How do I give feedback without making someone defensive?

Perhaps we are aiming at the wrong thing

I have given and received enough feedback over the years to be wary of any technique that promises to make difficult feedback painless. Sometimes feedback is uncomfortable because it matters. If I tell someone that the way they are behaving is making it harder for colleagues to contribute, that their work is repeatedly arriving late, or that something they are doing is damaging an important client relationship, I may be asking them to reconsider something they believe about their competence, intentions or behaviour. They might be surprised, disagree with my interpretation, feel embarrassed or become angry. None of those reactions automatically tells me that I gave the feedback badly.

This is why I think the question How do I give feedback without making someone defensive? needs a slight adjustment. A better question is How do I give feedback without creating unnecessary defensiveness? Some of the discomfort in a feedback conversation belongs to what is being discussed. Some of it is created by how I introduce, describe and pursue it. I cannot control the first completely, but I can take considerably more responsibility for the second. The aim is not to engineer a conversation in which the other person never feels uncomfortable or challenged. It is to make something important as possible to hear, examine and act on without adding unnecessary threat.

There is a tension here that I regularly see in leaders. At one extreme, people become so concerned about making someone defensive that they cushion the feedback until the message almost disappears. The conversation is warm, careful and respectful, but the recipient leaves without being entirely sure what the manager wanted them to understand. At the other extreme, “I’m just being direct” becomes permission to pay little attention to how the message has been constructed or received. Neither is particularly useful. The challenge is to remain clear enough that the issue can be understood while being open enough to the possibility that your interpretation of what happened is incomplete.

Know where observation ends and interpretation begins

One of the simplest ways to create unnecessary defensiveness is to confuse what happened with what we think it means. Imagine that I have just watched a meeting in which one member of the team interrupted a colleague several times. I could say, “You interrupted Maria four times during the first ten minutes.” I could move one step further and say, “You weren’t listening to Maria.” Or I could go further still and conclude, “You’re dismissive of other people.” The first is reasonably close to an observation, the second is my interpretation of what the behaviour means, and the third has become a judgement about the person.

The problem is not that we interpret behaviour. Leaders have to interpret what they see all the time. The problem is forgetting that we have interpreted it. The movement from observation to interpretation can happen so quickly that, by the time I give the feedback, my explanation feels like an objective description of reality. If I begin the conversation with “You’re dismissive of other people”, the recipient has very little room other than to accept or defend themselves against my judgement. If I begin with what I actually noticed, we have something much more examinable.

This distinction between observation and interpretation also sits behind several of the feedback approaches I use with leaders. The AID feedback model, for example, starts with the Action that was actually observed before considering its Impact and what might need to Develop (o Desired Behaviour). The BOOST feedback model approaches the same challenge from a slightly different direction, asking whether feedback is observed, objective and specific. Neither model removes judgement from feedback, but both make it harder to begin with a sweeping conclusion about the person rather than something they actually did.

A useful way of slowing this down is to separate four things in your own mind before starting the conversation: What happened? What have I made it mean? Why does it matter? What don’t I yet know? In the example above, that might become: I noticed you interrupted Maria four times. I started to wonder whether her perspective was being given enough space. After the fourth interruption she contributed very little, and I was concerned because we needed her expertise. What I don’t know is what was happening from your perspective. Those distinctions do not remove judgement from the conversation. They make the judgement visible rather than disguising it as fact.

Make your reasoning visible, but keep it open

This is also why I am slightly uncomfortable with the common advice to “stick to the facts”. If all I say is “You interrupted Maria four times”, a perfectly reasonable response might be “So what?” I am raising the issue because I think those interruptions mattered. Pretending that I am simply a neutral recorder of observable behaviour can make the conversation strangely artificial. The better discipline is not to hide my interpretation, but to distinguish it from what I observed and make my reasoning available for examination.

I might therefore say: “I noticed you interrupted Maria four times during the first ten minutes. After the fourth interruption, she contributed very little, and I became concerned that we were losing a perspective we needed. What was happening from your side?” Now the recipient can see how I moved from an observation to a concern and, importantly, they can challenge that reasoning. Perhaps they tell me that Maria had asked them before the meeting to keep moving her on because she knew they had very little time. That is useful information. It does not mean I was wrong to raise what I noticed. It means my interpretation has encountered information I did not have when I formed it.

This is an important distinction because feedback is not simply the transfer of objective truth from someone who knows to someone who does not. It is information about behaviour, performance or impact from a particular vantage point. Sometimes that information is highly reliable and the implications are clear. Sometimes it is incomplete. Often it is somewhere between the two. A useful feedback conversation therefore allows me to be clear about what I have seen and why I think it matters while remaining genuinely interested in information that might strengthen, modify or occasionally overturn my interpretation.

If you want more structure for this kind of two-way conversation, the CEDAR feedback model is particularly useful. After establishing the context and examples, its Diagnosis stage deliberately asks both people to explore what happened rather than assuming the giver already possesses the complete explanation. That is particularly valuable when you have a legitimate concern but also recognise that there may be context you do not yet have.

Don’t confuse disagreement with defensiveness

Someone saying “I don’t think that’s an accurate description of what happened” is not necessarily being defensive. They may simply think you are wrong. The distinction matters because once I label someone as defensive, it becomes remarkably easy to discount anything they say next as further evidence that they cannot receive feedback. Their disagreement becomes proof of my original diagnosis.

If I regard my feedback as important information rather than an infallible verdict, disagreement becomes information too. I can ask which part they disagree with, what they think I am missing and how the situation looks from their perspective. Sometimes what I hear will change my view. Sometimes it will add context without changing the central issue. Sometimes I will listen carefully and still disagree. That is also a legitimate outcome. “I understand why you see it differently, and I still think we need to address what happened” allows both things to be true. Curiosity does not require me to surrender my judgement, just as clarity does not require me to stop listening.

Remember that power is already in the room

Feedback does not take place between two abstract, equal participants. The same sentence from a manager can carry a very different weight from the same sentence spoken by a peer. If I have influence over someone’s performance rating, opportunities, reputation or progression, they may be listening not only to what I am saying but also trying to work out what it means for them. Is this a serious performance issue? Has this already been discussed with other people? Is this going to appear in my review? Does my manager now think differently about me? What looks like defensiveness may sometimes be an attempt to understand the consequences of a conversation in which the stakes are not equal.

The greater the power difference, the more cautious I think we should be about interpreting silence as agreement or a calm response as evidence that the feedback has landed well. It also makes clarity particularly important. If this is an observation I want the person to consider, say so. If it is an expectation that needs to be met, be equally clear about that. Inviting someone’s perspective should not be used to disguise a decision that has already been made, just as managerial authority should not be used to turn one person’s interpretation into unquestionable truth.

When defensiveness appears, don’t prosecute

Even when you have prepared carefully, described something specific and explained why it matters, the other person may still say, “That’s completely unfair.” There is a powerful temptation at that point to prove that it isn’t. We produce another example, and then another. The recipient responds to each one. Before long the conversation resembles a courtroom, with one person prosecuting and the other constructing a defence. The irony is that our response to defensiveness has created even more material for the person to defend themselves against.

I would usually become more curious before becoming more forceful. Which part feels unfair? What do you think I’m missing? How does this look from where you’re sitting? What is different about your reading of what happened? These are not tricks for calming someone down. They need to be genuine questions. The possibility that I have missed something has to be real.

Sometimes their explanation changes my view. Sometimes it adds context but does not change the issue. Sometimes, after listening carefully, I still disagree. Listening does not require withdrawing the feedback. It simply prevents me from responding to resistance by turning the conversation into a case I have to win.

Give the person somewhere to go

Compare “You need to communicate better with Finance” with “I’m noticing that Finance is getting involved very late, and we’re then having to rework decisions. What’s happening from your side?” The first identifies a deficiency but gives the recipient very little to work with. “Communicate better” could mean almost anything. The second names a pattern, describes why it matters and opens the door to information.

Perhaps Finance repeatedly ignores invitations. Perhaps responsibilities are unclear. Perhaps the person has avoided involving them because previous interactions have been difficult. Or perhaps they simply have not recognised the consequences of leaving Finance out. We do not yet know.

Good feedback should increase someone’s capacity to act. If the conversation leaves them feeling judged but unclear about what they could do differently, we have created discomfort without much learning. Once the situation is clearer, the conversation can move towards agency: What could you do differently next time? What would make that easier? What part of this is within your control? What would you like to try?

This does not mean every feedback conversation ends with the other person choosing whether change is optional. Managers sometimes have to be explicit about expectations. But even where an outcome is required, there is often considerable room for ownership over how someone gets there.

Be particularly careful when feedback touches identity

Some feedback lands much harder than its apparent size would suggest. Tell someone who sees themselves as highly reliable that they have missed an important deadline and you may not simply be talking about a deadline. Tell a leader who prides herself on being inclusive that people experience her as dominating meetings and you may not simply be talking about airtime. Tell someone known for their attention to detail that their report contained significant errors and you may be colliding with something central to how they see themselves.

That does not mean you should avoid the feedback. Often it makes the conversation more important. But it does mean that behavioural specificity matters. Compare “You’re careless” with “There were three figures in yesterday’s report that came from the previous reporting period.” The second gives us somewhere to work. We can investigate how the errors happened, what their impact was and what needs to change. The first invites a much larger argument about whether the person is careless.

What does current research suggest?

The research on workplace feedback is more complicated than the simple message that “feedback improves performance”. A major 25-year review by Anseel and Sherf (2025) concluded that feedback research consists of several streams built on different assumptions, and that findings do not always map neatly onto how feedback is actually experienced in organisations. That complexity is useful. It reminds us that there is no magic wording that reliably turns difficult information into improvement.

Feedback can help, but it can also make performance worse

One of the most important findings in the field remains Kluger and DeNisi’s (1996) meta-analysis of feedback interventions. Across 607 effect sizes and more than 23,000 observations, feedback improved performance on average, but more than a third of the feedback interventions actually reduced performance.

Their Feedback Intervention Theory offers an important explanation. Feedback becomes less effective as it shifts attention away from the task and towards the self. That provides a strong research foundation for keeping difficult feedback close to observable behaviour, its consequences and the work that needs to change, rather than turning it into a judgement about the person.

“You are careless” invites attention towards Who am I? “Three figures in this report came from the previous reporting period” directs attention towards What happened, and what needs to change? That distinction is more than good conversational etiquette. It may influence where the recipient’s attention goes.

Receptivity depends on more than whether the feedback is accurate

A review by Fulham, Krueger and Cohen (2022) makes another useful distinction. For feedback to produce learning, recipients need both to be receptive to it and able to discern whether it is true and useful. The authors identify barriers arising from the feedback itself, the person giving it and the recipient. Importantly, responsibility does not sit entirely with either side.

That supports the argument that the giver should reduce avoidable barriers without assuming responsibility for making the recipient accept the message. The giver can make feedback specific, honest and examinable. They can explain their reasoning and listen to additional information. But the recipient also has work to do in considering what they have heard.

Credibility and the characteristics of the feedback matter

A systematic review and meta-analysis by Zyberaj (2026) examined 24 empirical studies and 75 effect sizes relating supervisory feedback characteristics to different stages of feedback processing. One striking feature of the literature was the attention given to source credibility. The review also found positive associations between feedback quality and how feedback was perceived, accepted and acted upon, while warning that much of the evidence is correlational.

This matters practically. People do not receive feedback as disembodied information. They also make judgements about the person giving it. Do they understand my work? Have they actually observed what they are describing? Do I believe their intentions are constructive? Are they applying standards fairly? Is their interpretation worth taking seriously?

This is another reason why concrete observation and transparent reasoning matter. “You’re not collaborative enough” asks the recipient to trust a large judgement. “I noticed X, I saw Y happen afterwards, and this is why I’m concerned” makes more of the reasoning available for examination.

People differ in their orientation towards feedback

A 2023 meta-analysis involving 46 independent samples and 12,478 workers found that feedback orientation, broadly someone’s receptivity towards feedback, was positively related to feedback seeking, job satisfaction and work performance (Katz, Moughan and Rudolph, 2023).

That is a useful corrective to the idea that the giver controls everything. The same feedback will not land identically with everyone. People differ in their willingness and readiness to seek, process and use information about themselves. So when someone becomes defensive, the answer is not necessarily to search frantically for a better sentence. The recipient brings their own history, self-perception, goals and concerns into the room.

The emerging picture is relational rather than mechanical

Taken together, the research points away from the idea that there is one perfect feedback formula. Effective feedback depends on what is said, what it directs attention towards, how credible and useful it appears, who is receiving it, the relationship between the two people and what happens after the initial message.

That is why I think the best stance is neither “say it nicely” nor “just tell people the truth”. It is this: Be specific enough that there is something real to examine. Be clear enough that the issue does not disappear. Be open enough that your interpretation can encounter information you do not yet have.

Five questions before you give difficult feedback

Before starting the conversation, I would ask myself:

1. What did I actually observe? Strip away the labels and identify the behaviour, event or outcome you could reasonably describe.
2. What have I inferred from it? Separate what happened from the story you have constructed about why it happened or what it says about the person.
3. Why does this matter? Be able to explain the impact on the work, team, client or relationship.
4. What might I not know? Identify where your interpretation could be incomplete before entering the conversation.
5. How is this person likely to react, and how should I adapt my approach? Consider what the feedback may threaten for them, what history or power dynamics may shape the conversation, and what will help them hear the message. Adapt the timing, setting, pace and opening without diluting what needs to be said or assuming you can control their response.

These questions are a preparation tool rather than another feedback model. If you want a broader practical overview of the capability itself, my Feedback Giving guide looks at the wider skill of preparing for and conducting effective feedback conversations. The five questions here have a narrower purpose: helping you think more clearly before a conversation in which defensiveness may be a particular risk.

Three questions for reflection

1. When you anticipate defensiveness, do you tend to soften the message until it becomes unclear, or push harder until the conversation becomes an argument?
2. Think of a piece of feedback you need to give. What did you actually observe, what have you inferred, and what information might still be missing?
3. What could you change about the timing, setting or opening of the conversation to reduce unnecessary threat without weakening the feedback?

Conclusion

You cannot make difficult feedback entirely safe, and trying to do so can create its own problems. The message becomes vague, the concern is disguised, or curiosity becomes a technique for manoeuvring the other person towards a conclusion you have already fixed.

Your responsibility is more demanding and more realistic. Describe what you have actually noticed. Show how you reached your concern. Explain why it matters. Invite information that could change your understanding. Stay clear when the issue still needs to be addressed. And remember that the recipient’s reaction is influenced by far more than the sentence you choose.

If the challenge is less about defensiveness and more about choosing how to structure the conversation, I have brought together the different approaches I use and write about in my guide to feedback models and effective feedback at work. It includes AID, BOOST, CEDAR and several other approaches for giving, receiving and working with feedback.

Three questions for reflection

1. When you anticipate defensiveness, do you tend to soften the message until it becomes unclear, or push harder until the conversation becomes an argument?
2. Think of a piece of feedback you need to give. What did you actually observe, what have you inferred, and what information might still be missing?
3. What could you change about the timing, setting or opening of the conversation to reduce unnecessary threat without weakening the feedback?