DEI Without Depth Is Decoration

Most DEI programmes at scale produce measurable improvements in survey scores and no measurable improvement in senior-level demographic composition. This is not because the programmes are insufficiently rigorous. It is because the work they are designed to do cannot actually address what is producing the outcomes they are trying to change.

The issue is not programme design. It is the gap between procedural interventions and psychological ones.

What the current model is doing

The standard DEI intervention operates at the procedural level: policies, training, representation targets, recruitment criteria, diverse slate requirements, unconscious bias awareness. These interventions address what organisations can procedurally control. They are, within that frame, often well-designed.

What they do not address — because they cannot, by construction — is the psychological material that actually produces the outcomes at the senior layer. The procedural interventions create the conditions under which diverse talent can enter an organisation. They do not address the conditions that determine whether diverse talent progresses within it. Progression, at senior levels, is governed by judgements that are considerably less procedural than entry.

These judgements are made by panels of incumbent leaders, most of whom have received unconscious bias training, most of whom report that they do not hold significant bias, and most of whom nevertheless consistently promote candidates who resemble the existing senior profile more than candidates who do not. The gap between the training and the outcome is not hypocrisy. It is the specific limit of what procedural intervention can achieve.

Why bias awareness does not address bias

Unconscious bias training teaches participants to recognise bias in the abstract. The format is usually: here is a common pattern, here is how it operates, here is what to look for in yourself. Participants learn the concepts, can usually describe them accurately, and often report a meaningful shift in awareness.

What the format cannot do is make the participant’s own bias visible to them in the specific moment it operates. This is not a failure of the curriculum. It is a structural feature of what bias is. Bias, by definition, is not available to the conscious self at the moment it influences a judgement. The conscious self experiences the judgement as a reasoned assessment of the specific case. The bias that shaped the assessment is, at the moment of assessment, invisible.

Training teaches the pattern. The specific instance remains undetectable from inside the person making the assessment. Which is why unconscious bias training, however well-delivered, does not address the bias in the specific decisions that determine senior composition.

What would address it is different. It would address what the bias is defending against in the person making the judgement. The senior leader who cannot recognise their bias in the moment is not cognitively deficient. They are, usually, carrying unresolved material that the bias is serving to protect. Until the underlying material is addressed, the bias continues to operate, undetected, in each specific case.

The psychological material underneath

Three patterns recur with enough frequency, in my experience working with senior leaders at close range, to be worth naming.

Disowned qualities. A leader has spent decades not expressing a specific quality — ambition, directness, conviction, authority, vulnerability — because their early professional conditioning taught them it was dangerous to express. When they encounter a candidate who carries that quality competently, they experience friction that reads, to them, as a response to the candidate. Underneath, it is the activation of the material the leader has been organising their professional identity around not expressing. The candidate who would have been promoted is read as “not quite right” for reasons the leader cannot articulate.

Cultural specificity treated as universality. A leader has built their career operating within a specific cultural performance of leadership — a particular register of confidence, pace of speech, style of assertion, form of intellectual engagement. They have, over time, come to experience this performance not as one cultural style among several but as what leadership itself looks like. Candidates whose leadership expresses through different cultural registers are experienced as lacking something the leader cannot quite specify. The specification is absent because what is missing is conformity to a specific cultural performance, which the leader has never held as contingent.

Incumbent class maintenance. A leader is carrying, usually below conscious recognition, a specific investment in the maintenance of their current class position. Advancing candidates who genuinely differ from the incumbent profile threatens that position in ways the leader is not consciously aware of. The threat registers as discomfort, discomfort registers as assessment of fit, assessment of fit produces the decision not to advance. The class dimension is never named. The outcome replicates the class.

None of these patterns is addressable by training. All of them are addressable by the developmental work that unconscious bias training is not designed to do.

What would work at senior level

The intervention that would actually move senior demographics is not a training programme. It is depth work with the leaders making the senior advancement decisions — sustained engagement with the psychological material that produces their assessments, conducted with the specificity and duration that genuine integration requires.

This is considerably more expensive than training. It is also considerably more uncomfortable. The leader who enters this work discovers, reliably, that their previous assessments of candidates who did not advance were shaped by material they would not have defended if asked to articulate it. This is not a comfortable discovery. For most leaders, it produces a period of significant reconsideration.

The organisations that have genuinely shifted senior demographics over sustained periods — not in the recruitment funnel, but in actual senior composition — have almost always combined procedural intervention with something developmental at the leadership layer. The ones that have only done procedural work have almost always seen the scores move without seeing the composition move.

This pattern is visible in the data. It is rarely named because naming it would require commissioning an expensive intervention rather than continuing to commission the less expensive one that produces scoreboard motion.

The honest question for organisations

Before the next DEI investment cycle, ask one question honestly.

In the last three years, how many senior leaders in this organisation have done serious depth work — sustained, structurally supported, psychologically specific work on their own judgement apparatus?

If the answer is zero or near-zero, the DEI programme is operating without the layer that would allow it to produce senior-level change. That is a choice. It is defensible. But it is important to make it consciously, rather than to believe that the procedural interventions will, on their own, accomplish what they structurally cannot.

Organisations that are serious about senior composition eventually come to this. The ones that do not remain in the pattern of rising training scores and stable senior demographics for years, occasionally commissioning larger versions of the interventions that have not worked before.

The cheaper answer is to recognise the gap earlier and invest in what closes it.