CoachNed
Two empty armchairs and coffee cups

Firm guides

McKinsey PEI Rehearsal: Story Bank, Probe Ladder, Scores

Build a McKinsey PEI story bank, drill a follow-up probe ladder, self-score each rep, and run a two-week plan on CoachNed's four-probe simulator.

UpdatedReviewed by Ned

Most candidates polish a 90-second opener until it sounds like a TED talk. The interviewer scores the four follow-ups after that. If your story dies when asked what the other person said next, the polished opener was wasted airtime.

This page is the rehearsal system only: how to build a story bank, climb a probe ladder, self-score every rep, and run a two-week schedule that includes CoachNed's PEI simulator at /behavioral. For what the format is and which dimension each prompt hits, use the PEI guide and the question bank. After this, you should be able to walk into a loop with four stories that survive depth, not eight summaries that sound good once.

Build the bank before you rehearse anything

McKinsey's interviewing page tells candidates to prepare two personal examples for each of the four areas the firm assesses, and to discuss experiences in detail with focus on your specific role and the key actions that mattered. That is a depth requirement, not a cue to write eight speeches.

Start with inventory, not prose. Write every candidate experience on one line: setting, the decision you owned, the friction, and one checkable result. Then run four filters. Keep only stories that pass all four.

FilterKeep if…Cut if…
OwnershipYou can name what would have failed without youThe honest pronoun is "we" for every beat
FrictionSomeone resisted, delayed, or disagreedThe project ran smoothly and you coordinated
RecallYou still remember what was said in the roomYou only remember the outcome slide
RecencyRoughly the last two to three yearsYou need a script to reconstruct the details

Aim for four primary stories first (one that can carry Leadership, Connection, Drive, or Growth), then add a second example per dimension as insurance. McKinsey suggests two per area; a colleague who passed interviews for the Vietnam office described picking situations she was proud of, then practicing them orally alone and with friends rather than memorizing a script (Jenny Phung, Careers Blog). Depth on four beats thin coverage on eight.

Map each keeper to a primary dimension and a backup. Do not force a campus club story into Drive if the real material is Connection. The bank is a casting list, not a spreadsheet you fill for symmetry.

The follow-up probe ladder

Yale's Office of Career Strategy tells students to build a bank of stories and practice them aloud with a partner, not on paper only (Preparing for the Behavioral Interview). McKinsey PEI goes further: the interviewer stays on one story and climbs. Rehearse the climb, not the first paragraph.

Use this ladder on every story in the bank. Answer each level cold, out loud, in under 60 seconds. If you stall, write the missing fact before you polish language.

LevelWhat the interviewer is testingSample probePass signal
0OpenerTell me about a time you…60–90 seconds; stakes clear; you own the decision
1JudgmentWhy that option, not the other two?You name rejected options and a trade-off
2OwnershipWhat did you do that nobody else did?A concrete action only you took
3Other peopleWhat did they say, and what did you do next?Dialogue, not "they were difficult"
4ReflectionWhat would you change now?A specific change you already applied once

McKinsey's own tip sheet is blunt about levels 2 and 0: focus on your specific role, spend most of the time on critical actions, and resist a long setup (Thursday Tip). In the room I stop listening when level 3 gets a character summary instead of a line of dialogue. "She was skeptical" is not a PEI answer. "She said the pilot would make her look indecisive to her VP" is.

Build a probe card per story: six to ten questions across the ladder, written in the interviewer's voice. Rehearse the card before you re-rehearse the opener. Candidates who reverse that order usually discover their holes on interview day.

Score the rep, not the feeling

After each rehearsal, score yourself on a 12-point sheet. Do not grade "confidence." Grade the evidence the interviewer can write down.

Criterion012
Setup lengthOver 60 seconds of context45–60 secondsUnder 45 seconds, stakes clear
Personal actionTeam verbs dominateMix of "I" and "we"Clear "I" actions with timing
Friction namedNo resistanceVague pushbackNamed person, objection, your reply
Checkable result"Went well"Soft outcomeNumber, decision reversed, or durable change
Level-3 survivalStall or inventGeneric recoverySpecific dialogue and next move
Level-4 reflection"Nothing, it was perfect"Vague lessonConcrete change already used again

Worked example. Same Connection story, two reps, scored cold.

Rep A (scripted opener, thin under probes): setup 70 seconds on the client org chart; actions described as "we aligned stakeholders"; when asked what she said, "she was hesitant"; reflection "I'd communicate more." Score: Setup 0, Action 0, Friction 1, Result 1, Level-3 0, Level-4 0 = 2 / 12.

Rep B (same night of your life, probe-first practice): setup 35 seconds; you ran the usage cut overnight and proposed a two-week pilot framed as protecting her VP commitment; she said reversing the full launch would look like she had mismanaged the plan; you walked her deputy through the pilot first; the pilot caught a pricing error before rollout; next project you flagged the same risk at draft stage. Score: 2+2+2+2+2+2 = 12 / 12.

The story did not change. The rehearsal target did. Track scores in a simple log: date, story, dimension, total, weakest criterion. Retire a story only when it clears 10 / 12 twice with different probe orders.

Ned's rule. If you cannot answer "what would have happened if you had stayed quiet" in one sentence, you do not have a PEI story yet. You have a project summary. Fix the story before you fix the delivery.

A two-week practice schedule

Fourteen days is enough if you stop rewriting openers and start climbing the ladder. Keep case practice separate; this block is PEI only.

DayFocusOutput
1–2Inventory and filters8–12 candidate lines; cut to 4 keepers
3Bank mapPrimary + backup dimension per keeper
4–5Probe cards6–10 probes written per story
6Cold ladder, Story AScore sheet; fix fact gaps
7Cold ladder, Story BSame
8Cold ladder, Stories C and DSame
9Shuffle probesSame stories, new order; no notes
10Partner or simulatorFull PEI slot on two stories
11Fix weak criteriaOnly the rows that scored 0–1
12Second simulator passDifferent dimension prompts
13Light openers only90-second versions; do not re-script
14Rest the voiceOne skim of probe cards; sleep

A McKinsey colleague who prepared personal-experience examples for Latin America interviews described choosing roughly five experiences, writing short precise stories, then practicing repeatedly with feedback from people who would interrupt (Andrés, Careers Blog). That interrupt is the product. Silent reading does not create it.

If you have less than two weeks, compress days 1–5 into one evening and spend every remaining session on levels 2–4. Openers are cheap; dialogue under pressure is not.

Use CoachNed's PEI simulator in the free week

When you need a partner who will not politely stop after your opener, use /behavioral. The PEI simulator stays on one story and runs four follow-ups, which matches the probe depth that empties thin banks. Everything on CoachNed is open for seven days with no card; after that it is $120 for a recruiting season or $49 a month.

How to use the free week without wasting it on scripts:

  1. Day 1 of access: load one keeper story. Take the four follow-ups cold. Score the sheet immediately.
  2. Day 2–3: same story, different prompt wording for the same dimension. You are training transfer, not memorization.
  3. Day 4–5: rotate the other three keepers. Log scores.
  4. Day 6: re-run the weakest story only. Stop when it hits 10 / 12.
  5. Day 7: one full pass per story, then close the tab. Do not binge openers.

Free live AI interview

Rehearse with a live AI interviewer

A live behavioral interview with follow-up questions. First question is free.

Start free

Live voice · Question 1 of 6

Pair the simulator with the free PEI workbook if you still need structured slots to fill. The simulator drills depth; the workbook fills the bank. Neither replaces reading McKinsey's own interviewing page once.

Your assignment for tonight

Pick one story that already passed the four filters. Write a probe card with at least two questions at each of levels 1–4. Record yourself answering the opener once, then answering the eight probes without pausing the recording. Score the 12-point sheet. If any criterion is 0, write the fact before you record again. Tomorrow, run the same story on /behavioral and compare the two scores.

CoachNed is not affiliated with McKinsey & Company.

Frequently asked questions

How many PEI stories should I rehearse for McKinsey?

McKinsey suggests two examples for each of the four areas it assesses. In practice, deepen four keepers first until they survive levels 3 and 4, then add backups. Eight thin stories lose to four deep ones.

How long should each PEI rehearsal take?

Budget 12 to 20 minutes per story once probes start: roughly 90 seconds for the opener, then short answers to a ladder of follow-ups. A five-minute monologue with no interruptions is not a PEI rehearsal.

Can I practice PEI alone if I have no partner?

Yes, if you force interruptions. Write the probe card, shuffle it, answer cold on a recording, and score the sheet. A tool that asks four follow-ups, such as /behavioral, closes the gap faster than silent review.

Should I memorize my PEI answers word for word?

No. McKinsey colleagues who write about preparation warn against becoming scripted; practice orally so the narrative stays conversational (Jenny Phung). Memorize facts and decisions, not sentences.

What score on the self-sheet means I am ready?

Treat 10 / 12 twice, with probes in a different order, as the bar for a keeper. Anything under 8 usually means a missing fact, not a delivery problem.

Sources