How do I give negative feedback without damaging the relationship?

When a difficult conversation goes badly, the conversation is rarely the cause. It is the weeks after it. Many difficult feedback conversations are handled with care and still leave trust weaker than before.

I have seen the same sequence more than once. The feedback is clear and kind. The other person nods. Nobody walks out. Then, over the next month, they stop raising concerns, or they agree with everything too fast, and you are left wondering what happened.

In most of these cases, the person was not sure why the conversation was happening, or felt you had either held back or come on too strong. Some leaders avoid the second risk by softening critical feedback until it disappears. The point never lands, and the other person is left to guess what you meant.

So when you give constructive negative feedback, ask which mistake you are more likely to make. Soft feedback erodes trust slowly. Blunt feedback breaks it quickly. Most of us lean one way, so it is worth knowing which before your next conversation.

Warm, candid or both?

Scott (2017) offers one of the best-known versions of this idea. Her model asks two things: how much you care about the person, and how directly you name what needs to change. Put those on two axes and you get four kinds of conversation.

Warmth runs left to right, and candour runs bottom to top.

Under pressure, people drift. The useful question is what the drift protects. Sometimes it protects the other person. Often it protects you.

This is a practitioner model, not a tested one, and it leaves a lot out. Its value is in separating two failures people often confuse: saying too little, and saying it without enough care.

When kindness becomes the problem

All of the leaders I work with are genuinely kind. You notice when someone is struggling, and you decide now is not the moment for the hard point. The next few conversations go smoothly. The issue never comes up. You tell yourself you are protecting the relationship.

What usually follows is quieter. The other person can tell something is being held back, even if they could not say what. They edit what they share, agree a little too quickly, or ask a colleague for the honest version. The relationship has not broken. It has become less honest, and that is harder to spot than one bad conversation.

The fix is seldom to get blunter. It is to get specific. Start with one thing you saw, the effect it had, and what you want to be different. The AID model (action, impact, desired outcome) helps here, because it keeps the conversation on what happened, not on your general worry about how someone is doing. One clear example is easier to say out loud, and easier for them to act on.

When honesty arrives without warmth

Direct feedback without warmth can damage trust quickly. The content may be accurate, but the person hears the criticism and spends the rest of the conversation defending themselves or protecting their standing.

The problem is rarely the content. It is the absence of any signal that you want them to succeed. An observation can sound like a verdict when it comes without context, without any acknowledgement of what they were trying to do, or in a tone that suggests frustration rather than purpose.

Where firmness is needed, the BIFF model keeps the message respectful. It asks feedback to be brief, informative, friendly and firm, so the challenge is clear without becoming personal. Friendly does not mean soft. It means saying plainly and sincerely that you think they can do better, and that you want to help them do it.

Before the conversation, check whether you can state your care for them in one honest sentence. If you cannot, the feedback will probably land as criticism, however carefully you word it.

Staying honest over time

A single missed deadline can be discussed and forgotten. A recurring pattern needs a different conversation, because the other person has probably stopped noticing it.

Take a colleague whose reports arrive a day late most weeks. One conversation about the latest report will not change the pattern. A CEDAR conversation names the pattern, agrees the outcome you need, explores what gets in the way, and sets a review date three weeks out. The structure gives them room to respond, and the review date shows you are still paying attention. Feedback with no review date tends to fade. You can find all the models in our Feedback Hub.

Repairing trust after the conversation

The conversation is only part of the work. What you do afterwards tells the person far more about your intent than anything said in the room. This is the part most leaders leave out. Ask directly how the feedback landed: “Was that useful, and how did it come across?” That invites an honest answer, and it shows you care about the effect, not only the delivery.

Follow up within a few days. Check progress without hovering. Keep the small commitments, such as the resource or the check in you promised. Trust rebuilds through a run of ordinary things done as promised, not through one grand gesture. For more, see How to Build Trust at Work.

When trust was already low

Sometimes feedback is necessary when trust is already thin. Perhaps the person has had a hard year, or an earlier decision left them feeling overlooked. Even a carefully judged conversation may meet suspicion, and the feedback alone will not restore the relationship.

Start by acknowledging the wider context honestly. That does not mean apologising for every earlier decision. It means recognising that they may be listening with a different ear. Naming the strain often lowers the temperature more than any careful phrasing. Then the feedback becomes one step in a longer process of rebuilding. For more, see Rebuilding Trust in Leadership.

Feedback and trust: how to tell whether trust is recovering

The clearest sign is what the person does, not what they say. Watch for the signals in the table below.

The third row needs the closest attention. If those signals show up, the relationship may still be under strain, even if nobody has said so.

Try a simple habit. Two or three weeks after the conversation, have a short, unhurried check-in. Ask one question about how the feedback is working in practice, then listen without defending the original point.

What does the research suggest?

Research on feedback goes back decades, and its findings are mixed. Kluger and DeNisi’s (1996) meta-analysis of feedback interventions found that feedback improved performance on average, but that over a third of the interventions reduced it. The effect is not reliable enough to assume.

Edmondson’s (1999) study of 53 teams at one office furniture manufacturer found that psychological safety was the strongest predictor of learning behaviour. The study did not test feedback conversations, so applying it to them is an inference. It fits the concern, though, that people hold back when they do not feel safe to speak.

The review of feedback seeking by Ashford et al. (2003) organises the research around three motives: achieving goals, protecting ego, and managing image at work. If image concerns shape who asks for feedback, a colleague who stops asking for your view may be protecting their image rather than losing interest in your opinion. That is a reason to keep the door open and ask directly.

Five questions to ask yourself after giving negative feedback

These five questions are a short check to run after any feedback conversation, ideally within a day. They are not a test of how well you performed. Each one points at a common way trust gets lost, so your answers show where to look before the next conversation.

1. Did the person know what the conversation was for before it started?
   If you cannot say what the purpose was, the other person probably could not either. People who are unsure why they are being spoken to tend to fill the gap with the least generous explanation available. This question shows whether the purpose was clear to them, or only clear in your head.
2. Was the feedback based on something I observed, rather than something I assumed about them?
   Assumptions about motive are one of the fastest routes to defensiveness. Reviewing this question shows whether you described what happened or what you concluded about the person. If you cannot point to the evidence, the feedback may need repairing.
3. Did I say what I value in them, and was that honest?
   Telling someone you value them only helps if it is true, and if they can see it in what you do. Praise that feels like a cushion is quickly recognised for what it is. This question checks that the warmth was real.
4. Did I ask how the feedback landed, and did I listen to the answer?
   You can only learn how feedback landed by asking. Reflecting on this question shows whether you asked at all, and whether you could hear the answer without defending yourself. Even a short or guarded answer is useful information.
5. What did I commit to doing next, and have I done it?
   Trust usually follows the small things done as promised. This question checks whether you named a next step, and whether you have taken it. A promise kept is one of the clearest signals you can send.

Conclusion

A difficult conversation is one event. What people remember is how you behaved in the days and weeks that followed. If they can see you care about the outcome and about them, they will bring you the hard things next time. That is the relationship worth protecting, and it is how you maintain trust over the long run.

Three questions to reflect on

1. Which pattern do you drift into most often, and what is it protecting?
2. Which feedback conversation are you putting off, and what is the delay costing?
3. When you asked how your feedback landed, what did the person say, and what did you change?

Sources

Ashford, S.J., Blatt, R. and VandeWalle, D. (2003) ‘Reflections on the looking glass: A review of research on feedback-seeking behavior in organizations’, Journal of Management, 29(6), pp. 773–799. doi: 10.1016/s0149-2063(03)00079-5

Edmondson, A.C. (1999) ‘Psychological safety and learning behavior in work teams’, Administrative Science Quarterly, 44(2), pp. 350–383.

Kluger, A.N. and DeNisi, A. (1996) ‘The effects of feedback interventions on performance: A historical review, a meta-analysis, and a preliminary feedback intervention theory’, Psychological Bulletin, 119(2), pp. 254–284. doi: 10.1037/0033-2909.119.2.254

Scott, K. (2017) Radical Candor: Be a Kick-Ass Boss Without Losing Your Humanity. New York: St Martin’s Press. ISBN 978-1-250-10350-5.