Effective feedback: why a third of all feedback actually reduces performance, and how to do it differently
There is something strange about giving feedback in the workplace. Everyone says it’s important. Every leadership book devotes chapters to it. Every HR strategy includes a section on feedback culture. And yet, even in organisations that claim to have ‘got it right’, it remains one of the most dreaded moments in working life. People put it off. They say it wrong. Or they say it too late, in a formal performance review where the other person should have been prepared long ago.
The underlying assumption is that giving feedback is essentially a good thing. That, as long as you do it, it leads to better performance. That assumption is largely incorrect. And that is not a gut feeling, but a scientifically established finding that still has hardly any impact in practice.
Anyone who genuinely wants to improve feedback within an organisation must first understand where it goes wrong. The solution lies not in giving more feedback, but in giving better feedback. Feedback that is specific, behaviour-focused and immediately actionable. And above all: feedback that you keep practising in real life, until it becomes a natural part of how people work together.
The uncomfortable secret of feedback
In 1996, Avraham Kluger and Angelo DeNisi published a meta-analysis in Psychological Bulletin that turned the feedback literature on its head in one fell swoop. They examined hundreds of results from years of research into feedback and reached a sobering conclusion. On average, feedback did indeed lead to an improvement in performance, but in more than a third of all cases, giving feedback actually led to a deterioration in performance. That is no minor deviation. It is a fundamental phenomenon that most managers are unaware of.
What Kluger and DeNisi uncovered is that feedback is not a neutral intervention. The way it is given, at what point, by whom, and focused on which level (the task, the self, or the process) determines whether it helps someone progress or actually hinders them. Feedback focused on the recipient’s self, for example: ‘you are too detail-oriented’, is more likely to lead to defensiveness and reduced performance than feedback focused on the task. For example: ‘this email lacks a clear question in the first paragraph’.
That distinction may seem minor, but in practice it makes a huge difference. And it explains why organisations that ask their managers to ‘be more direct’ sometimes end up with worse results than before. And not so much because directness is wrong, but because directness without skill leads to precisely the kind of feedback that the meta-analysis has already identified as detrimental to performance.
Why feedback goes wrong: three mechanisms we underestimate
There are three patterns that repeatedly undermine feedback in the workplace, and none of them has anything to do with the content of what is said.
The first is timing. Research shows that feedback given closer to the moment of action is consistently more effective than feedback given weeks later. And yet the standard model in most organisations is exactly the opposite: annual appraisal interviews with observations accumulated months ago. An employee who made a mistake in March is told in November that they should have done better. The brain has long since filed that situation away as closed, and the feedback comes across as a criticism of who someone is rather than what they did.
The second is specificity. General feedback (“you need to show more initiative”) gives the recipient no starting point for changing their behaviour. Specific feedback (“in Tuesday’s meeting, you waited until three colleagues had spoken before making your point; I’d like to hear from you sooner”) does. The difference lies not in how it sounds, but in whether you can actually do anything with it. In other words: concrete behaviour.
The third is reciprocity. In organisations where managers give feedback but never ask for it themselves, a dynamic arises that psychologists consistently link to a reduction in trust and openness. Research in the Harvard Business Review shows that 57% of employees would rather receive corrective feedback than purely positive feedback, whilst managers are actually reluctant to give that feedback. That gap between what people want and what they get is one of the weakest points in most feedback cultures.
What constructive feedback actually does to the brain
To understand why constructive feedback works and ‘bad’ feedback does not, it helps to look at the brain. When someone receives feedback that is interpreted as an attack on the self, this activates the amygdala, the area of the brain responsible for the fight-or-flight response. At that moment, the prefrontal cortex’s capacity to process and integrate information is reduced. In other words: just when you want someone to learn, the brain shuts down to learning.
This is no excuse for giving vague feedback or avoiding difficult messages. It explains why a specific approach is needed. Giving constructive feedback means: sticking to the substance, but choosing the form in such a way that the recipient is actually able to take in what you are saying. That requires skill, situational leadership and practice.
What often works in practice is a structure that anchors the feedback in specific behaviour, identifies its effect, and invites dialogue about alternatives. Models such as COIN (Context, Observation, Impact, Next steps) or the simple situation-behaviour-impact format help with this. But as with all skills, knowing the model is the easy part. Applying it, whilst your heart rate rises and the other person becomes more defensive, requires something fundamentally different.
Why training on giving feedback rarely leads to different behaviour
An average feedback training programme lasts a day, sometimes two. Participants learn models, practise once or twice in a role-play with a colleague, are given a handout, and return to work. Three weeks later, a team lead is sitting down with a colleague who is missing deadlines, and without exception falls back on the patterns he had before the training. Not because the training was poor, but because a one-off exercise is not enough to change behaviour.
This is the transfer problem faced by the entire L&D sector. Research estimates that less than ten per cent of what is learnt in training actually translates into new behaviour in the workplace. For feedback skills, that percentage is likely even lower, because application depends so heavily on factors such as the conversation partner, the context and emotional self-regulation. And when we talk about emotional regulation, people often forget that this is only developed through repeated exposure.
Psychologist K. Anders Ericsson demonstrated in his research on expertise development that skills such as giving feedback do not arise from insight or motivation. They arise from deliberate practice: focused, repeated practice on specific aspects, with immediate feedback on one’s own actions. That is precisely what cannot be offered in a single day of training, and also what most organisations were unable to organise until recently.
How AI role-plays measurably improve feedback
This is where PractAIce comes in. The platform allows employees to practise feedback conversations with an AI avatar that reacts just as a real colleague would – with defensiveness, with emotion, and with unexpected counter-questions. That sounds simple, but it fundamentally changes how feedback skills develop.
A manager preparing for a difficult conversation with a team member can run through the scenario three times before the actual conversation takes place. She experiments with different ways of opening the conversation. She observes the difference between a closed question and an open question. She experiences how a direct way of phrasing things comes across differently from a more complex one. And most importantly: after each attempt, she receives detailed, behaviour-focused feedback on how she handled it. Not: “you did well”, but specifically which moments were effective and which weren’t, and why.
For L&D professionals, this adds another dimension: measurability. PractAIce generates data on specific behavioural indicators for each practice session. Was the feedback specific enough? Was there sufficient room for response? How was defensiveness handled? That data, collected across multiple practice sessions, shows for the first time how a team’s feedback skills are developing. And not based on self-reporting, but on observed and factual behaviour.
This also changes the nature of the conversation about feedback culture. No longer based on survey questions where people state that they are ‘open to feedback’, but on observable improvement in how feedback is actually given and received.
Frequently asked questions about giving feedback
What is the difference between feedback and criticism?
The difference lies not in the content, but in the intention and the form. Criticism is aimed at judging the other person; feedback is aimed at enabling improvement. In practice, this means: feedback identifies specific behaviour, describes its effect, and leaves room for a response. Criticism often does only the first, or the second without the first.
How do you give feedback to a colleague who is senior to you? Giving feedback to someone above you requires care in how you phrase it, not compromises on the content. What works in practice: first ask explicitly if there is room for your observation, describe specifically what you have noticed, and leave the interpretation to the other person. “I noticed that I didn’t have the chance to finish my point in this morning’s meeting; I just wanted to mention it to see if that rings a bell” comes across differently from a judgemental statement.
How often should you give feedback?
Research consistently points to frequency over intensity. Short, frequent feedback closer to the moment works better than feedback saved up for a formal meeting. For most working relationships, daily to weekly is a good frequency. Not as a ritual, but as a natural expression of engagement with the other person’s work.
Can you practise giving feedback using AI?
Yes, and in many ways it’s more effective than traditional training. AI role-plays offer the repetition, safety and immediate feedback that deliberate practice requires. What AI cannot replace is the human nuance of an experienced coach or mentor. What AI does enable is for employees to practise the basic skills of giving feedback repeatedly before applying them in real conversations. And this is something that has never before been possible at scale in most organisations.
In conclusion
Giving feedback is a skill whose importance is recognised across the entire working world, yet one in which virtually no one receives systematic training. This is not due to any ill intent; it is a consequence of the fact that, until recently, we had no scalable way of allowing people to practise properly.
AI role-playing has closed that gap. It becomes possible to develop feedback skills in the way science has long prescribed: repeatedly, in realistic situations, with immediate feedback on one’s own actions. For organisations wishing to build a feedback culture, this is a shift that makes all the difference.
Are you unsure how this would work in your organisation? A 15-minute demo of PractAIce shows how a feedback session with an AI avatar unfolds and what data it generates.