The feedback paradox

In 1996, researchers Avraham Kluger and Angelo DeNisi published a finding that should have fundamentally changed how organisations think about feedback.

Their meta-analysis, involving 607 effect sizes and more than 23,000 observations, found that feedback improved performance on average, with a moderate positive effect (d = 0.41).

But there was a catch. More than a third of the feedback interventions actually made performance worse.

Not simply less effective. Worse.

Thirty years later, much of our response has been to train managers to deliver feedback more frequently, more constructively and more skilfully.

Certainly, we have become better at teaching the techniques. We have models for structuring conversations, frameworks for difficult feedback, and guidance on being specific, timely and balanced.

But have we paid enough attention to the conditions in which those conversations take place?

Kluger and DeNisi’s research suggested that feedback becomes less effective when it shifts attention away from the task and towards the individual’s sense of self. How feedback is received, and where it directs attention, matters enormously.

We would argue there is a broader organisational question we have been neglecting.

What if the problem isn’t primarily that managers don’t know how to give feedback, but that we haven’t created the conditions in which good feedback can happen?

In 2004, researchers Lisa Steelman, Paul Levy and Andrea Snell introduced the concept of the feedback environment.

Their research identified seven characteristics of the everyday conditions that shape how feedback works between managers, employees and colleagues.

These go well beyond the quality of an individual conversation.

Now consider which of these seven dimensions a typical feedback skills workshop can meaningfully improve.

Delivery? Certainly.

Quality? To a considerable extent.

But what about the rest?

Availability depends on managerial workload and competing priorities. Credibility is influenced by trust, competence and how managers are selected and supported. Feedback-seeking depends on workplace norms and incentives. The willingness to offer or receive difficult feedback depends partly on psychological safety.

These are not simply individual skills. They are also properties of the environment in which people work.

Yet we continue to invest heavily in improving the conversation while paying considerably less attention to the conditions surrounding it. We have been treating a problem of organisational capability and capacity as though it were primarily a problem of individual technique.

If you think about it most organisations have designed their feedback processes around the line manager.

Managers conduct performance reviews, hold development conversations, allocate work, assess performance and influence career opportunities. Colleagues, customers and others can provide valuable feedback. But the manager remains the dominant channel through which feedback becomes part of formal performance and development systems.

And that manager channel is under increasing pressure.

Gallup’s State of the Global Workplace 2026 report found that global employee engagement fell to 20% in 2025.

More concerning was the decline among managers.

Manager engagement fell from 31% in 2022 to just 22% in 2025. Over the same period, engagement among individual contributors declined by only one percentage point.

The traditional engagement advantage enjoyed by managers has almost disappeared.

Think about what that means for the feedback environment.

A manager struggling to keep up with operational demands may simply have less time to observe performance, offer coaching or follow through on development conversations.

A manager absorbed in constant change may have less energy to provide thoughtful, timely feedback.

A manager who feels unsupported may find it more difficult to create the trust and openness needed for people to raise concerns or admit mistakes.

This isn’t to suggest disengaged managers are incapable of giving good feedback. Nor does Gallup’s engagement data directly measure feedback quality.

But it raises an uncomfortable question.

Are we building our feedback systems around the very people whose capacity to sustain them is being steadily eroded?

There is further evidence worth considering.

A 2021 meta-analysis involving 112 independent samples and more than 31,000 workers found that a positive feedback environment was associated with better supervisor-rated performance, stronger workplace relationships and lower burnout.

The relationship with burnout was particularly substantial, with a correlation of −0.51.

That does not establish causation. But it does suggest that the quality of the feedback environment is closely connected to people’s experience of work.

My interpretation is that the relationship may be mutually reinforcing.

Managers under pressure struggle to create positive feedback environments. Poor feedback environments, in turn, can add uncertainty, frustration and stress for the people working within them.

The result can become a self-reinforcing cycle.

Why the system is struggling

I see three organisational trends contributing to this problem.

1. We keep stretching managerial capacity

As organisations flatten structures and widen spans of control, managers are expected to support more people while maintaining operational performance.

There may be good reasons for reducing layers of management. But there are consequences.

A manager with eighteen direct reports has a very different opportunity to observe, coach and develop individuals than one with seven.

This isn’t necessarily a question of commitment or skill. It is a question of time and attention.

No amount of feedback training creates more hours in the day.

2. We promote people into management without adequately preparing them

Gallup reports that only 44% of managers globally say they have received management training.

That is a significant capability gap.

We still frequently promote technically capable people into roles where success depends increasingly on enabling others. Then we expect them to coach, develop, motivate and provide useful feedback, often without enough preparation or ongoing support.

Training matters. But it needs to develop judgement and relationships, not simply provide another conversational script.

3. We keep adding responsibilities without removing anything

Managers have absorbed successive waves of responsibility: organisational change, wellbeing, hybrid working, performance management, increased reporting requirements and, more recently, AI adoption.

Few organisations have been equally disciplined about deciding what managers should stop doing.

We talk about empowering managers while continuing to fill their days with administration, coordination and competing demands.

The result is a familiar organisational contradiction.

We tell managers that developing people is one of their most important responsibilities. Then we design their jobs so that it is often the first thing sacrificed when workloads rise.

Will AI solve the feedback problem?

AI offers some genuine opportunities.

It can help managers capture observations, identify patterns, prepare for conversations and reduce the administrative effort associated with performance reviews.

Used thoughtfully, it may free up time for the human interactions that matter most.

But there is an important limitation.

More feedback information is not necessarily better feedback.

An AI system can help produce a more polished performance summary. It cannot, by itself, create trust between two people.

It can identify a pattern in performance data. It cannot reliably substitute for the judgement, context and care needed to discuss what that pattern means.

And it cannot compensate for a manager who is unavailable, lacks credibility or is working within a culture where speaking openly carries personal risk.

There is another complication.

Gallup’s research on AI adoption found that employees whose managers actively support AI use are around 1.7 times as likely to use it frequently and more than seven times as likely to strongly agree that AI helps them do what they do best.

Yet only around 36% of employees in organisations integrating AI strongly agree that their manager supports their team’s use of it.

In other words, the effectiveness of AI adoption itself is strongly associated with managerial support.

This creates a paradox.

We are looking to technology to relieve pressure on managers, but successful technology adoption frequently depends on those same managers having the capacity to lead it.

AI can help. But it won’t repair a poorly designed management role.

What should organisations do differently?

The answer isn’t to abandon feedback training.

It is to stop expecting training to solve problems it was never designed to address.

I would suggest five priorities.

1. Diagnose the feedback environment before prescribing a solution

You could do worse than start with the seven dimensions of the feedback environment mentioned here.

Where are people experiencing difficulty?

Is feedback too infrequent? Do people distrust its source? Is it inconsistent? Are managers unavailable? Do employees hesitate to ask for help or raise concerns?

Different weaknesses require different interventions.

If delivery is the problem, training may be the answer.

If managers lack time, the solution may involve redesigning workloads or reallocating responsibilities.If people do not feel safe speaking openly, introducing another  feedback process is unlikely to help.

The starting point should be the real constraint, not the intervention we happen to have available.

2. Treat managerial capacity as a strategic resource

Senior leaders need to be more deliberate about what they expect managers to do.

That means examining spans of control, the administrative burden, competing priorities and the amount of time genuinely available for developing people.

It also means asking what managers should stop doing.

We cannot keep adding responsibilities to a role and assume that better skills will somehow make everything fit.

3. Build stronger peer feedback

One of the overlooked strengths of Steelman’s model is that it recognises feedback from both supervisors and coworkers.

Yet most organisational feedback systems remain heavily dependent on the line manager. Colleagues often see aspects of performance that managers don’t. They observe how people collaborate, solve problems, respond to pressure and contribute to the work of others.

Creating better opportunities for peer feedback can improve the quality and range of information available.

It also reduces the organisation’s reliance on a single feedback channel.

This need not mean another formal 360-degree process. Often, it means establishing team practices where reflection, constructive challenge and learning from one another become normal parts of work.

4. Build psychological safety alongside feedback capability

Feedback depends on people’s willingness to speak, listen, question and acknowledge mistakes.

Research published by Mental Health First Aid England in February 2026 found that 45% of surveyed UK employees felt unable to raise mistakes or risks at work, while 35% did not feel safe asking for help.

New Zealand is not immune. The 2024 Ethics at Work survey found that 42% of employees who reported workplace misconduct experienced some form of retaliation or personal disadvantage. Only 49% were satisfied with the outcome of raising their concerns, down from 58% three years earlier.

More recent New Zealand research, involving over 11,600 workers, has also linked low psychological safety, excessive workloads and poor change processes with increased psychological distress and reduced performance.

Although these studies measure different aspects of workplace experience, they point towards a common challenge.

Feedback depends not simply on whether people are encouraged to speak, but on whether they trust what will happen when they do.

And that trust is shaped as much by organisational behaviour and leadership culture as it is by the feedback skills of individual managers.

Those findings should concern anyone interested in performance and organisational learning.

A workplace where people conceal problems is not simply a poor feedback environment. It is a risk to decision-making, improvement and organisational effectiveness.

Leaders need to demonstrate that feedback is something they seek as well as give.

How managers respond when challenged, corrected or presented with uncomfortable information may matter more than how elegantly they deliver a difficult message.

5. Develop managers within a system that supports them

There remains a strong case for investing in feedback and coaching capability.

Managers benefit from understanding how to frame feedback, focus attention on the task, invite reflection and help people translate insight into action.

But development should be embedded in the realities of managerial work.

That includes opportunities to practise, receive coaching, reflect with peers and apply new behaviours over time.

Most importantly, senior leaders must reinforce those behaviours through the way managerial performance is assessed and rewarded.

If we say people development matters but reward managers almost exclusively for short-term operational results, we should not be surprised when feedback and coaching fall down the priority list.

The uncomfortable conclusion

Gallup’s 2026 report offers one final finding worth reflecting on.

While global manager engagement stood at just 22%, it reached 79% in organisations Gallup identified as best-practice workplaces.

That is an extraordinary gap.

It would be simplistic to attribute the entire difference to organisational design. Leadership quality, selection, development and culture all contribute.

But it does challenge the idea that declining managerial engagement is inevitable.

The conditions organisations create matter.

And that brings us back to feedback.

For years we have focused on making managers better at having feedback conversations.

That is useful, but insufficient.

We also need to ask whether managers have the time, credibility, support and the organisational conditions needed to make those conversations meaningful.

We need feedback systems that draw on more than one source, encourage openness, reward learning and recognise the limits of human capacity.

Perhaps the biggest improvement we could make to feedback has little to do with how managers give it.

It may have far more to do with how we design their jobs.

Like to know more about how to use practical neuroscience to drive more impact?

Register for one of our public leadership programmes

Connect with us on LinkedIn, or sign up to our e-newsletter