Week 7 — Performance & Calibration

60-Second Summary

Week 7 of the 12-week program. Theme: Defensible ratings, defensible promotions.
Monthly performance journal entry per report — the ritual you install this week.
60 min pre-read + 90 min cohort + Friday homework with a falsifiable artefact.
Reviewed by HR Director, line manager, and OB faculty lenses.

Calibration is where managers reveal whether they actually manage. A weak manager goes into calibration with vague impressions ('she's a strong performer'); a strong one goes in with specific evidence, observed behaviour across the cycle, and a level case that a peer can challenge productively. Calibration is also where bias is most likely to enter — and where it's most expensive when it does.

What the evidence says

Castilla (MIT) on the paradox of meritocracy: when organisations believe they are meritocratic, biased decisions actually increase. Structured calibration is a partial corrective.
Kahneman et al. (Noise): 'noise' — variation between decision-makers given the same evidence — accounts for at least as much error in performance ratings as systematic bias does.
Performance research consistently finds: ratings from a single manager have low reliability; calibration across 3+ managers + structured evidence raises reliability substantially.

Pre-read (60 minutes)

Read: Calibration sessions run well — 20 min.
Read: Writing a level case (the promotion packet you wish your manager had written for you) — 20 min.
Read: Five biases that show up in calibration (recency, halo, similar-to-me, severity, leniency) — 15 min.
Reflect (10 min): your direct reports. Without thinking too hard, rank them. Now name three pieces of evidence for each ranking. If you can't, you have a calibration problem.

Live session (90 minutes)

Cohort flow with a senior-manager coach

1
The level case (25 min)
Coach walks through writing a level case: behavioural evidence per competency, specific projects, impact, and trajectory. Cohort works on one of their own reports live.
2
Bias hunt (15 min)
Each manager identifies which biases they're most prone to. Coach shares mitigations: blind review of ratings before discussion, calibration on specific competencies first, attention to under-represented voices in the room.
3
Calibration sim (35 min)
Cohort runs a mock calibration with a coach playing the skeptic. Each manager defends a rating; coach challenges with 'where's the evidence?' and 'have you compared to [X] across teams?'.
4
Promotion conversations (15 min)
Practice two conversations: 'you're getting promoted' (specific, evidence-rich, not a coronation) and 'you're not getting promoted this cycle' (specific, with a development path, not a soft no).

The ritual you install this week

Monthly performance journal entry per report

Once a month, spend 5 minutes per report writing: what evidence have I observed this month — strengths and developmental? File it. At performance time, you have 6 months of dated, specific evidence rather than 'the last two weeks I happened to remember'. This single ritual transforms calibration quality.

Modern tools for this skill

Category	Examples (2026)	Use
Performance platform	Lattice, Leapsome, Culture Amp Perform, Workday Talent	Reviews, calibration, 360
Calibration prep	Notion DB, Lattice talent profile	Evidence per competency
Continuous notes	Tability, 15Five, weekly journal	The monthly entry above
AI evidence summariser	Claude/ChatGPT on 1:1 + project notes	Draft a level case for you to refine

Copy-paste AI prompt

Here are my 1:1 notes and project updates for [report] over the last 6 months [paste]. Write a draft level case in the format: competency, observed behaviour with date, impact, gap to next level. Flag where I have insufficient evidence and need to gather more.

Friday homework — falsifiable artefacts

Level case written for one report — at least one piece of dated evidence per competency.
Monthly performance journal started — one entry per report in your private notes.
Calibration prep doc completed for your next review cycle.
Self-audit completed: rate your reports, then check whether you have evidence — or just impressions.
Submitted to coach: redacted level case + your self-audit findings.

Success signal

By end of week 7, you can defend any rating for any report with at least three pieces of dated, specific evidence. In a calibration challenge, you change your view when challenged by better evidence — and you challenge others' views with the same standard.

Reviewer notes

HR Director (15+ yrs)

When a performance decision ends up in a dispute, the only thing that matters is contemporaneous documentation. A manager who has 6 months of monthly entries with dated, specific evidence walks into the dispute prepared. A manager who has 'it was clear to everyone' walks into a fair-process complaint that the company often loses.

Line Manager (20+ yrs)

I journal 5 minutes per report per month. Sounds trivial. It's the single highest-ROI habit I've kept in 20 years. Performance reviews go from days to hours, conversations are richer, and I never have to bluff a manager-of-manager when asked about a specific person.

OB / HR Professor (25+ yrs)

Kahneman's 'noise' work has reshaped how we think about evaluation. Two managers given the same information often rate differently — that's noise, not bias, and it requires different mitigations: structured rubrics, independent ratings before discussion, and calibration across raters. The monthly journal addresses the most common cause: recency-weighted memory.

Written by Pawan Joshi.Sources cited inline.

First published 23 Jun 2026See site changelog →