Why behaviour-only 360° feedback misses part of the picture

What a classic 360° actually measures

Most 360° instruments are competency models with a questionnaire: they define the leadership behaviour the organisation wants to see and ask supervisors, peers and direct reports how often they observe it. That is not a flaw; it is the definition. Bracken, Rose and Church (2016) define 360° feedback as a process for collecting, quantifying and reporting co-worker observations about a person – raters’ perceptions of specific behaviours. Observation is the raw material. Nothing more was ever promised.

DefinitionBehaviour-only 360° feedback

A multi-rater process in which several groups of raters score a leader’s observable behaviour against a competency model, usually alongside a self-assessment on the same items. It answers one question: what leadership behaviour do others observe? Why it occurs and when it flips are not questions it asks.

A good 360° describes this outer ring reliably, given enough raters. What such a report says is rarely wrong. What it does not say is the problem: it shows a pattern and is silent about its cause. "Delegates too little" is an observation. Whether a missing skill sits behind it, a belief, or a pressure reflex that only appears in crunch weeks – the report has no data on that, and that is where development starts. What is 360° feedback? covers the basics.

Three things a behaviour report cannot show

1. The belief behind the pattern

Leadership behaviour follows beliefs – about the role, the team, what happens when you let go. Kegan and Lahey call these competing commitments: a leader wants to delegate and is at the same time, usually unspoken, committed to never losing control of an outcome. The second commitment wins when it matters. Others cannot rate this layer: beliefs are not observable from outside. You have to ask the leader – separately from behaviour, or they rate their own behaviour a second time. More in Beliefs versus behaviour.

2. The capability that exists but is not used

A behaviour report cannot tell "cannot" from "does not"; both appear as low frequency. A leader who lacks a skill needs training. A leader who has it and does not call on it under certain conditions needs an understanding of those conditions. Bandura’s self-efficacy work offers the way in: whether a leader believes they can show a behaviour is a question of its own. Only comparing that confidence with what others observe reveals whether this is a training, a calibration or a retrieval problem – see The gap between capability and behaviour.

3. The pressure condition under which the pattern flips

Almost every pattern that shows up as a weakness in a 360° is conditional: it appears when the deadline closes in or the client escalates – and disappears when the pressure lifts. A frequency question over the past months averages that condition away: "controls everything in crunch weeks" and "otherwise gives plenty of room" become an unremarkable mean. The clinical, psychodynamic approach to leadership that Kets de Vries stands for takes exactly these reaction patterns seriously – with the assumption that under pressure people fall back on older strategies. Which ones they are cannot be observed – but it can be asked; see Leadership under pressure.

Why this matters for development

Feedback without a why produces resolutions. The leader reads "delegates too little", resolves to delegate more, and does – until the next crunch week. Then the pattern returns, and the next report looks like the last. That is not a character flaw but the predictable outcome of an intervention that names a symptom and leaves the cause untouched.

The research supports this scepticism more clearly than the industry would like. Kluger and DeNisi (1996) showed across 607 effect sizes that feedback interventions improve performance on average (d = .41) but that more than a third made performance worse. Their feedback intervention theory explains when: effectiveness declines as attention shifts from the task towards the self. A behaviour-only report is easily read as a verdict on the person: it offers no mechanism to work on.

Smither, London and Reilly (2005) meta-analysed 24 longitudinal studies of multisource feedback. Ratings from others improved only modestly afterwards: corrected effect sizes of .15 for direct reports, .05 for peers, .15 for supervisors and −.04 for self-ratings, with confidence intervals including zero for every source except supervisors. The authors warn against expecting large, widespread improvement. Change becomes more likely when the feedback signals that change is needed and recipients see it as necessary and feasible, set goals and take action. Two studies they cite found that managers who worked with an executive coach improved more.

Five questions instead of one

If behaviour is the end of a chain, an instrument has to ask along the whole chain. For us that means five questions; only the leader can answer the first and the third.

What a whole-system 360° asks differently

LEADBeyond 360° is a consultant-led 360° leadership assessment built along this chain – Belief → Capability → Trigger → Behavior → Outcome, with Well-being as a sixth layer and energy buffer. Only the leader answers the Belief and Trigger items. The Trigger layer asks in ten items which reaction patterns take hold under pressure, grouped into five defence patterns: Over-Control, Pleasing, Over-Performance, Distancing and Avoidance. Capability is captured in the self-view as confidence and compared with what raters observe; the difference is read as a calibration question, not as a verdict on untapped potential. Details on the methodology page.

Where even a whole-system 360° reaches its limits

• Self-report stays self-report. Only the leader knows their beliefs and triggers – and they can be wrong, self-protective or aspirational. These layers are not an X-ray but a structured self-description that gains weight only against the view from outside.
• Links are hypotheses. That a belief blocks a behaviour is an assumption of the model, tested in the conversation – not an empirical finding about this one leader.
• Self–other agreement has inconsistent findings. Fleenor and colleagues (2010) show that interest in self–other agreement stems from its purported links to self-awareness and leader effectiveness, but that the field has used inconsistent metrics and produced discrepant findings. A gap is a prompt for conversation, not a metric – see Self-image and the view of others.
• More layers do not cure a bad process. Too few raters, distrusted anonymity or a report without a conversation remain problems – see How a 360° assessment works.
• Consultant-led means: not for every use case. A whole-system 360° is built for development – not for unaccompanied mass roll-outs, not for selection or promotion decisions. Free-text comments are not analysed by AI in the current version.

Is it worth asking about the why, if all it yields is hypotheses? Yes – because the conversation that follows them is a different one. A report that says "you delegate too little" ends in a resolution. A report that asks "what do you believe happens when you let go – and exactly when do you step back in?" is where development begins.