Knowles Knowledge Centre•Role Mapping
A Record Audit Matrix for Employees, Managers and Leaders
The same five stages look different at each level. This matrix sets out what each one actually looks like as observable behaviour.
Organisations frequently commission the same bias development for everyone and are then surprised that it lands unevenly. The cause is usually not the content. It is that a stage such as “fix the rule” describes a different behaviour for an individual contributor, a line manager and a senior leader — and only one of the three can remove a discretion that will not hold.
This guide maps the five stages across three organisational levels, states the observable behaviour at each intersection, and identifies where development priorities genuinely differ.
About this guide
Why one bias course for everyone underperforms
The method is shared. The situation in which it has to be used is not.
A manager taught an individual contributor’s version of this method will audit their own allocations conscientiously and go on accepting their reports’ accounts of theirs, which is the one thing the method says cannot be relied on.
The stage matrix
Five stages by three levels. Each cell describes observable behaviour — what somebody at that level visibly does when the stage is being done well.
Stage
Employee
Manager
Leader
Say the rule
Writes one scoreable sentence, dated, before opening any record
Asks for a written rule rather than an assurance of fairness
Requires every discretionary decision in a process to have a written rule somewhere
Pull twenty
Takes the last twenty in sequence, including the awkward ones
Ensures the record exists to be pulled, and that it is not the decider’s own notes
Funds the recording of decisions that currently leave no trail
Score them
Hands their rows to somebody else rather than marking them alone
Scores a report’s rows against the written rule, declining the circumstances
Treats a scored audit as ordinary management information, not as an incident
Face the gap
States the number before accounting for any of it
Lets the number stand when it is explained at them for ten minutes
Publishes their own audit before asking anybody else for theirs
Fix the rule
Amends their practice, or amends the rule and defends it openly
Decides which of the two the team is actually going to do
Removes the discretion entirely where the practice will not hold
Read down a column for a level’s development profile. Read across a row to see why a single course cannot serve all three.
Where the levels genuinely diverge
Three transitions cause most of the difficulty, and none of them is solved by more content.
From auditing yourself to scoring somebody else
The move from individual contributor to manager changes the unit from “my rows” to “theirs.” A newly promoted manager who audits honestly will often score a report’s rows generously, because they know the circumstances of each case and because marking a colleague down feels like an accusation.
It is not an accusation, and the generosity destroys the instrument. Development at this transition is almost entirely about marking against the written rule and declining the explanation — which is mechanical, uncomfortable, and the whole reason a third party scores the rows at all.
From scoring rows to removing discretion
The move to senior leadership shifts the unit again, from a record to a design. A leader who receives a dozen audits all showing the same gap on the same rule, and responds by asking people to try harder, has chosen the one option the evidence does not support.
This transition is where the fifth stage matters most, because an individual can amend their own practice and only somebody at this level can conclude that a discretion should not exist.
Exposure inverts the usual pattern
For most of this method, latitude increases with seniority. The willingness to publish a number about yourself frequently does not. A junior analyst’s audit is a development exercise; a director’s is a document, and the more senior the decision, the less often it is recorded at all.
The development need is therefore reciprocal rather than hierarchical: junior staff usually need a record that exists and a scorer who is not them, and senior staff usually need to notice that their own most consequential decisions are the least audited in the organisation. Neither is served by the same session.
The three levels are not stages of increasing skill. An employee who writes a rule, pulls twenty rows, has them scored and amends their practice has completed the method. Treating the leader column as “advanced” produces development that is irrelevant to the people making most of the decisions.
Using the matrix to set priorities
The matrix is a scoping tool. Used properly it narrows a programme rather than expanding it.
Start from the noticed pattern, not from the grid. Identify the distribution somebody has remarked on, then locate the one or two cells describing the behaviour that was missing. A programme built from a whole column will be too broad to change anything.
Check the level before the stage. When a team’s audits all show the same gap, establish first whether the rule is workable. Developing the deciders will not help if the rule has never been applied faithfully by anybody.
Expect different formats, not just different content. Employee-level behaviour responds well to one real audit of the participant’s own last twenty rows. Manager-level behaviour needs practice in scoring somebody else’s rows while they explain the circumstances, which is uncomfortable and has to be rehearsed. Leader-level behaviour is usually better served by auditing one decision they make themselves, and publishing it, than by a course at all.
A practical sequence: diagnose the gap using how to identify the gap between your stated rule and your last twenty decisions, locate the affected cells in this matrix, then scope development to those cells only.
Frequently asked questions
Conclusion
The five stages do not change with seniority. What changes is whose record is being counted:
Most disappointing bias programmes trace back to a mismatch between the level the programme was designed for and the level the audience actually occupies.
Naming the cell before commissioning the work removes most of that risk, and usually makes the programme smaller rather than larger.
The underlying five-stage structure is set out in you cannot inspect your own bias, but you can audit your record. Organisations that have identified the level and the stage they need can review the Understanding Unconscious Bias Training Course in Singapore, checking that the design matches the audience level rather than the method label.

