirements that make a given employer different, and those are often the questions you are least prepared for.

Pull from these places, in this order:

1. The posting's requirements or qualifications section. The DOE module points candidates here. Turn each requirement into the capability it implies. "Partner with sales and legal to close enterprise deals" becomes cross-functional influence.
2. Federal announcements. For federal roles, competencies "may be listed as KSAs or in the 'qualifications required' section of the USAJOBS announcement." KSA stands for knowledge, skills, and abilities.
3. The employer's published values. Amazon publishes 16 Leadership Principles, numbered from Customer Obsession to Success and Scale Bring Broad Responsibility. The list has grown before, so read the live page each time rather than an old copy. One recruiting firm's Amazon kit advises interviewers to ask one or two questions from 3 to 5 of the most relevant principles per interview.
4. The employer's published interview method. Google documents its structured-interview approach on re:Work. Where an employer publishes its method, read it for what it scores.

How often private-sector postings name competencies directly is not established publicly. Check for yourself: read the requirements section of each posting you are interviewing for and count how many lines name a capability rather than a credential. If the answer is few, lean harder on published values and on the recruiter's description of the loop. If your applications already run through Refolk, the fit score it gives each posting shows which requirements your history covers thinly. That is a first draft of the gap list you will confirm in the coverage matrix.

Who is on the other side of the table

Refolk's index shows who claims structured-interviewing skill, and the sample breakdown matters for how you prepare:

The percentages come from 25-profile samples, so read them as indicative only. My reading, which is an inference I have not verified: smaller companies put founders in the loop, so at those companies you may face interviewers with less training and less predictable probes. That makes the probe layer more important there, not less. The concentration of bar raisers at Amazon, which runs dedicated cross-team interviewers, also makes it the cleanest worked example of a coverage matrix. Its 16 published principles turn a target posting into a fixed competency list.

The build, step by step

The build runs in eight stages, from competency extraction to a pressure-tested bank, and each stage has a pass bar you can check. Start from the competencies, then mine your history without filtering, so the postings set the rows and your records supply the stories.

Sources disagree on the order. One guide inventories first and maps competencies afterwards, while OPM and DOE start from competencies. I start from competencies so the matrix rows come from posting text, but I keep the inventory unfiltered so that the list does not narrow what you remember.

On timing: no published source gives per-stage minutes. One vendor guide claims the whole build takes 3 hours. Another spreads it across a week with days 1 to 3 as the core. The minutes below are Refolk's proposed budget, not a measured standard. The first four stages come to roughly 2 to 2.5 hours. Fact sheets, the probe layer, and rehearsal grow with the number of stories, and they deserve the most time because they are where stories get tested.

For the inventory, draw on "situations from your experiences on your resume", as MIT advises, then go beneath the resume to the records behind each line: old performance reviews, launch emails, project trackers, and closed tickets. A resume line usually hides three or four stories. If Refolk wrote your resume from your history, that history is already one place, which makes the first 30 entries mostly a matter of splitting lines into separate incidents.

What goes on a story fact sheet

A story fact sheet is a one-page record of a single story in which every figure, date, and name traces back to a document. The 30-second and 90-second versions are drawn from it. It is the bottom layer of the story, and the layers above it are only as strong as it is.

What each field proves, and how it can mislead:

• The result figure proves impact only if it comes from a dashboard or ticket. A remembered figure misleads because it is usually rounded up and missing its time window.
• Year and scope prove the story is relevant at your target level. They mislead when an old story described in today's vocabulary hides the fact that you were junior at the time.
• Actions starting with "I" prove your personal contribution. They mislead when "I" is grammatically there but the action was a group decision you attended.

No source sets a recency cutoff. Practitioner advice ranges from "the last two or three years at the level you are targeting" to the last five years. Record the year and scope, and let the level test decide.

Story ID: S01
Working title:
Year and duration:
Scope at the time (my level, team size, budget):
Primary competency:
Secondary tags:
Situation (who was involved, what led up to it):
Task (what I was accountable for):
Actions (each line starts with "I"):
- I
- I
- I
Result (number, unit, time window):
Source of record for the result:
Other people in the story and their roles:
What I would do differently:
Failure or conflict story? (yes/no):
30-second version:
90-second version (bullets only):

How to map interview stories to competencies: the coverage matrix

The coverage matrix is a grid with each loop's required competencies as rows and your bank stories as entries. Every required competency must show at least two stories that hold it as their primary tag. Any row that falls short is a named gap with a specific plan to fill it.

Build it in this order:

1. Rows come from posting text. Use the posting's or the employer's own wording. Where two loops name the same capability in different words, merge them into one row and note both phrasings.
2. Count primary tags only. Secondary tags tell you which stories you can stretch in the room. They do not count toward coverage.
3. Apply the bar per loop. Practitioner guidance says "every common competency should have at least one story, ideally two." I treat two as the bar, because a single story leaves you nothing if it lands badly or has already been told earlier in the loop.
4. Name every gap with a fill plan. The options, in order of preference: go back to the inventory and look for a missed entry, promote a story whose secondary tag is strong enough to rewrite as primary, use a well-documented non-work story, or, as a last resort, prepare an honest answer that explains how you would approach it.

Deciding what makes the cut

Going from 30 entries to 8 to 12 is a judgement between two variables: how much your target loops want a story's competency, and how well its claims are documented.

Before you finalise, check the cut contains at least one failure story and one conflict story. As one guide puts it, "Interviewers will ask about failures, conflicts, and mistakes. A bank of only wins leaves you improvising the hardest questions."

Competency (posting wording) | Loop | Primary stories | Count | Secondary-only | Gap? | Fill plan
Customer Obsession | Amazon | S03, S07 | 2 | S11 | No | -
Frugality | Amazon | S05 | 1 | S02 | Yes | Re-mine inventory for cost entries; promote S02 if its record holds
Cross-functional influence | Loop B | S01, S09 | 2 | S04 | No | -
Handling ambiguity | Loop B | - | 0 | S06 | Yes | Rewrite S06 with ambiguity as primary; confirm dates in project tracker
Failure / mistake | All loops | S08 | 1 | - | Yes | Add a second failure from inventory with a documented lesson

How to prepare for behavioral interview follow-up questions

Prepare a probe layer for every bank story: written answers to the follow-ups structured interviewers ask under situation, action, and result, at least two to three levels deep. OPM tells interviewers to "make a list of acceptable probe questions for every competency interview question", so the follow-ups are part of the scoring, not small talk.

"What was your role?" appears in the interviewer guides from OPM, Harvard Medical School HR, Caltech, Binghamton, UVA, and LinkedIn. A ranked frequency count of probes is not established publicly, but that one recurs across all of them. OPM explicitly tells interviewers to "seek clarification when candidates say 'we did…' or 'our group…'", so write your we-to-I answer before anyone asks for it.

The follow-up questions are the scoring layer. A story that only works without interruptions is a script, not evidence.

Rehearsal and the pressure test

Practise the 30-second version until you can open any answer with it. Practise the 90-second version from bullet points, because one practitioner warns that scripting or memorising stories "may limit your ability to adapt." Then run a mock interview in which a friend interrupts each story twice with "why?" A story passes when you can answer both interruptions from the fact sheet and get back to the result without losing the thread. Founders and executives make up a large share of self-identified behavioral interviewers in Refolk's sample, and they are less likely to follow a probe script, so this unscripted test is the closest you will get to the real room.

How a story bank goes wrong

Story banks fail in seven predictable ways. Each one looks fine on paper and only shows up under follow-up questions. Check every story against all seven before the bank counts as finished.

1. "We" stories. The story describes the team's work, and "What was your role?" exposes it within one probe. OPM trains interviewers to push on exactly this. Check: every Action line on the fact sheet starts with "I".
2. Unanchored numbers. The result figure is remembered, not documented, and a dig-deeper probe ("what did you find when you looked into it?") breaks it. The figure looked convincing until someone asked where it came from. Check: every figure on the fact sheet names its source of record.
3. Inflated coverage. One versatile story is tagged with five competencies and counted five times, so the matrix looks full while several rows have no real story. Check: recount using primary tags only.
4. Only wins. The bank has no failure story, so you improvise on "what would you have done differently?", the question where improvising costs most. Check: at least one failure story and one conflict story are in the bank.
5. Over-rehearsed scripts. Smooth delivery looks like readiness, but a story that "collapses under two follow-ups" was polish without depth. Check: a mock interview with two why-interruptions per story.
6. Generic competency lists. Mapping to a generic six-theme list misses posting-specific requirements, such as Frugality at Amazon. Check: every matrix row traces to posting text or the employer's published values.
7. Stale or wrong-level stories. A story from six or more years ago, or one pitched below the target level, gets probed for scope you did not have at the time. Check: each story is tagged with its year and scope, and passes the level test.

Failure modes 2 and 3 are the dangerous pair. Together they produce a bank that looks finished and checks out at a glance, but fails the first time someone probes a number in a competency you thought you had covered.

Keeping the bank current across loops

The bank is a living document. Update it after every round, and re-run the competency extraction whenever a new posting enters your pipeline. A bank built for last month's targets will drift out of coverage as your targets change.

After each interview, note which stories you told, which probes landed hard, and which competencies came up that your matrix did not predict. Feed those back in. Add a probe answer where a follow-up surprised you, fix a fact sheet where a figure felt shaky, and add a matrix row where a new competency appeared. When a new posting arrives, extract its competencies, merge them into existing rows where the wording maps, and check whether any row has fallen below two primary stories. When you tailor your resume for that posting, in Refolk or by hand, check that the claims it now emphasises have a matching bank story. Interviewers use behavioral questions to validate resume claims, so every emphasised line should have a story ready behind it.

Once the bank passes the checklist below, it is ready for the per-question work: the Behavioral Story Pick decides which story leads for a given question, and the Behavioral Answer Standard grades how you tell it.